Commissioning and Optimization of a Total Skin Electron Therapy Technique Using a High Dose Rate Electron Facility

Total skin electron therapy (TSET) is used for the treatment of Mycosis Fungoides. Several techniques have been developed, in order to achieve homogeneous dose distribution over the complete body surface. To implement a TSET technique, one has to optimize a variety of parameters. Monte Carlo simulation of TSET can facilitate this optimization. The aim of this study was to commission and optimize a TSET technique using the 4 and 6 MeV electron and the high dose rate facility on the Elekta Precise accelerator. The EGS4nrc/BEAMnrc Monte Carlo code was used. The beam data were calculated and measured at two different scoring planes for a single beam. The Model was validated by comparing the simulation with measurements. Two different vertical angles were used to obtain a uniform dose. The angle was optimized for best dose uniformity. The Rando phantom is placed on a rotating platform and rotates 60 degrees apart to facilitate the six patient position orientations. The doses delivered in a phantom by complete treatment were measured with Kodak EDR2 films and TLDs. The dose distribution varied among various scanning directions by 2 3 mm and 3 4 mm for 4 and 6 MeV respectively. The composite percentage depth dose of all six dual fields for the 4 and 6 MeV yielded an R80 of ~4 mm and ~6 mm, respectively. Dose uniformity was ±6% for 4 MeV and ±5% for 6 MeV. The bremsstrahlung contamination was 0.9% 1.3%. Good agreements were found with published literature and inline with international protocols.

I specially thank Mr Kobus van der Walt and Claude Wain wright from the National Hospital mechanical workshop for their design of the platform and solid water phantom.

Radiotherapy
Radiotherapy or radiation treatment is defined as the treatment of diseases (mostly malignant) with ionizing radiation. The radiation may be applied as beams from the outside of the body, a process known as external beam radiotherapy, or by introducing radioactive sources into the body cavities, which is called intracavitary or intraluminal radiotherapy. Sources may be inserted into the patient's tissue to give interstitial radiotherapy. Occasionally radioactive fluids are introduced into the body either via a vein or into the cavity.
The type of treatment used depends partly on the body site requiring treatment. These types of radiotherapy treatment are practiced in most radiotherapy departments; radical radiotherapy, adjuvant radiotherapy, chemoradiotherapy, intraoperative radiotherapy and palliative radiotherapy. Radiotherapy is usually prescribed according to the intention, required for each patient (Griffiths and Short, 1994).
There are various types of ionizing radiation used in radiotherapy such as x-rays, gamma rays, electrons, neutrons, etc. Ionizing radiation is capable of damaging the genetic material (DNA) in vivo without significant deleterious effects on normal tissues. Usually, x-rays are produced in a linear accelerator by stopping fast electrons in a target material such as tungsten, or gamma rays generated in a TeleCobalt unit. Radiation can cure or control cancer by damaging the cancer cells so they cannot divide or reproduce. About fifty to sixty percent of patients with cancer will require radiation at some time or other during the course of their disease. Radiation is a safe and effective form of treatment for patients of all ages (Rath, 2000). Radiotherapy combined with surgical and medical disciplines improve treatment outcome better than surgery or radiotherapy alone. The radiotherapy specialty was born immediately after the discovery of Roentgen rays or xrays by Wilhelm Conrad Roentgen in the year 1895. The first generations of low energy x-ray generators were very inefficient in penetrating deep-seated tumours. Subsequently the discovery of radium in 1898 by Marie Curie gave birth to the specialty brachytherapy.
It was the discoverer of the telephone, Alexander Graham Bell, who proposed the concept of using a radium source inside the tumour.
Following 1945, from the expenence of radar systems, the concept of the linear accelerator evolved. More and more refined x-ray generators (Van de Graaff generator and linear accelerators) have developed afterwards to make radiation more penetrating than the previously available low energy x-ray generators. Artificially prepared radionuclides such as cobalt-60 (60CO) and Caesium-137 (l37Cs) have been in use as sources of radiation in the past seven decades.
In the past, the understanding about radiation safety was not clear. People used radiation casually to treat patients with cancers and non-cancerous conditions. Radiation sources were used widely over several years for brachytherapy purposes until the introduction of radiation safety principles in the 1950s. From the experiences of radiation hazards, afterloading systems for brachytherapy evolved, making radiation therapy safer without the risk of exposure to the medical personnel. With growing technology and better understanding of radiation biology, radiotherapy achieved many milestones at a faster rate. Since the early 1990s, radiation oncology has increasingly become technology oriented. This has resulted in accurate target localization and precise delivery of radiation to the target area resulting in better tumour control, minimal normal tissue complications and to some extent improved survival rates (Rath, 2000). Now radiation therapy plays an important role in cancer management. Today about 45 percent of all cancer patients can be cured, about one half of them are cured by radiation therapy applied alone or in combination with surgery or chemotherapy (Wambersie and Gahbauer, 1995). The clinical experience accumulated in decades shows that, to be efficient, the radiation treatment must be delivered with a high physical selectivity. At present, electron linear accelerators are the primary equipment of a modem radiotherapy department, and are used to irradiate a large proportion of the patients for at least part of the treatment. Photon beams of about 6-20 MV have in general a sufficient penetration in the tissues to treat most of the tumours with an adequate physical selectivity. A combination of several beams adequately oriented allows the radiation-oncologist to deliver the prescribed dose to the "target volume" (tumour) without exceeding the tolerance of the surrounding normal tissues. Conformal therapy, which needs wellequipped and well-staffed centers, further improves the physical selectivity of the treatment, and offers definitive advantages at least for some tumour types and/or locations. Finally, modem linear accelerators are used to maximize accuracy in dose delivery to obtain better therapeutic results in radiotherapy (Wambersie and Gahbauer, 1995), to deliver a high dose to a target volume (tumour) and spare as much as possible the normal surrounding tissue.
In general the radiotherapy aims to deliver enough radiation to the tumour to destroy it without irradiating normal tissue to a dose that will lead to serious complications (morbidity) (Rath, 2000). Study has shown that the dose-response curve is quite steep and there is evidence that a 7 to 10 % variation in the dose to the target volume may result in a significant change in both the tumuor control and normal tissue complication probabilities (Kutcher, 1992).
Radiation therapy demands more accurate dosimetry for good patient care (Metcalfe et al, 1997). The demand has increased tremendously with the advent of computer technology like CT scanners, which allow detailed knowledge of the geometry and densities of the body to be irradiated (Ma et al, 1999). Taking into consideration the steepness of the dose response curve as mentioned above, methods that can be employed for the accurate determination of absorbed dose distributions in the patient have a big role to play (Awusi, 2000, Metcalfe et al, 1997. In fact Monte Carlo simulation is fast becoming the next generation of dose calculation engine for radiation treatment planning systems in routine clinical practice (Mohan, 1997 andJiang, 1999).

Types of rad iotherapy treatment 1.2.1 Radical radiotherapy
Radical radiotherapy is used in early stages of cancers for curative purposes. The radiation oncologist takes a lot of time to accurately delineate the tumour volume, analyze image data, simulate, perform dosimetrie analysis of a plan and actual radiation dose delivery. It usually takes about 6-8 weeks to complete a course in multiple sequential phases called the shrinking field technique. Some common tumours treated by radical radiotherapy are cancers of the larynx, nasopharynx, uterine cervix, skin, bladder, breast, and prostate. Radical radiotherapy involves multiple hospital visits, a prolonged course of treatment up to normal tissue tolerance, and the patient has to expect and accept some degree of acute and chronic side effects (Griffiths and Short, 1994).

Adjuvant radiotherapy
The word adjuvant is derived from the Latin verb 'adjuvere' meanmg 'to help'. In situations where radiotherapy is utilized for the improvement of the results of another modality (usually surgery) it is called adjuvant radiotherapy. Radiotherapy can be delivered before surgery (preoperative radiotherapy), after surgery (postoperative radiotherapy), during surgery (intraoperative radiotherapy) and as a combination of preoperative and postoperative radiotherapy (sandwich radiotherapy). When radiation therapy is administered during surgery, the microscopic and minimal macroscopic disease in the tumour bed get sterilized and thereby local control and ultimately survival is improved. The commonly encountered cancers requiring adjuvant radiotherapy are rectal cancers, head and neck cancers, breast cancers, and brain tumours (Mohan B B, 1999).
Radiotherapy is however, most frequently used postoperatively.
Surgeons find difficulty in excising an infiltrating tumour, because their excision may not be pathologically complete. They are likely to leave residual disease, or spill tumour to the adjacent areas during handling of the tumour. In this situation radiotherapy frequently helps surgeons to circumscribe the tumour and overcome the above difficulties.
Radiotherapy treatment has a higher failure rate at the tumour center which contains radioresistant tumour clonogens. In contrast, radiotherapy is efficient in the eradication of a small number of well vascularized tumour cells at the resection margin. Hence combination of radiotherapy and surgery sounds logical. The best example of postoperative radiotherapy is demonstrated in stage-I seminoma of the testis. By giving prophylactic postoperative radiotherapy, the relapse rate reduces from 15% to near zero percent. The other example is in post excision breast cancer. In this situation the breast relapse rate reduces from 35% to less than 10% after postoperative radiotherapy (Mohan BB,1999).

Chemoradiotherapy
Sometimes anti-neoplastic drugs when given in conjunction with radiotherapy, enhance the efficiency of radiation. When radiation is given concurrently with chemotherapy the cancer cell kill increases by two fold. These principles are used in the organ preservation techniques in anal canal cancer, bladder cancer, esophageal cancer and cervical cancers (Mohan B B, 1999).

Intraoperative radiotherapy
Radiation can be delivered during operation resulting in sterilization of the malignant cells in the tumour bed. The irradiation of the tumour using this technique is superior to percutaneous external beam radiotherapy in multiple doses. Sometimes, electron beam irradiation and interstitial brachytherapy are used to improve local control. This principle of radiotherapy is used in soft tissue sarcoma, pancreatic cancers, stomach cancer and retroperitoneal sarcomas (Mohan B B, 1999).

Palliative radiotherapy
In very advanced cancers, there are poorly defined generalized symptoms which are difficult to manage. In this situation, cure is not possible and the concern is with the issues of quality of life. The aim is therefore the minimization of discomfort, called palliative treatment. This form of therapy should be simple, should not produce

Electron beam radiotherapy
morbidity, and improve quality of life without necessarily prolongation of life expectancy. Palliation can involves some of the following: surgical diversion procedure, nerve block, analgesic medication, transcutaneous electrical nerve stimulation and radiotherapy. Chemotherapy is rarely utilized for palliation in chemosensitive tumours (Griffiths and Short, 1994).
The total skin electron therapy technique can be used for radical, palliative cases as well as adjuvant radiotherapy.
The features of the electron beams that make it a unique therapeutic tool are related to physical characteristics rather than to any special biological effectiveness of the electrons.
The most attractive characteristic in radiotherapy is the shape of the depth dose curve.
The curve displays a moderately flat in plateau in the first few centimeters of tissue, followed by a rapid fall in the absorbed dose to a small "tail" produced by x-ray. With high energy electrons the fall in depth dose after the initial plateau, is less steep. The advantage to be drawn from the depth dose pattern are, therefore, greatest at low energies, making the use of electrons for irradiation of sub-dermal tumour with the benefit of sparing the underlying tissues.
The characteristic which are of particular significance in clinical applications are: (i) The dose distribution from a single beam is such as to allow the treatment of the surface slab of tissue to relatively uniform doses whilst sparing underlying, deeper regions of healthy tissues.
(ii) The depth dose curve with electrons of lower energy offer rapid and simple treatment set-up, with the use of one field in many cases.
(iv)There is no difference in biological effectiveness of electrons compared with megavoltage photon radiation.
(v) The build-up of absorbed dose below the skin is rapid: thus the skin sparing effect is smaller than with high energy photons.

7
(vi) The dose distribution in tissue suffers perturbation if tissue inhomogeneities are presents with in the beam.
The principal applications of electrons are in (ICRU report 42, 1987): (i) The treatment of skin cancers.
(ii) Chest wall irradiation for breast cancers.
(iii) The treatment of head and neck cancers.
Although many of these sites can be treated with superficial x-rays, irradiation using electron beam offers distinct advantages in terms of uniformity of the dose in the target volume and in minimizing dose to deep seated tumours (Mohan, 1999).

Electron therapy treatment planning
The complexity of electron-tissue interactions does not make electron beams well suited to conventional treatment planning algorithms, because of their difficulty in modelling and predicting the dose for oblique incidence or tissue interfaces.
The early methods of electron dose distribution calculations were empirical and based on water phantom measurements of percentage depth doses and beam profiles for various field sizes, similar to the Milan and Bentley method developed in the late 1960s for use in photon beams (ICRU report 35, 1984). Inhomogeneities were accounted for by scaling the depth dose curves using the Coefficient of Equivalent Thicknesses (eET) technique (Khan, 2003). This technique provides useful parametrization of the electron depth dose curve but has nothing to do with the physics of electron transport that is dominated by the theory of multiple scattering. The Fermi-Eyges multiple scattering theory (Jette, 1983) considers a broad electron beam as being made up of many individual pencil beams which spread out laterally in tissue, approximately as a Gaussian function with the amount of spread increasing with depth. The dose at a particular point in tissue is calculated by an addition of contributions of spreading pencil beams. This algorithm can account for tissue inhomogeneities, patient curvature and irregular field shape.
Rudimentary pencil beam algorithms dealt with lateral dispersion, but ignored angular dispersion and back scattering from tissue interfaces. Subsequent analytically advanced algorithms refined the multiple scattering theory through applying both the stopping powers as well as the scattering powers but nevertheless generally failed to provide accurate dose distributions in general clinical conditions.
The most accurate way to calculate electron beam dose distributions is through Monte Carlo techniques. The main drawback of the current Monte Carlo approach as a routine dose calculation engine is its relatively long calculation time. However, with the everincreasing computer speed combined with the decreasing hardware cost, one can expect that in the near future Monte Carlo-based electron dose calculation algorithms will become available for routine clinical applications (Podgorsak, 2004).

Monte Carlo simulation techniques
Monte Carlo (MC) simulation is one of the most accurate methods available at the moment for obtaining the dose distribution due to a radiation beam. The method can precisely model the physical processes involved in radiation therapy and is powerful in dealing with any complex geometry (Ma et al, 1999 andJohns andCunningham, 1982).
The MC method is a statistical simulation method (Bushberg et al, 1994). It simulates the tracks of individual particles by sampling appropriate quantities from the probability distribution governing the individual physical processes using machine-generated random numbers. By simulating large number of histories, information can obtained about average value of macroscopic quantities such as energy deposition. Moreover, since one follow individual particle histories, the technique can be used to obtain information about the statistical fluctuation of particular kinds of events. It is also possible to use Monte Carlo to answer questions which cannot be addressed by experimental investigation, such as "what fraction of these electrons were generated in the collimator versus the filter" or "how often have certain photons undergo Compton scattering".
MC method consists of computer simulations that involve transport of a photon, or electron beams through a medium and calculating the deposition of energy within the phantom by using the laws of probability and the known physical characteristics (Nahum, 1988). The transport of an incident particle, and of the particles that it subsequently sets in motion, is referred to as a particle history and in MC each history is uniquely followed by random selection from the probability distribution that control each possible interaction (Metcalfe et al, 1997). The histories of a very large number of individual photons or electrons as they interact, scatter and eventually disappear are tracked (Rogers, 2002). Because the MC method requires modeling a stochastic set of events, the computer essentially rolls the dice to determine how each particles interacts and what the fate of that particle will be after the interaction (Nelson, 1988, Mohan, 1988. In contrast, MC simulation of photon transport is much faster compared to electron transport (Nahum, 1988). Photons on average undergo a moderate number (tens) of interactions and also the cross-section data needed for most applications are known to a high degree of accuracy (Andreo, 1991). While the electron transport it is timeconsuming to simulate each interaction individually because an electron undergoes a large number (thousands) of elastic scattering from nuclei during its history (Rogers et al, 1990). Moreover in the photon simulations, the electron transport consumes most of the computing time for high energies where the electron range is large (Mackie, 1990). This is because there are usually many short electron transport steps corresponding to each photon step (Awusi, 2000). Therefore the simulation of electron requires a different approach involving a combination of multiple scattering and stopping power theories.
Berger, 1963 first introduced the condensed history technique in which electron histories were "condensed" into a series of steps in which the effects of many scattering events were considered at once and a multiple scattering theory used to account for the elastic and inelastic scattering during this step (Rogers et al, 1990).

Advantages of Monte Carlo simulation
The main advantages the Monte Carlo method for the calculations of the dose distributions in a patient (Rogers 1991;Ma and Jiang, 1999;Andreo, 1991;Metcalfe et al, 1997), are: (a) Monte Carlo method can be accurately model the physics of radiation therapy transport and can be applied to any absorbing medium, geometry and radiation beam.
(b) Electrons and positrons produced in case of photon interactions can also be tracked.
(c) Information about macroscopic quantities such as particle fluence can be obtained.
(d) The method can be used to obtain the information that cannot be measured experimentally.
(e) The Monte Carlo methods can handle backscatter from high-density materials such as bone and scatter perturbations by air cavities more accurately than any other existing dose calculation model (Rogers and Bielajew 1990).
(f) The method can be predicate some of the experimental investigation, such as what fraction of the electrons was generated in the collimator versus the filter, or how often have certain photons undergone Compton scattering.
(g) The Monte Carlo can provide information such as fluence, energy fluence, energy spectra and angular distributions of the radiation beam which is almost difficult to measure.
(h) The Monte Carlo method allows the generation of the energy spectrum, not only in the central part of the beam, but also in regions away from it.
(i) Using Monte method it possible saving in manpower at the expense of computer time.
Monte Carlo simulation is therefore the method of choice for solving complicated recitation transport problems (Williamson, 1989).

Limitations of Monte Carlo simulation
(a) The method required a large number of histories to achieve adequate statistical uncertainty in the distribution. The lower the uncertainty the smoother the depth dose or cross beam profiles obtained from the distributions (Metcalfe et al, 1997  . Hopefully with these new developments the dose to be delivered to a patient will be calculated in a few minutes.

Monte Carlo simulation of linear accelerator head
There are several simulation codes available which can be used to simulate therapy units.
In all of the above simulation codes, the EGS4 code, is the most widely used MC code in medical radiation physics. The EGS4 code is written in MORTRAN language, which is based on FORTRAN, but has extensions to make it more flexible and easier to use (Metcalfe et al, 1997).
MC simulations of the radiation beam output for radiation treatment machine head, offer a practical means for obtaining energy spectrum and angular distribution of the photon and electron beam, which are important in radiation dosimetry (Nahum, 1988). The user has to set up the problem geometry, which includes arrangement and description of the various relevant components of the head, in a manner that can be understood by the computer program (Awusi, 2000). One of particular advantage of the BEAM code is the way it was designed and simplified in such a way that it can be accurately used to simulate treatment heads by other individuals with minimum effort .
The other advantage of the BEAM code is that the generated phase space files can be reused by the BEAM itself, allowing the user to simulate a treatment head output in separate steps to reduce CPU time.
In this study the Monte Carlo simulation codes, were used to simulate the 4 and 6 MeV electron beams from an Elekta Precise linear accelerator and to calculate the dose distribution in a phantom. The BEAMDP code was used for the analysis of the phase space files (PSF), as well as the CT based PhantomlCTCREATE program was used in the simulation of CT based models.

Aim of the study
The aim of this study was to commission and optimize a high-dose rate electron (HDRE) facility on an Elekta Precise linear accelerator for a Total Skin Electron Therapy (TSET) technique.

Specific objectives
The specific objectives required to achieve the above aim are as follows: (a) Measurements of reference beam data for the 4 and 6 MeVelectron beams in high dose rate mode in a water phantom at isocentre (100 cm SSD).
(b) Measurements of reference beam data for the 4 and 6 MeVelectron beams in high dose rate mode at the position of the treatment plane (350 cm SSD).
(c) Monte Carlo simulation of the Elekta Precise linear accelerator using the BEAM code to obtain phase space files at the isocentre for the 4 and 6 MeVelectron beams.
(d) Simulation of beam data in a water phantom using the phase space files and the DOSXYZ code and comparing the results to the measured reference data at isocentre, so that Monte Carlo simulation parameters can be optimized.
(e) Simulation of beam data at the treatment plane distance and comparing the results with the measured reference data to validate the simulation parameters.
(f) Simulation of a multiple beam treatment on a Rando phantom and comparing the results with film measurements to verify the accuracy of the simulation.
(g) Optimizing the treatment parameters by simulation of alternative treatment set-ups.

Introduction
The electron linear accelerator (linac) was developed at early 1950s by several different research groups (Metcalfe et al, 1997). The basic design of these machines is similar to a heavy-ion accelerator. The linac uses high frequency electromagnetic waves to accelerate charged particles such as electrons to high energies through a linear tube. There are several types of linear accelerator designs but the ones used in radiotherapy accelerate electrons either by traveling or stationary electromagnetic wave of a frequency 3 GHz, giving a wavelength = 10 cm in a vacuum (Khan, 2003).  (Khan, 2003). These pulses are delivered to the magnetron or klystron and simultaneously to the electron gun. Pulsed microwaves produced in the magnetron or klystron are injected into the accelerator tube or structure via a wave guide system. At the proper instant electrons, produced from the electron gun are also pulse injected into the accelerator structure. The accelerator structure consists of an evacuated copper tube with its interior divided by copper discs or diaphragms of varying aperture and spacing (Khan, 2003, Metcalfe et al, 1997, Johns and Cunningham, 1983. As the electrons are injected into the accelerator structure with an initial energy of about 50

Principles of operation
KeV, the electrons interact with the electromagnetic field of the microwaves. And the electrons gain energy from the sinusoidal electric field (Johns and Cunningham, 1983). As the High-energy electrons emerge from the exit window of the accelerator structure, they are in the form of a pencil beam of about 3 mm in diameter. In the low energy linear accelerators (up to 6 MeV) have relatively short accelerator tubes, the electrons are allowed to proceed straight on to strike a target for X-ray production (Khan, 2003). In the higher-energy linear accelerators, the accelerator structure is too long and, therefore, is placed horizontally or at an angle with respect to horizontal. The electrons are bent through an angle (usually about 90°or 270°) between the accelerator structure and the target using bending magnets, focusing coils and other components such that the beam emerges facing down wards (Khan, 2003, Johns andCunningham, 1983).

The Linac photon and electron beam
In the photon beam, after the accelerate of the electron beam to relativistic velocities within the linear accelerator wave guide they strike a target and photons with a broad energy spectrum forward peaked fluence are emitted due to bremsstrahlung production (Khan, 2003). X-rays are produced when high energy electrons are incident on a target of a high Z material such as tungsten. The target is thick enough to absorb most of the incident electrons and as a result the electron energy is converted into a spectrum of Xray energies (Johns and Cunningham, 1983). In the electron mode of an accelerator, this beam, instead of striking the target, is made to strike an electron scattering foil (usually of lead) in order to spread the beam as well as get a uniform electron fluence across the treatment field (Khan, 2003).

Beam collimation and monitoring
The treatment beam is first collimated by a fixed primary collimator located immediately below the X-ray target. In case of X-rays the collimated beam then passes through the flattening filter whose main function is to modify the forward peaked X-ray beam to a uniform beam and to filter the low energy X-ray spectrum (Khan, 2003). In the electron mode the flattening filter is moved away and replaced by a scattering foil whose main function is to spread the electron beam (Khan, 2003, Metcalfe et aI, 1997. The flattened X-ray beam or the electron beam is incident on the dose monitoring chambers, whose main functions are to monitor dose rate, integrated dose and field symmetry. After passing through the ion chamber, the X-ray beam is further collimated by a continuously movable collimator consisting of two pairs of lead or tungsten block jaws that provide a rectangular opening. For electron beams an applicator of appropriate size is used. The field size localizer is provided by a light source system in the treatment head (located between the ion chamber and the jaws) which is a combination of a mirror and a light source (Khan, 2003, Johns andCunningham, 1983).

Photon interaction processes
Attenuation of a photon beam by an absorbing material is caused by four interactions describe photon absorption in tissue: the photoelectric effect, Compton effect, pair production and coherent scattering.

Photoelectric Effect
The ohotoelectric effect is a phenomena in which a photon interacts with an atom an ejected one of the orbital electrons from the atom. In this process the entire energy, hy, of the photon is transferred to the atomic electron. The kinetic energy of the ejected electron (called the photoelectron) is equal to hv-EB, EB is the binding energy of the electron.
Interactions of this type can take place with electrons in the K, L, M, or N shells. Figure   2.2. shows the the Photoelectric effect phenomena.
Ejected Photo-electron

Compton Effect (Incoherent)
The Compton process, the photon interacts with an atomic electron as though were a "free" electron. The term "free" here means hat the binding energy of the electrons is much less than the energy of the bombarding photon. In this interaction, the electrons receives some energy from the photon and is emitted at angle e. The photon, with = hvoreduced energy, is scattered at an angleó (see figure 2.3).
The Compton process can be analyzed in terms of a collision between two particles, a photon and electron. By applying he lows of conservation of energy and momentum, one can drive the following relationship.

Pair production
In this process, a photon interacts with the nucleus of an atom, not an orbital electron.
The photon gives up its energy to the nucleus and, in the process, creates a pair of positively and negatively charged electrons. The positive electron (positron) ionizes until it combines with a free electron. This generates two photons that scatter in opposite directions. The probability of pair production is proportional to the logarithm of the energy of the incoming photon and is dependent on the atomic number of the material.
The energy range in which pair production dominates is ;:::;25MeV. This interaction does occur to some extent in routine radiation treatment with high-energy photon beams.

Coherent scattering
The coherent scattering, known as known as classical scattering or Rayliegh scattering, is

Electron interaction processes
As an energetic electron traverses matter, it interacts with the matter through Coulomb interactions with atomic orbital electrons and atomic nuclei. Through these collisions the electrons may lose their kinetic energy (collision and radiative losses) or change their direction of travel (scattering). Energy losses are described by stopping power; scattering is described by scattering power. The collisions between the incident electron and an orbital electron or nucleus of an atom may be elastic or inelastic. In an elastic collision the electron is deflected from its original path but no energy loss occurs, while in an inelastic collision the electron is deflected from its original path and some of its energy is transferred to an orbital electron or emitted in the form of bremsstrahlung (Podgorsak, 2004).
Energetic electrons experience thousands of collisions as they traverse an absorber; hence their behavior is described by a statistical theory of multiple scattering embracing the individual elastic and inelastic collisions with orbital electrons and nuclei.

Collisional energy loss
Collisional energy loss occurs when a secondary electron passes close enough to an atomic electron to eject it from it from it is shell either permanently or temporary.
Collisional energy losses in which the electron loses a small amount of energy are very frequent. The rate of energy loss by this mechanism depends on the electron energy, the number of the atomic electrons per unit volume, and slightly on the ionization energy of the atoms in the medium (Metcalfe et al, 1997). Less frequently, large energy losses occur when a much higher portion of electron energy is transferred to an orbital electron.
The ejected electron is known as a 8 ray, which itself causes ionization and excitation.

Incident electron
Bremsstrahlung photon

Radiative energy loss (bremsstrahlung production)
In these types of energy losses, Coulomb interactions between the incident electron and nuclei of the absorber atom result in electron scattering and energy loss of the electron through production of X-ray photons (bremsstrahlung). Figure 2.6 illustrates the mechanism of the bremsstrahlung production. The probability of this interaction increases as the distance of the electron from the nucleus is decreased. The energy of the Bremsstrahlung photon cannot be larger than the incident electron energy (Metcalfe et al, 1997).

Electron stopping powers
Stopping powers are widely used in radiation dosimetry, but they are rarely measured and have to be calculated from theory. The linear stopping power is defined as the dEI expectation value of the rate of energy loss per unit path length, IdZ of the charged particle. Stopping power can be divided into collisional stopping power and radiative stopping power. When the stopping power is divided by the density the ratios are called the mass collisional stopping power and mass radiative stopping power. The total mass stopping power has units of MeV ern" g-l. The total mass stopping power is given by (ICRU Report 37, 1984

Mass collision stopping power
The mass collision stopping power is (%LIresulting from electron-orbital electron interactions (atomic excitations and ionizations). The mass collision stopping power can be given by the formula: where; f3 is the velocity of the electron relative to the speed oflight c.
re is the classical electron radius Z is the projectile charge in units of electron charge I is the mean excitation energy.

Mass radiative stopping power
The mass radiative stopping power is the rate of energy loss by electrons that results in production ofbremsstrahlung. The radiative stopping power increases with higher atomic number and higher energy. The radiative stopping power is given by the following formula (Metcalfe et al, 1997): where E is the energy, A very approximate expression of the ratio of radiative to collision stopping power is given by the following formula, The equation predicts that the two stopping powers will be equal when the energy is approximately equal to 800/Z MeV.

Restricted stopping power
In radiation dosimetry the concept of restricted stopping power (~) is introduced which accounts for that fraction of the collisional stopping power (%)colthat includes all the soft collisions (in which only a small amount of the incident particle energy is transferred to the secondary particles, in the form of excitation energy) plus those hard collisions (in which a large fraction of the incident particle energy is transferred to a secondary electron causing what is called a "delta ray"), which result in delta rays with energies less than a cut-off value 1:1.
The restricted stopping power (also referred to as a linear energy transfer) Lt:. of a material, for charged particles, is the quotient of dE tJ by dl., where dEtJ is the energy lost by charged particles due to collision in traversing a distance dL minus the total kinetic energy of charged particles released with kinetic energy in excess of~: The restricted stopping power is the restricted linear collision stopping power divided by the density of the material.

Range concept (CSDA)
A charged particle such as an electron is surrounded by its Coulomb electric field and will therefore interact with one or more electrons or with the nucleus of practically every atom it encounters. Most of these interactions individually transfer only minute fractions of the incident particle's kinetic energy and it is convenient to think of the particle as losing its kinetic energy gradually and continuously in a process often referred to as the continuous slowing down approximation (CSDA).
The CSDA range (or the mean path-length) for an electron of initial kinetic energy Eo can be found by integrating the reciprocal of the total stopping power: Where, Eo is the starting energy of the particle.
The CSDA range thus represents the mean path-length and not the depth of penetration in a defined direction.

Radiation dosimetry
This thesis considers the dosimetry of a Total Skin Electron Therapy (TSET) technique.
It is therefore appropriate, before we proceed any further, to define radiation "dose". The absorbed dose, D is defined as energy absorbed in a medium per unit mass or (Johns andCunningham, 1983 1983).

2.10
The dose can also be derived from a particle fluence. The fluence, <1>, is defined as the number of particles crossing a surface of unit area perpendicular to the direction of motion. The energy fluence, If', is sometimes used and this is the total energy passing through a surface of unit area.

2.11
Differential fluence, <l>E,U' is an element of fluence possessing an energy, E, and a trajectory, U. Often it is sensible to separate a fluence into such components. When we examine a beam spectrum or the in-air distribution of a point source, we are referring to a differential fluence.
When a particle interacts with the medium, it does so with probability per unit length, Jl, and _!_ is the mean free path between interactions. If Ea,b is the mean energy absorbed by Jl the medium, then the absorbed dose is defined as: Dose can also be calculated from the divergence of the vectorial energy fluence, 'P, (Rossi and Roesch, 1962), and this is the manner in which the fluence transport equation is used to generate analytic models of dose distribution.

Total skin electron therapy
Skin cancer is rarely a fatal disease (a notable exception being melanoma). The majority of lesions grow slowly and 90 per cent of the lesions arise in exposed areas. They are usually diagnosed in an early stage of development. These tumours are often readily curable and therefore the selection of the optimum treatment modality should be considered with respect to the expected effects, the relative comfort, time and cost of treatment of the patient, as well as upon the probability of cure. Curability of the common skin tumours by competent surgery is not questioned by reasonable radiation therapists nor should it be challenged (Walter, 1967). 25 Radiation therapy has a special role in the palliation of the widespread cutaneous lymphomas and some multicentric diseases such as Kaposi sarcoma. Some tumours are radioresistant or are located in sites that tolerate irradiation poorly. These include the position of the extremities. Ionizing radiation is preferred as it has a high probability of eradicating the lesion and preserving normal tissues. All this can be achieved with minimal time and cost to the patient. This demands a good clinical knowledge of anatomical factors, radiobiology, and a judicious computation of all the physical factors in radiation therapy, including treatment volume, quality of irradiation, dosage, overall treatment time and fractionation (Walter, 1967).
Malignant skin diseases, such as mycosis fungoides and cutaneous lymphomas (La TCM, et al 1979, Richard, 2003, are often treated with nitrogen mustard and Photo Ultra Violet type-A (PUV A), but the most effective treatment is Total Skin Electron Therapy or TSET (Richard, 1997). TSET (AAPM, 1988) is an external beam therapy. It is a complex technique for which special irradiation and dosimetry conditions have been studied based on the particular methods implemented (AAPM, 1988). Technical challenges in setting up a TSET program arise primarily from the unusual target volume of the disease that often includes the whole-body skin surface extending to a depth of about 5 mm. Because of the shallow depth of the disease, low-energy electrons that have a limited penetration are the favoured choice of radiation source. The goal is then to deliver a relatively uniform dose (e.g., ±10%) to the skin of the entire body amid the ever-changing curvature of the body surface and the unavoidable self-shielding among the body structures. In addition, the X-ray contamination, produced by the inevitable interactions of the electrons with materials in the beam path, has to be kept low to prevent serious radiotoxicity arising from whole-body X-ray exposure (Chen et aI, 2004). A high dose rate is required in order to treat the patient at 3 to 4 meters SSD in the time that a conventional field would be treated at the isocentre. For TSET, in order to achieve a flat profile in the vertical axis, two beams may be combined such that the 50% point of one beam overlaps the 50% point of the other (Richard, 1997).
Monte Carlo (MC) methods have been widely used to design radiotherapy beams because of their accuracy and efficiency in estimating the performance of various designs under consideration (Sung et al, 2005). It is well recognized that MC dose calculations are the most accurate way of computing patient relative dose delivery. Nevertheless, to achieve reliable results, some MC experience as well as the use of a powerful computing facility is needed (Faddegon et al, 1998, Ma andJiang 1999).  (Pavón, et al, 2003, Sung et al, 2005).

Irradiation beam requirements
When using electron beams, the most common situation would be to treat superficial tumours according to certain conditions and requirements of the radiation beam. These requirements involve characteristics of the treatment electron beam, the disease entity and the patient population. They include specification of: field size, penetration, energy, dose, dose rate, field flatness in the treatment plane, x-ray contamination, and the need for and nature of boost fields. The central requirement is to treat virtually the entire body surface to a limited depth and to a uniform dose using electrons with a low x-ray contamination.
These requirements are coupled with the varied obliquity of body surfaces, beam directions, patient self-shielding, etc, (Hoppe et al, 1979). The required penetration depth of TSET treatment is usually varied with the stage and type of the disease and as well as the body surface. This penetration depth ranges from approximately 5mm to 15 mm or more at the 50 percent isodose surface which encompasses most lesions. It appears advantageous to provide more than one TSET beam energy to cover this range of depth. The electron beam incident on the exit window of the accelerator can be characterized by a relatively narrow distribution of energy fluence whose peak is termed the accelerator energy, Ëa (see fig 2.8). As the beam passes through the exit window and different materials between the exit window and the phantom surface, the energy will decrease and the energy spread will increase. The energy fluence distribution of such a beam arriving at the treatment plane (phantom surface) is characterized by its peak, or most probable energy, E P,o' and a lower mean energy, Eo' The value of E P,o can be obtained by subtracting the most probable energy loss in the energy-degrading materials traversed from the accelerator energy, Ea' or from the range-energy equation given below. In this 2.20 low energy range, the most probable energy loss for the low-energy TSET electrons is just the mean collision ionization energy loss for an electron of energy E (Khan, 2003).
The range-energy relationship:

2.19
is used to relate the most probable energy at the phantom surface, Ep,o' in MeV, to the practical range, Rp, in cm of water. The mean energy at the phantom surface (treatment plane), Eo, in MeV is related to the half-value depth, Rso, in cm of water by: The treatment beam traversing the patient or phantom further degrades and spreads out in energy. lts mean energy can be estimated as a function of depth, z, and the mean entrance energy, Eo, by the equation:

2.21
As noted, equation (2.19) is used to relate the most probable energy at the surface, E; 0 , in MeV, to the practical range, in cm of water. The mean energy at the phantom surface (treatment plane), in MeV is related to the half-value depth and is usually in the range 3 to 7 MeV with accelerator energies, Ëa, ranging from about 4 to 10 MeV. Occasionally, lower energies have been employed. Most irradiation techniques involve significant electron energy loss from the sequence of materials traversed by the electron beam, as much as several MeV between the accelerator vacuum and the patient treatment plane (AAPM, 1988). Often, there are body areas shielded in part by other body sections or inadequately exposed because of limitations of the geometry of the treatment technique. Small supplementary boost fields of electrons or orthovoltage x-rays are therefore frequently needed. The accompanying megavoltage x-ray contamination is penetrating and forward directed; it often exposes much of the body volume and should be as low as reasonably achievable. It is roughly proportional to the number of fields used since all fields contribute penetrating x-rays; often it can be estimated prior to the selection of the technique. The average x-ray dose can be reduced by angling the beam axes so that the peaks of the forward-directed x-rays lie outside the body. A desirable x-ray contamination level averaged over the body volume is 1% or less of the total mean electron dose at dose maximum. This may be difficult to achieve with some equipment and techniques. Most TSET procedures are time-consuming to carry out because of the Average dose rates from 0.25 to several grays per minute at the depth of dose maximum are used, with the lower end of this range usually considered only marginally acceptable.
Some patients require physical support devices to ensure their safety as well as correct positioning in a standing position. Radiation shielding of specific anatomical surfaces or organs may also be required. Commonly, finger and toe nails, tops of feet, and the eyes are protected during at least part of the treatment, with the use of shielding being dependent on the extent of disease (Karzmark et al, 1988).

Irradiation room requirements
Providing good dose uniformity over the height and width of a patient usually necessitates the use of large distances between scatterer and patient, typically 2-7 meters, with the distance being technique dependent. Hence existing treatment room layouts may restrict the choice of a TSET technique. The TSET procedure involves significant ozone production from ionizing large volumes of air in the treatment room. Frequent exchange of the air in the treatment room is essential for confining ozone exposure to acceptable limits. For most installations, the shielding provided by megavoltage x-ray treatment rooms has been found adequate for TSET therapy, which involves bringing a large fluence of energetic electrons into the treatment room. However, measurements must be made to ensure that radiation protection for TSET is adequate (Karzmark et al, 1988).

Irradiation techniques
The patient population requiring TSET is relatively small; therefore, the technique is available only in major radiotherapy centers. Prior to the use of electron beams, lowenergy x-rays were used for total skin electron therapy. They presently have limited usage. The clinical results using a variety of such x-rays were less than encouraging because it was difficult to treat the entire skin area adequately. There was maximum field-size and field-junction limitations, and it was not possible to treat to an adequate depth without a large x-ray integral dose.
During this period, a number of TSET techniques adapted to the equipment available have been developed. Historically, machine-producing electrons have been used with an accelerator energy range, Ea, from 1.5 MeV to 10 MeV for TSET. The Van de Graaff generator, which was the first accelerator employed for TSET, has been largely supplanted by the isocentrically mounted electron linac. Electron beams from accelerators show the typical characteristics of a dose maximum occurring just below a normally incident skin surface and a rapid fall-off of dose with depth to maximum range determined by the incident electron energy (Ekstrand and Dixon, 1982).
There were different groups of techniques used in TSET treatment, such as translational techniques, in which the patient is placed in a horizontal position and is then translated with respect to a beam whose dimensions cover the patient laterally. In large field techniques, the patient is usually irradiated standing upright in the path of one or more fields, aiming to cover the upper and lower parts of the body. A number of large-field treatment techniques have been developed, some of which are very complex and timeconsuming (Christina, 2005). However, the Stanford technique which utilizes a large horizontally directed electron beam produced by a medical linear accelerator and treats patients at extended distance in six standing positions has been adapted as the standard technique for TSET treatment (Richard, 1997).

Beta particles
In this technique the patient lies on a motor-driven couch and moves relative to a downward-directed beam at a suitable velocity. Beta particles from radioactive sources, such as 90Sr, provide an alternative electron source which are preferred because of their wide spatial divergence, broad spectrum of energies and low average energy (1.12 MeV) and have a limited penetration depth in tissue (Haybittle, 1957, Proimos, et al, 1960. A 24-Ci 90Sr jJ source in the form of a 60 cm linear array is used. The source is contained in a source shield housing and is positioned above the couch. The maximum energy of beta particles emitted by 90Sr is 2.25 MeV. However, due to the spectral distribution of jJ -ray energies the effective depth of treatment In this case IS only a fraction of millimeter. In a beta-particle unit, the 24 ei source is spread over an area 53 cm long by 2 cm wide (Haybittle, 1957), a treatment distance of 40 cm was used, and the source was arranged horizontally with its long axis perpendicular to its direction of motion as it traversed the length of the recumbent patient. Although beta particles have been successfully employed for TSET, the majority of patients are treated with electrons from accelerators at this time. Long exposure times, lesser average penetration associated with their energy spectrum and poorer uniformity characterize beta-particle treatments. High output and the variable electron energy feature of linacs have led to their increasing adoption for TSET.

Narrow rectangular beams
In this technique, the Van de Graaff accelerator is in a fixed position with vertically downward narrow rectangular beams and patients are translated on a motor-driven couch placed under the electron beams (Williams et aI, 1979). The energies used in this technique are about 1.5 to 4.5 MeV.
Another Van de Graaff TSET technique, made use of a wide cone with the beam scanned magnetically in vacuum transversely in the X direction while the patient is moved longitudinally under the beam in the Y direction. The dose distribution across the beam in a treatment plane was uniform to an extent dependent on the distance below the cone but at least as good as ± 5%. The energy of the Van de Graaff accelerator was adjusted to control the depth of penetration for treatment. Treatment times were about one minute for each full length pass and less for small treatment areas (Andrews and Swain, 1957).

Large electron field techniques
In order to achieve good uniformity in TSET treatment, the electron beam should cover the whole length of the patient. This requires a large electron field, which is produced by scattering electrons through a wide angle and using large treatment distances. The field is made uniform over the height of the patient by vertically combining multiple fields or vertical arcing. The patient is treated in a standing position with four or six fields directed from equally spaced angles for circumferential coverage of the body surface.

Scattered single beam
Different techniques have been employed to achieve uniform skin treatment by using a large electron field. The large electron beam can be produced by scattering a single field or parallel fields. Patients treated with a single electron field at extended distance showed excellent clinical results. A scattered single electron beam technique employing alinac for a standing stationary patient (Tetenes and Goodwin, 1977), in order to obtain a flattened beam with an electron energy of 4 MeV at the treatment plane, and initial accelerator energy of 6.5 MeV is used with a titanium scattering foil 0.15 mm thick placed 10 cm from the accelerator exit window. A shaped polystyrene scatterer beamflattening filter is mounted on the front of the treatment head with a distance of 7 meters between the accelerator beam exit window and the treatment plane. The measured transverse uniformity in the treatment plane for this technique was ± 1% within a 40 cm radius around the central axis and within ± 8% for a 200 cm diameter circle. The maximum dose rate at the treatment plane with both the normal linac scatterer and the added scatterer in place was 3 Gy/min.

Pair of parallel beams
In this technique, two horizontal parallel beams are used and their axes are contained in a vertical plane at a treatment distance of about 2 meters. The technique was developed for an 8 MeV linear accelerator and includes the use of carbon energy degraders located just beyond the exit window of the accelerator. By using different thicknesses of carbon degraders, the depth of penetration was adjusted from about 2 to 25 mm to meet the requirements ofthe individual patient. Energy degraders (decelerators) produce less-rapid fall-off of depth dose, as well as a reduction in the beam energy; two horizontally directed beams, with a central axis vertical separation of 150 cm, were used to obtain ± 5% uniformity for a treatment plane 200 cm high. The X-ray dose was about 2% of the

Patient rotation
peak value for each field when using a 2 cm thick carbon decelerator. For thinner decelerators, the integral dose from electrons increased, but that due to x-rays showed little variation (Szur et al, 1962).

Pendulum-arc
An isocentrically mounted 8 MeV linac has been used in this technique (Sewchand et al, 1979). The accelerator is rotated continuously during treatment in a 50°arc about the isocenter starting from an initial angle with the beam axis aimed below the feet to a final angle with the beam aimed above the head of the standing patient. It may be feasible to vary the dose rate, or gantry rotation speed at constant dose rate, automatically, as a function of gantry angle so as to vary the dose rate and hence, optimize the dose uniformity in the vertical direction. A large Plexiglas sheet 1 cm thick placed 5 cm from the patient is used to degrade the beam energy further and provide large-angle electron scattering near the patient skin. A six-arcing-field technique is described with the total xray dose measured at 10 cm depth equal to 4.2% of the average electron dose at the depth of maximum dose, Dmax.
Studies of treatments involving patient rotation about a vertical aXIS for total skin irradiation include the work of Tetenes and Gooddwin, 1977, Podgorsak et al, 1983and Kumar, 1978. These groups use a single horizontal beam, the first with a single scatterer located near the beam exit window and a 7-meter treatment distance. The latter two groups have a first scatterer placed near the beam exit window and a second large planar scatterer located 20 cm from the treatment plane, which is located 265 cm and 3 m, respectively, (Kumar andPatel, 1978 andPodgorsak et al 1983) from the beam exit window. Podgorsak et al, 1983, have developed analytical expressions for rotational dose distributions using stationary depth-dose data and a variety of phantom and patient cross sections. The calculated and measured dose distributions show close agreement. With an accelerator electron energy of 6 MeVand a depth-dose curve equivalent to 3.5 MeV in the treatment plane, the x-ray background amounted to 4% compared to 2.2% for the Tetenes and Goodwin method.

Stanford technique (rotational technique)
In the Stanford technique, the patient is treated with six fields (anterior, posterior, four oblique) positioned 60 degree apart around the circumference of the patient. Each field is made up of two component beams, oriented at a suitable angle with respect to the horizontal. The patient treatment positions for the full six-dual field treatment cycle, is shown in figure 2.9. The Stanford technique of six -dual fields, requires modification of the accelerator such as removing the scattering foil and installing a scatterer at the front of the collimator. These changes would require safety interlocks to prevent operation of the accelerator in this configuration for conventional electron beam treatments.
Most institutions have adopted the Stanford technique In principle without making alterations in the accelerator hardware. Because the regular scattering foils and various interlocks are left in place, no special precautions are required in preparing the machine for Total Skin Electron Therapy. In some accelerators a high dose rate mode is installed to allow an output of more than 2000 monitor units per minute. This significantly speeds up the treatments. Because conventional electron cones are not used, the electron field is collimated by a special wide-aperture insert attached at the end of the collimator. It is preset via interlock to a wider jaw setting and specific electron energy, selected for high dose rate mode of operation.
To shorten the treatment time, the patient is treated with three dual fields per day, for example day 1: one dual field from the anterior, two dual oblique fields from the posterior; day 2: one dual field posterior and two dual fields anterior oblique. A complete cycle of six dual fields is thus completed in two days. A source-to-patient distance of about 4 m is sufficient for this technique (Khan, 2003).

Dual field angle
A low energy electron beam is considerably widened in size by scattering in air. As an example, a 9 MeVelectron beam (of unspecified initial width), after transversing 4 m of air and an acrylic scatter plate, attains a Gaussian dose profile measuring a 90% to 90% isodose width of about 60 cm, which is usually sufficient to cover a patient's width.
Along the height of the patient, two fields, one directed toward the head and the other toward the feet, are angled such that in the composite dose distribution a ± 10% dose uniformity can be obtained over a length of about 200 cm (Khan, 2003).
A method of determining dual field angle by film dosimetry has been described by Khan, 2003. A series of dosimetry films in their jackets are mounted in a vertical board, larger than the height of a typical patient, and are positioned at the treatment distance. The scatter plate is placed in front of the films as in actual treatment. The films are exposed to a single electron field directed at a 10 degree to 15 degree angle with respect to the horizontal axis. The films are scanned for an optical density profile in the vertical direction. The profile is then placed side by side with its mirror image and separated by a distance such that the combined profile shows not more than ± 10% variation within about 200 cm. The separations between the two profiles gives the desired angle between the dual fields. A confirmatory composite profile is then measured by exposing the films to the dual fields with the interfiled angle determined above.

Calibration
A thin window (~0.05 g/cm'') plane-parallel chamber is a suitable instrument for measuring the depth dose distribution for the low energy beams used for this technique.
Because plane-parallel chambers are presently not calibrated by calibration laboratories, they may be calibrated by intercomparsion with a calibrated Farmer-type chamber, using a high energy (;::::10 MeV) electron beam.
The AAPM, 1988 recommends that the total skin electron therapy dose be measured at a

2.25
The skin dose for a water phantom, (Ds)w can be calculated from equation 2.22 and 2.23.
The factor B can also be determined by taping a film strip in its light-tight paper to the surface of a cylindrical polystyrene phantom, and exposing it to a single dual field.
Another film strip taped around the phantom is exposed to six dual fields. By relating the optical densities to doses in the two cases, factor B can be determined from equation 2.23.
The composite depth-dose distribution for the six dual fields may be determined by sandwiching dosimetry film (in its paper jacket) in the cylindrical polystyrene phantom and cutting the excess film so that the edges conform to the circular surface of the phantom. A black tape is wrapped around the phantom over the film edges to make the film light-tight. The phantom, with the film parallel to the horizontal axis, is exposed to the six dual fields, duplicating the actual treatment conditions. After appropriate processing, the film is scanned for optical density distribution, which is related to the dose distribution by a reference sensitometrie curve.

In vivo dosimetry
Although an overall surface dose uniformity of ± 10% can be achieved at the treatment distance, in a plane perpendicular to the horizontal axis and within an area equivalent to a patient's dimensions, there are localized regions of extreme nonuniformity of dose on the patient's skin. Excessive dose (e.g., 120%-130%) can occur in areas with sharp body projections, curved surfaces, and regions of multiple field overlaps. Low-dose regions occur when the skin is shielded by other parts of the body or overlapping body folds.
From in vivo measurements, areas receiving a significantly lower dose can be identified for local boost. If eyelids need to be treated, internal eye shields can be used, but the dose to the inside of the lids should be assessed, taking into account the electron backscatter from lead.
Thermoluminescent dosimeters (TLD) are most commonly used for in vivo dosimetry.
For these measurements the TLD must be thin «0.5 mm) to minimize the effect of dose gradient across the dosimeters. TLD chips are commercially available that meet these specifications. These chips can be sealed in thin polyethylene sheets to avoid contamination. The reference chips may be calibrated in a polystyrene phantom using an electron beam of approximately the same mean energy incident on the TLDs as in the in vivo measurement conditions. The desired mean energy may be obtained by selecting an appropriate incident beam energy and depth (Khan, 2003).

Dosimetrie parameters in large field techniques
Dosimetrie characteristics of TSET beams are examined for a single horizontal beam, a single angled dual-field beam, and the full array of six dual-field beams. These parameters include depth dose, isodose distributions, field flatness in the treatment plane and X-ray background.

Field flatness
Low energy electron beams are considerably widened by scattering in air. For example, a 6 MeV 20x20 cm 2 electron beam, after passing through 4 m of air, achieves a Gaussian intensity distribution with a 50% to 50% width of approximately 1 m (Holt and Perry, 1982). This usually gives adequate uniformity over the patient width. If two such fields are joined together vertically at their 50% lines, the resultant field will be uniform over a height of approximately 1 m. A proper combination of more such fields or a continuous arc can lead to a larger uniform field, sufficient to cover a patient from head to feet. The size and shape of an electron beam developed at a distance by air scatter can be estimated by multiple scattering theory. Holt and Perry, 1982, have used this approach to obtain a uniform field by combining multiple profiles in proper proportions and angular separation. In addition to air, the electron beam is scattered by the scattering foil inside (or sometimes, by design, an additional scattering foil outside) the collimator. However the x-ray contamination would be increased, because unnecessarily wide beams waste electron flux to the sides.

X-ray contamination
X-ray contamination is present in every therapy electron beam and becomes a limiting factor in total skin electron therapy. Ordinarily, these X-rays are contributed by bremsstrahlung interactions produced in the exit window of the accelerator, scattering foil, ion chambers, beam-defining collimators, air, and the patient. The Bremsstrahlung level can be minimized if the electron beam is scattered by air alone before incidence on the patient. This would necessitate some modifications in the accelerator, such as removing the scattering foil and other scatterers in the collimation system. Various safety interlocks would be required to make this separation feasible for routine clinical use (Khan, 2003).

Dose distribution
The depth-dose distribution in a single large field incident on a patient will depend on the angle of incidence of the beam relative to the surface contour for an oblique beam, the depth dose curve and its dmax shift toward the surface. When multiple large fields are directed at the patient from different angles, the composite distribution shows a net shift with apparent decrease in beam penetration. This shift of the relative depth dose closer to the surface has been explained by Bjarngard et al, 1977. as been due to the greater path length taken by obliquely incident electrons in reaching a point. Although dose uniformity of ± 10% can be achieved over most of the body surface using the six field technique, areas adjacent to surface irregularities vary substantially due to local scattering. Areas such as the inner thighs and axillae, which are obstructed by adjacent body structures, require supplementary irradiation. The total bremsstrahlung dose in the middle of the patient for the multiple field technique is approximately twice the level of a single field. This factor of two has been experimentally observed by a number of investigations.

Introduction
In the previous chapter the interaction processes of the electrons and photons were discussed. Although the physics of the interactions are well understood, in general it is impossible to develop an analytic expression to describe particle transport in a medium.
This is because the electrons can create both photons (e.g., as bremsstrahlung) and secondary or knock-on electrons (ê-rays) and conversely, photons can produce both electrons and positrons. In addition, both electrons and photons scatter a great deal. MC techniques are becoming more and more widely used. In general this is because the cost of computing continues to decrease dramatically. In medical radiotherapy physics this increase in use is also because of availability of general purpose and specialized code systems, such as EGSnrc (Rogers, 2002) and BEAMnrc .

The EGS4 code
The EGS (Electron-Gamma-Shower) code system for MC simulation of electron and photon transport (EGS4) is a FORTRAN based, electron and photon transport, Monte Carlo code (Rogers, 2002). The routines making up the package are coded in MORTRAN language, a macro extension to FORTRAN language (Richard, 1997).
Macros are used as a faster method of coding frequently called geometry routines and the random number generator. EGS4 is comprised of internal and user oriented routines.

3.1
The EGS4 internal routines deal with specific elements of the particle transport. Standard EGS4 contains algorithms for photons to determine the consequences of Compton scatter, the photoelectric effect and pair production. For electrons bremsstrahlung, Moller scatter, Bhabha Scattering, annihilation events and multiscattering/ condensed history steps are considered. The EGS4 preprocessing package (PEGS4) calculates the relevant cross section data by reference to a suitable input file that contains among other variables the atomic composition, physical density and energy range over which the data must be generated for photons and electrons. The state of materials can be mixtures and compounds and can be generated of arbitrary density. The input requires setup for each material through which transport simulations are to be carried out. PEGS4 generates cross section data and the hatch routine places this data in accessible arrays for the EGS4 photon and electron interaction routines to use (Richard, 1997).

Random numbers generator (RNG)
Monte Carlo calculations attempt to stochastic nature of particle-particle interactions by sampling in a random fashion from known particle interaction cross-sections. This requires a random number generator (RNG) capable of producing a sequence of truly random numbers. A significant proportion of the computing time in a simulation is spent generating random numbers so it is important that the code used for generating these numbers is efficient (Metcalfe et al, 1997). For the EGS4 code, the random number generator is of the multiplicative congruential type. The nth random number is found recursively using the expression: where a is a constant multiplier and K is the integer word size ofthe computer, "mod" is the modulo function. The first number in the sequence, X o , is called the seed and is usually specified by the user.

Electron transport
As an electron slows down in a material, it can undergo hundreds of thousands of scattering events and a large number of interactions. In order to simulate its energy loss and scattering through a medium efficiently a different approach is needed to account for energy loss and scatter (Metcalfe et aI, 1997). The EGS4 code utilizes the condensed random walk method. The electron loses energy in a continuous fashion by combining its interactions. The appropriate scatter angle is sampled from electron elastic multiplescattering distribution functions (Nelson et al., 1988).
In an actual MC simulation the user must define the lower bound of the electron's kinetic energy transport by specifying the variable ECUT. An ESTEPE variable is also specified that indicates the maximum fractional energy loss an electron can experience ). An algorithm called PRESTA was also introduced that optimizes the accuracy of energy deposition with simulation speed Rogers, 1988 andNelson et al, 1988).

Running the EGS4 code
The beauty of the EGS4 system is that a structured set of subroutines handles all of the physics in the simulation in a manner which allows users to write their own geometry and scoring routines without actually altering the EGS system itself. The user is responsible for writing the main program that contains two subroutines, HOWF AR and AUSGAB (Rogers, 2000). HOWFAR determines the geometry of the simulation and AUSGAB determines the dose scoring zones. These subroutines have been generalized in user friendly codes such as DOSXYZ and BEAM that allow the use of these powerful EGS4 transport simulation codes without any knowledge of MORT RAN.

Introduction
The BEAMnrc code has been designed to simulate the radiotherapy beams from any radiotherapy sources, including low energy x-rays, 60Co units, and both electron and photon beams from accelerators . BEAMnrc is based on the PRESTA algorithm extension of the EGS4 / MC code system for simulating radiation transport.
The BEAM model can be build from a series of individual component modules (CMs), each of which operate completely independently of the other component modules and contains between two planes which are perpendicular to the z-axis and which can not overlap (Rogers et aI, 1995). The beam axis is usually defined from the origin of the center of the beam as it exits from the accelerator window. The CMs in the BEAM code can be used for modeling the accelerator components such as; exit window, scattering foil, primary collimator, ionization chamber, mirror, jaws, etc. For each CM, all the geometrical and physical properties of the materials are specified at run time.

Running the BEAM code
There are three main stages are involved running the BEAM code (Rogers et aI, 2005):

a) "Specifying" Accelerators:
Before compiling and running a BEAM accelerator simulation, user must "specify" which component modules are to be used and in what order. The user may use the same CM as often as needed, as long as the identifiers used are unique. The user also has to specify the accelerator model via giving it a name representing the machine they are modeling.

b) Building an accelerator:
Once the accelerator has been specified, it must be "built". Building an accelerator consists of gathering all the source code, automatically editing it to avoid duplicate names (Awusi, 2000). The user has to specify the modules to use.

c) Compiling an accelerator:
Once the accelerator specified and built, the accelerator must be compiled. This required the MORTRAN and FORTRAN code. The user has to specify the name of the specific module they are running and once that is done the user is presented with a compilation menu with several options to select and modify before BEAM is compiled.
At this stage of the simulation with the BEAM code, some of the main tasks of the user are to: (a) Specify the geometry of the accelerator, which are the dimensions and materials of the treatment head.
(b) Specify the variance reduction techniques.
(c) Specify the transport parameters (the PCUT and ECUT values).
(d) Specify the physical properties of each material to be used in the simulation.
(e) The track of the particle's history by using the LATCH feature.
(f) Define the location of the scoring planes of interest to the user.

Efficiency and variance reduction techniques
The efficiency of a particular simulation is inversely proportional to both the variance of the results and the time taken for the simulation (Metcalfe et aI, 1997). If the uncertainty of a result is 0' and the computing time used in the simulation is t, then the efficiency e is defined as: where K is a proportionality constant.
since t is directly proportional to N, by assuming ê to be a constant, then equation 3.2 can be rewritten as where k=.JK!ê is another constant.
Since the fractional uncertainty is inversely proportional to the square root of the incident numbers, it is also inversely proportional to the square root of the simulation time. This means that efficiency is invariant with simulation time for a particular simulation type.
It is possible to increase efficiency by using variance reduction techniques, where either, (a) the variance for particular simulation time is reduced or (b) the time required to obtain results with in a certain variance is reduced (Metcalfe et aI, 1997). There are various techniques for variance reduction (Bielajew et al, 1988), but those employed by the BEAM code are: i) Range rejection: Range rejection can save significant quantities of computing time for electron transport calculations. The basic method is to calculate the residual range of a charged particle and terminate its history if it can not escape from the current region (Rogers et aI, 1995). The BEAM code includes a subroutine, which is used to pre-compute the splitting were not used .
iii) Photon forcing: The BEAM code offers an option whereby the user can force photons to interact in specified Clvls within a simulation. One of the main purposes of implementing this option was to study the generation of contaminated electrons in a photon beam. In BEAM, photons can be forced to interact in any subset of a component module .

Phase space files
One of the most important outputs of the BEAM code simulation is the phase space file (PSF). A phase space file contains data relating to particle position, direction, charge, etc.
for every particle crossing a scoring plane (Rogers et aI, 2005). The PSFs created can be analyzed by another program BEAMDP.

BEAMDP program
Another important program for processing phase space files is the BEAMDP program.
The BEAMDP (BEAM Data Processor) is a subsidiary of the BEAM program (Ma and Rogers, 2005). BEAMDP helps the BEAMnrc users to analyze the beam data obtained by the Monte Carlo simulation of the coupled transport of photons and electrons in a clinical accelerator and to derive the data required by the simplified sub-source models of these electron beams for use in Monte Carlo radiotherapy treatment planning.
When running BEAMDP, the user is having several options to analyze from the phase space file; some of them are: Usually a plot of particles per bin versus angle in degrees is given.
Also the user has to describe the input parameters e.g., the file name, field type, particle type, type of graph, scoring regions for spectrum and angular distributions, energy ranges and range of angles for angular distributions.
The BEAMDP program analyzed the phase space data and generates energy distributions for particles inside and outside the treatment field.

DOSXYZ code
The DOSXYZ is an EGSnrc based Monte Carlo simulation code for calculating a 3D absorbed dose (Walters et al, 2005). The co-ordinate system is cartesian. The geometry consists of a rectilinear volume with perpendicular x, y and z directions. The dimensions of the voxels are completely variable in all three dimensions. The material in each direction can be specified and its density can be varied. The code allows sources such as a monoenergetic diverging or parallel beam, phase-space data generated by a BEAM simulation, or a model-based beam reconstruction produced by BEAMDP. DOSXYZ also has parameter ISMOOTH that can be used to redistribute PSF particles when used more than once during a simulation.

CT Based Phantoml CTCREATE
The CT phantom is an option of DOSXYZ code that allows calculation of dose distributions in phantoms that are derived from CT data sets (Waiters et al, 2005). This allows simulations in realistic anthropomorphic phantoms. At this point in time, the system fully supports data in sets in the ADAC pinnacle format, the CAD PLAN format and DICOM format. It is possible to generate a CT phantom from CT data sets since all the required material and geometry information is contained in the data set.
There are several input parameters that the user has to specify in order to use the CT Phantom/ CTCREATE program: (a) The Format of the CT data.

52
(c) The number of the materials and ramps that are to be read from the file.
(d) The transport parameters e.g. the ECUT and PCUT values.
After the successful data input of CT PhantomlCTCREA TE, the appropriate CT phantom information is stored in "file.egs4phant", where "file" is the original CT data set name.
This CT phantom file created can then be read in and used by the DOSXYZ program to calculate the dose distribution in each voxel of the phantom. The CT phantom option in DOSXYZ is used by setting the number of materials (nmed) of the DOSXYZ input file to zero, which causes the program to execute differently than when nmed > O. Instead of geometry, material and density data being input in the DOSXYZ input file, DOSXYZ reads data from a CT phantom file that has been created using CTCREATE (Awusi, 2000 .

Introduction
In this study the Elekta Precise linear accelerator with High Dose Rate Electron (HDRE) mode was investigated for TSET treatment. This accelerator has three x-ray energies ( In this chapter, in section (4.2) the HDRE special procedure mode is discussed. In section

HDRE -special procedures mode
Our TSET unit in this study is based on the 4 and 6 Me V electron beams from an Elekta Precise medical linear accelerator, operating in the High Dose Rate Electron (HDRE) mode. The treatment head geometry is the same as in the standard 4 and 6 Me V electron modes.
As this technique requires the patient to be positioned at a considerable distance from the treatment machine (e.g. 3-7m), one of the main problems with this is that patient dose rate is considerably lower than at the isocentre so the treatment times are extended, making patient movement more likely. HDRE reduces this problem by allowing the electron dose rate to be increased.
The HDRE mode is controlled by an independent program, but shares the same scattering foil assembly with the conventional electron beams (Precise Operators Manual, 2003). To prevent unauthorized use of this mode, a HDRE key must be inserted at the control console and a HDRE applicator must be inserted into the head of the treatment unit. This applicator does not collimate or modify the electron beam in any way, although it may be used as a support plate for devices used to scatter the beam or degrade its energy (see The high dose rate interlocks prevent dose rates exceeding 1000 MU per minute during non-HDRE treatments. In addition to this feature, an interlock is provided which terminates radiation if the measured dose exceeds 125 percent of the nominal dose rate for a period of five seconds (Precise Operators Manual, 2003).
In order to deliver a high dose rate with the HDRE option, an electronic attenuator is introduced into the dosimetry circuit which divides the dosimetry signal by an approximate factor of 10, allowing HDRE dose rates of 3000 MU per minute at isocentre.
This means that during HDRE operation the dose and the dose rate reading at the console do not reflect the dose at the isocentre as for normal treatments, but that at a distance of approximately 3-4 meters from the isocentre.
In this Elekta Precise linear accelerator the HDRE beam is configured to a nominal energy of 4 MeV as option HDRE1, and 6 MeV as option HDRE2.

Experimental measurements 4.3.1 Depth dose measurements
All the depth dose measurements at the machine isocentre (100 cm SSD) were made with the gantry angle 0°(beam pointing down) and at the treatment plane ( readings were taken, and these reading were then averaged. The depth ionization curve was converted to a depth dose curve by applying stopping power ratios for water to air ; this curve was then normalized to its maximum value.

Beam profile measurements
The cross-beam profiles at the corresponding depth of the dose maximum at 100 cm SSD were measured in a polystyrene phantom for both 4 and 6 Me V energies. The ionization chamber was placed at a depth of dmax for each electron beam energy. The x-ray collimator field size was automatically set to 40x40 cm 2 projected at 100 cm SSD by using the high dose rate applicator. A series of readings were taken by positioning the ionization chamber at various off-axis distances from the central axis. Three readings were taken at each position and then averaged.

Multiple field measurements
The purpose of these measurements was: -To investigate the effective depth dose for multiple fields.
-To investigate the regional distribution of dose over a Rando phantom.
-To investigate the in vivo measurements using a Rando phantom. The TSET technique chosen was similar to the modified Standford technique (Karzmark et al, 1988), but with variation of the gantry angles used along with the irradiation fields.
Film was used to determine depth doses from multiple beams; the bare film was placed between the Rando phantom slices, cut flush to the phantom surface, and sealed with two layers of black electric insulation tape in the dark room. The Rando phantom slices containing the film were clamped tight to eliminate air gaps between the film and the phantom surfaces. The films were introduced into the Rando phantom between slices at four different levels namely, head, thorax, navel and pelvis level as shown in figure 4.5.
The Rando phantom was positioned on top of a rotating platform that was placed at the treatment plane at 350 cm SSD as shown in figure 4.6. Six beam pairs were employed (the major contribution at any given point is from three). After each irradiation of a beam pair the phantom was rotated through 60 degrees. Although film was not used for absolute measurements we were able initially to relate a dose from the TLD, which was calibrated by placing several dosimeters at the calibration position. The 12 fields were applied and 200 MU was delivered per field. Six fields were delivered with a gantry angle of 104°and 106°, and the other six fields were at 76°and 74°for 4 and 6 MeV respectively. After irradiation, the films were taken out in the dark room and were processed at the same time with the dose calibration films. The radiation pattern on the films was a series of concentric rings. The films were scanned using a Wellhëfer WP 102 Densitometer to determine the PDD and isodose curves.
For the cylindrical phantom the film was also introduced between two slabs of the phantom and sealed with black insulation tape and positioned at the treatment plane to obtain the beam data.
When six TSET beams are delivered in patient treatments, the dose delivered to any point on the patient surface will be larger than what would have resulted from a single beam due to the overlap of adjacent beams. The ratio of the dose delivered to the skin by the six-beam treatment to that delivered at dmax by a single TSET beam given the same beamon time was defined as the overlap factor (Cox et al, 1989 andKarzmark et al, 1988). An accurate determination of the overlap factor is important in TSET dosimetry as it determines the beam-on time of the TSET beams for a given prescribed skin dose. The overlap factor for this technique was measured at the level of the horizontal beam axis by using both radiographic film and TLD. Once the overlap factor was determined, the MU needed to give a prescribed skin dose in the TSET treatment can be calculated. Using the calculated MU, doses to other anatomie locations, from six-beam TSET treatment, were measured using TLD on the Rando phantom.
For TLD measurements, the TLDs were taped to the surface of the Rando phantom at the level of the beam axis for the six-beam irradiation. To minimize the statistical uncertainty of TLD, three TLD chips were placed at each anatomie location. In vivo dose measurements using TLD were performed according to an established procedure or protocol in our department, which is described briefly below. A group of 25 LiF TLD chips were used. The TLD chips were stored in a holder containing an aluminum tray.
Each TLD chip was numbered by writing on one side with a pencil. Before the dose measurements, the relative response of each TLD chip was quantified as follows: The TLD chips were annealed in the aluminum tray at 400°C for 1 h followed by 80°C

Absolute dosimetry
A plane -parallel ionization chamber (Roos chamber) with plastic phantom has been used in this study (according to the AAPM report No. 23 recommendations) (AAPM, 1988) for measuring the depth dose distribution. The calibration depths were measured from the depth dose curves for the 4 and 6 MeV beams. The x-ray contamination was measured from the tail of the depth dose curves according to the accepted method (ICRU Report 35, 1984).

Monte Carlo simulation of the Elekta Precise accelerator 4.4.1 Modeling of the radiation head of an Elekta Precise linear accelerator
The Elekta Precise linear accelerator was used in this study. The construction details of the treatment head were provided by the manufacturer.

Monte Carlo simulation of the Elekta Precise linear accelerator
The EGS4 Our EGS4/BEAM simulations consisted of two major stages (see figure 4.8); the first stage involved adjusting the primary electron beam parameters to match the beam data measured at 100cm SSD. In the second stage, these beam parameters were used to compute dose distributions at the treatment distance 350 cm SSD. Using the EGS4/BEAM code, two phase-space files were generated; one at the isocenter (100 cm SSD), and the other at the treatment plane (350 cm SSD). Each of the phase-space files contained spatial, spectral, and angular distributions of electrons, photons, and positrons.
In the first stage of the simulation, the accelerator head was simulated (from electron window up to the isocentre (100 cm SSD). In the second stage transport of phase space particles from isocentre to the treatment plane (350 cm SSD) was simulated. This is more effective than combining everything in one simulation, because the radiation head simulation only has to be done once.

a) The electron window
The "CONESTAK" CM was used for modeling the Electron window (Fig 4.9). The "CONESTAK" CM was used for modeling the primary collimator. The mam function of the primary collimator is to confine the electron beam to a useful size. This b) The primary scattering foil c) The primary collimator CM extended from Z =1.2 to 11.6 cm. The CONESTAK was used because the primary collimator is cone shaped. The medium inside the cone is air.

d) The secondary scattering foil
The "CONEST AK" CM was used for modeling the secondary scattering foil. The main function of the secondary scattering foil is to create a much more uniform electron beam than can be produced with a single foil and to reduce the need to eliminate lower-energy scattered electrons from the electron applicator (Khan, 2003). This CM extended from Z =11.6 to 15.0 cm.

e) The ionization chamber
The "CHAMBER" CM was used to model the ionization chamber (Fig 4.10 Because there are fewer boundaries for the particles to cross in the side walls of this CM than in other cylindrical planar or voxel geometries, the computing efficiency can be greatly improved by using the electron range rejection technique (Rogers et ai, 1995).
This CM extended from Z = 15 to 17.3 cm.

f) The mirror
The "MIRROR" CM was used to simulate the mirrors in the accelerator (Fig 4.11).The main function is to reflect light from a light source for field localization during patient positioning (Khan, 2003). MIRROR CM can have an arbitrary angle with respect to the Z-axis. The number of layers and their thicknesses and materials in the mirror can be arbitrary. The mirror is surrounded by air. The MIRROR outer boundary is a square centered on the beam axis. This CM extended from Z =17.3 to 26.85 cm.

g) The air gap between mirror and multi-leaf collimator
The "SLAB" CM was used to model the air gap between mirror and multi-leaf collimator (Fig 4.12). SLABS CM is also used for multiple slabs of arbitrary thickness and material, which are perpendicular to the z-axis. One single slab is a special case for SLABS.
SLABS has square symmetry about the beam axis. This is because SLABS is often used to model the bremsstrahlung target in photon accelerators. The outer boundary is a square. The air gap extended from Z = 26.85 to Z = 29.8 cm, that is, a thickness of 2.95 cm.

h) The multi-leaf collimator
The MLC was modeled for completeness, although the magnitude of its effect could not be estimated in advance.
The "MLCE" CM was used to model the multi-leaf collimators specific for Elekta machines (Fig 4.13). MLCE was designed as a variation of VARMLC (for Varian MLC) specifically for modeling Elekta MLC's. The tongue-and-groove in VARMLC has been replaced by interlocking steps. Also, unlike VARMLC, all leaves are identical and the sides of the leaves are focused (always to Z=O) by tilting each leaf about an axis that runs down the centre of its top surface. The entire leaf bank can also be tilted for off-axis focusing. This CM extended from Z =29.8 to 37.3 cm.   Rogers et al, 2005). .

i) The jaws
The "JAWS" CM was used for modeling the secondary adjustable jaws (Fig 4.14), whose main function is to collimate and define the field size at the isocenter (Khan, 2003). The accelerator jaws consist of upper (Y-direction) jaws and the lower (X-direction) jaws.
The positive and negative lateral coordinates at the jaws positions were adjusted such that a square field size of 40x40 cm 2 could be obtained (at the isocenter). This CM extended from Z =37.3 to 50.9 cm.

j) The air gap between the jaws and the isocentre
The "SLAB" CM was used to model the air gap between the jaws up to the isocenter plane cm. This air gap extended from Z = 59.9 to Z = 100 cm, that is a thickness of 40.1 cm. In other words the PPSF was created at Z = 100 cm at isocentre.
These CMs involved the first simulation stage of the accelerator, and at this stage of simulation about 1 x 10 8 histories were run. To avoid the discrepancies in the simulation, histories were divided into ten batches in order to be run individually and after that the PSFs generated from each batch were combined together to get the total number. Each of the PSFs was used as source file for the DOSXYZ program to obtain the beam data in the water phantom. Also the same PSFs were used as sources for simulation of the second stage to obtain the dose distribution for the CT based models using the CTCREATE program. 77

Second stage of simulation
In this stage the output of the accelerator was mode led from the previous scoring plane that was located at the isocenter, which is at Z = 100 cm, up to 350 cm SSD. This consists of an air slab with a thickness of 250 cm. Only one CM (SLAB) was used to model this air gap.
The primary PSFs collected in the first stage were used as the source input for this stage.
Bremsstrahlung splitting, Russian roulette or photon forcing were not used. The values for ECUT and PCUT as used in the first stage were maintained (0.7 MeVand 0.01 MeV respectively). The simulation in this stage resulted in a short simulation time compared with the first stage, because just the air gap component module was modeled.

Analysis of the phase space file
The obtained PSFs from the simulation of the Elekta Precise linear accelerator at a distance of 100 cm from the electron exit window, and at the treatment plane at 350 cm SSD were analyzed using the BEAMDP program (Ma and Rogers, 2005

Determining primary electron beam parameters
To find the parameters of the electron beam incident on the exit window, we followed published procedures by matching our calculated depth doses (PDDs) and cross-beam profiles to our measurements (Pavón et aI, 2003 andSung et aI, 2005). We started using beam parameters from the scientific literature where the simulations of linear accelerators were mode led, and made adjustments until the best match was found (fine-tuning). We assumed the electrons incident on the exit window to be monoenergtic and also investigated parallel circular beams with different energies having Gaussian radial distributions of full width at half-maximum (FWHM) of 0.05 cm. The beam parameters that yielded the closest agreement between simulations and measurements were considered as the true beam parameters and used for all subsequent calculations.

DOSXYZ -Calculation of dose distribution in 3D phantoms
The DOSXYZ code was used for the calculation of the absorbed dose in a water phantom. This allowed calculation of percentage depth doses and cross beam profiles for the large field size for the comparison with measurements. The PSFs generated by the BEAM program were used as source input files for the DOSXYZ code. The scoring plane for the PSFs was at the surface of the phantom.

The construction of the water phantom model
For MC calculation of the beam data two water phantom models were constructed: one at the isocentre and the other at the treatment plane. The isocentre (

79
For the water phantom at the treatment plane the boundaries in the X and Y direction are changed, but the voxel dimensions are the same as the one at the isocentre.

Transport control parameters
The electron and photon energy cut-off (ECUT and PCUT) was set at 0.700 and 0.01 MeV respectively, and the 700icru PEGS4 data file was used for the supply of cross section data for water and air media. The PRESTA algorithm was set "on" during simulations, because that proved to increase the history simulation rate quite effectively (Antolak et aI, 2002). The source was a full PSF from stage one incident on the front face. The number of histories was chosen so as to reduce the percentage error to less than two percent in all voxels in the field. This required about 600 million histories and the simulation rate was in the order of7.5x 10 7 histories per hour on a 2.41 GHz Pentium IV PC. 4.4

Number of histories and uncertainty
As we mentioned previously in section 3.7, the standard deviation in any Monte Carlo process is inversely proportional to the square root of the number of histories used (Bielajew and Rogers, 1988) therefore: Where (J is the standard deviation and N is the number of histories. Thus four times more histories are required to reduce the standard deviation by half.

4.5
-To find the number of histories that are needed to reduce the standard deviation to within 1%, given the standard deviation cri found with the trail run ofNl histories we let cr2 be equal to 1% and substitute in the equation 4.3: Thus the number of histories required to reduce the standard deviation to within 1% is equal to the initial number of histories multiplied by the square of the uncertainty.
From the above equations we can establish the appropriate number of histories to be used for the DOSXYZ program that will give a standard deviation which is within less than 2%. Several parameters were analyzed after simulation e.g. number of histories in simulation, histories simulation rate, CPU time, and computer memory space for each.
Full descriptions of simulation parameters are discussed in the results in chapter five.
The output file generated by DOSXYZ consists of a 3D dose array. The file is an ASCII file that can be read and displayed after writing suitable computer programs. After running the DOSXYZ program, graphs of depth dose and cross beam profiles were plotted. These profile data were compared with similar data obtained by measurements.

Comparison of Monte Carlo and measured profile data
The calculated beam data using DOSXYZ MC code and that obtained by the film and ionization chamber measurements were compared to each other. The PDDs were measured on the central axis for the two energies and by both methods. In order to obtain percentage errors less than 2%, the data were sampled in increments of 0.25 cm for the MC simulations up to a depth of 24 cm. The ion chamber measurements were done in increments of 0.15 cm, corresponding to the thickness of the available polystyrene sheets, up to a depth of 20 cm. Measurements were also performed using EDR2 film. The 83 measured depth doses data were then normalized to 100% at dmax for both. MC and ion chamber and the corresponding data were plotted on the same graph for comparison. The data for cross plane profiles were also sampled across the central axis with increments of 0.5 cm for MC and 1.0 cm for the chamber. In both cases the data were obtained at the maximum depths of each associated energy. All data were normalized to 100% relative to central axis dose, for both the MC method and chamber measurements. Corresponding data were then plotted on the same graph for comparison (see the results).

CTCREATE
The input beam data for patient models usually consist of a set of CT slices with the patient information expressed in terms of CT numbers. The conversion of these CT numbers into material properties is one of the main factors that determine the accuracy of the patient dose calculations (Du Plessis, 1999). Thus, before the data can be used directly for dose computation they have to be converted to electron densities relative to that of water.
For a patient model to be used in the DOSXYZ MC code, each voxel (volume element) has to be associated with a specific material entry in a PEGS4 cross-section file with a specific name and physical density (Walters et al, 2005). The PEGS4 data file contains the relevant cross section data for each material in the patient model. In the BEAM/DOSXYZ MC codes, a number of materials were used for the conversion of each voxel's CT number into electron density. The 700icru cross-section data for materials in the PEGS4 data was used in the simulation. After running of the CT phantom/CTCREA TE program, the CT phantom information is written into a file X.egs4phant, where X is the file name of the CT data set. DOSXYZ was run on these CT phantom files using any appropriate phase space file from the BEAM code as the source input file.

Comparison of measured and Monte Carlo dose distributions 4.7.1 Film registration
The method used for calculation of the film dose distribution was described in "Multiple field measurements" in section (4.3.3). The EDR2 films were introduced into four different slice levels along the length of the Rando phantom; Level I: head, level II: thorax, Level III: navel and Level IV: pelvis level as shown in figure (4.4). These films were scanned using the Wellhëfer WP 102 densitometer to determine the isodose distributions. The output-scanned data of the films were stored in a 2D array with a scan spacing 2 mm in the cross-plane direction for each level. The dose matrices consist of arrays of different dimensions for each slice. These dose array files were converted to ASCII files in order to be read and displayed on a PC.

The CT image
The original CT image data are in an integer array of 256x256 voxels obtained from the CADPLAN HP Unix work station. Thus, to be read and displayed on a PC the bytes had to be swapped. For the companson of the DOSXYZ and film results, the dose distributions were displayed (overlaid) on the CT image.

Normalization of the dose distributions
For comparison to be meaningful, the normalization in both the film data and DOSXYZ should be the same. The film dose was normalized at the calibration point, which is located near to the skin at the navel level of the Rando phantom (according to the AAPM TG-32 protocol recommendation). The average dose in 9 voxels was calculated at that calibration point and then the complete dose distribution was normalized relative to that average dose. A program was written in IDL to read, normalize and display the dose distributions. Thus the DOSXYZ dose distributions had to be normalized in a similar way to the film data.

Introduction
In the previous chapter the methods used for carrying out the experimental procedures and the MC simulation of the Elekta Precise accelerator and calculation of the dose distributions with MC methods were presented. In this chapter the results that were obtained for the experimental measurements and the MC simulation are given.

HDRE -special procedures mode
The operating parameters including linear accelerator dose and dose rate with the collimator setting at its maximum for the HDRE mode were investigated. The HDRE mode provides an adequately high dose rate with 4 and 6 Me V electrons at the patient treatment plane located at several meters distance e.g. 3 -3.5 m. This was investigated through several measurements of the output of the machine at different distances as shown in figures 5.l.a, 5.1. b and table 5.1.
We define the output factor of the machine in the high dose rate mode as follows:

( D95 )Norma/-dose-rate
Where (D SSD ) HDRE is the dose measured at 350 cm SSD in HDRE mode, for a single dual field.

(D95)
is the dose measured at 95 cm SSD, usinglOxlO crrr' applicator at Normal=dose-rate normal dose rate.  MeV respectively. The solid lines represent a power function fit to the data points. The dose rate of the machine in HORE mode at the isocentre is approximately ten times that in the normal treatment mode (Precise operators manual, 2003). Also the graph shows that the output of the machine decreases at a lower rate at extended distances and there is a relatively small difference in the output factor values between distances of 3 to 3.5 m. This is an advantage in implementing a TSET technique because of the reduced influence of variation of SSD at the patient skin surface.    The peak of the dose distribution moves to shallower depths as the SSD increases due to the divergence of the beam and energy loss in air. The ion chamber measured values of dmax agree quite well with the film data. For the ion chamber measurements the surface dose increases with increasing SSD for 6 MeV, but for 4 MeV the surface dose is almost constant. The precision of the measured data with the ion chamber is within about 1 mm.

Single field depth dose measurements
This corresponds to about 3% change in dose.
For the film measurements the characteristic curve for Kodak EDR2 film is shown in

Single field profile measurements
Beam profile at the isocentre The beam profiles for both energies at 100 SSD have a good uniformity compared to other published results (Peters et al, 1995, andSung et aI, 2005). This is probably due to the dual scattering foil design of the Elekta Precise machine. The beam profiles at 100 SSD of the 6 MeV beam have a better uniformity than those of 4 MeV due to less scattering in air of the higher energy (6 MeV) compared to lower energy (4 MeV). For the TSET single beam profile (350 SSD) the measured 90% beam widths for 4 and 6 MeV energies are 65.1 ± 1.1 and 69.8± 1.6 cm respectively. For the 60% beam width there is little difference between 4 and 6 Me V energies.  Off-axis distance (cm)

Ratio of average skin dose to calibration dose: The overlap factor (OV)
The overlap factor was calculated using calibrated TLDs around the outer surface of the cylindrical phantom and also using film. The dosimeters were irradiated with six dual fields at 60' intervals. The average dose was determined for all the TLD chips.
The overlap factor was also calculated for the Rando phantom using TLD. The following

Absolute dose measurement
The calibration depths were 6 mm and 10 mm for the 4 and 6 MeV respectively. These  AAPM, 1988 recommends an x-ray background of not more than 1%. The dual field technique reduces the total x-ray contamination at the centre because the main bremsstrahlung component is on the central axis of the beam. The x-ray contamination was measured from the tail of the percentage depth dose curves and it was found to be 0.9% and 1.3% for the 4 and 6 MeVelectron beam energies respectively.

Simulation geometry of the Elekta Precise linear accelerator
In this study the Elekta Precise linear accelerator model was built and compiled with the EGS4/BEAMnrc MC code. A graphical representation of the geometry of the Monte Carlo model for the Elekta Precise linear accelerator as produced by BEAMnrc is shown in figure 5.11 below.

Monte Carlo simulation of the Elekta Precise linear accelerator
The BEAMnrc simulation consisted of two major stages; in the first stage we obtained the primary phase space file (PPSF), where the scoring plane was at the isocentre (100 cm SSD). In this stage the beam simulation parameters were adjusted for a good match to the beam data measured at 100 cm SSD. In the second stage, the PPSF was used as source input for obtaining the secondary phase space file (SPSF) which was used subsequently as an input source to compute dose distributions at the treatment plane (350 cm SSD).
BEAMnrc calculates uncertainties of its results by dividing the total number of histories into 10 equal batches and determining the average value and standard deviation from the results of the ten batches. In order to obtain uncertainties below 2% it was found necessary to run 1 x 10 8 histories for the first stage of simulation. The PSFs generated from each batch were combined together to get the total number. The primary phase space files occupied 0.929 and 1.5 Gbyte of disk space for 4 and 6 MeV respectively. The secondary phase space files that were collected at the treatment plane occupied 2.0 and 2.2 Gbyte of disk space for 4 and 6 MeV respectively. Details of the simulation are shown in Table 5  1) The number of histories utilized in all the simulation stages were the same. (To obtain uncertainties below 2%).
2) The number of particles in the PPSFs (at the isocentre scoring plane) was the same for both energies. The number of particles in the SPSFs (At the treatment plane) was slightly larger for 6 MeV than for 4 MeV, due to more scattering and energy loss in air for 4 Me V electrons.
3) Simulation time for the PPSFs was longer than for the SPSFs at the isocentre because of the simulation of all head components of the accelerator comparing with a very short simulation time at the treatment plane because only one component module (the air slab) had to be simulated.
4) The simulation rates for the two energies for the same scoring plane are almost the same.

Analysis of the phase space files
By using the BEAMDP program (Ma and Rogers, 2005)  for 4 and 6 MeV respectively. These planar fluence graphs are slightly more uniform at the field center and start to decrease when we move further away from the central axis.

Spectral distribution
The spectral distribution gives the total number of particles scored in each energy bin within a specified energy range. The spectral distribution at SSD = 100 cm and at SSD = 350 cm for mono energetic incident electron beams are shown in figures 5.20 to 5.23. It can be noted that the spectra have two main peaks that are indicated for all particles (dark blue line) and electrons only (red lines). These two peaks are probably due to the geometry of the primary and secondary scattering foils respectively (Bjërk et al, 2002).

Validation of the Monte Carlo model
The parameters characterizing the electron beam incident on the exit window were found by matching our calculated beam data (percentage depth doses and cross-beam profiles) to our measurements in the water phantom at the isocentre (100 cm SSD). For the measured beam data of 4 and 6 MeV energies we found the best match (fine-tuning) for the calculated beam data by using 5 and 6.72 MeV energies respectively. Also we used parallel circular beams having Gaussian radial distributions with full width at halfmaximum of 0.05 cm. The electron source was considered mono energetic in both cases.
The MC beam models were then evaluated at the treatment plane by comparing the calculated PDDs and profiles with measurements as explained in chapter 4. Dose distributions in the Rando phantom were also compared.

Calculation of the beam data in a water phantom with the DOSXYZ code
The phase space files created in the first stage of the simulation for the 4 and 6 Me V at the two scoring planes were used to calculate percentage depth dose and cross-plane profiles in a water phantom using the DOSXYZ program.
The number of histories to be used for each scoring plane for each energy was determined by carrying out trial simulations as mentioned before in chapter 4 (section 4.4.5.2).
Simulations were done with two different phantoms, using 200 and 600 million histories for the phantoms at the isocentre and at the treatment plane respectively. This resulted in an uncertainty below 2% for both scoring planes. The results in table 5.4 show that the simulation time for 4 MeV is much shorter than for 6 MeV. That is mainly due to the difference in the specifications of the pes that were used in the simulation. MeV for a single horizontal beam. In both figures notice that the depth dose curves start from 0.25 cm because of the first pixel size 0.25 cm that was constructed in the water phantom. Also the depth dose curve moves towards the surface as we move to the extended distance. The uncertainty of the simulation is less than 2%.    Once the single beam parameters have been determined to be adequate, the next task is to find the optimum beam angles, which is very much like matching beams in conventional treatment planning. The objective is to find the appropriate projection angle that gives a flat (±10%) profile over approximately 2 m (AAPM, 1988). For TSET beams, we are helped by the fact that the beams are very broad and there is no distinct beam edge. By using a range of angles between the central axes of two beams, the angle that gave the best uniformity of dose was determined. The calculated combined beam profiles for the optimum angles, are shown in Fig. 5.34 and 5.35. The optimum angle was found to be 14°for 4 MeVand 16°for the 6 MeV. The dose uniformity was ± 5% for 4 MeVand ± 3% for 6 MeVover the range of -100 to +100 cm.   and 6 MeV respectively. The statistical uncertainties in the MC data are less than ± 2%.

Dual beam characteristics
The calculated values of dmax, Rso, and Rp, agree with measured data within about 1 mm. Tables 5.5 and 5.6 show measured and calculated characteristic parameters of the percentage depth dose curves for the two energies for 40x40 cm 2 at the isocentre (100 cm SSD) and the TSET beam at the treatment plane (350 cm SSD). These tables show a numerical comparison for the two electron energies for the PDDs that were shown in the above figures. As expected, there is excellent agreement in the Rso and R, depths between the ion chamber data and MC data. In the comparison, the average depth indicated in the tables represents the average for measured data using ion chamber and film. The difference is noted between this average depth for the two measurements and MC.  The computed beam profiles at the isocentre (lOO cm SSD) agrees with measurement within statistical uncertainties of 2%. We also verified the above-determined beam parameters at the extended distance. The statistical uncertainties were less than ±2 % near the beam. For the beam profiles at the treatment plane the agreement between MC and measurements was nearly as good as at 100 cm SSD.

Carlo and the Measurements
Beam Profile at 100 cm SSD, 6 MeV

Head level
The PDD curves at the head level are shown in figures 5.44 and 5.45 for 4 and 6 MeV energies respectively. The measured and calculated PDD curves for both energies show some disagreement. For the 4 MeV, the measured surface dose is less than that calculated by MC. This deviation is due to the 4mm aperture diameter of the film densitometer detector. The region above and below the 50 % depth shows some difference of the dose between MC and film dose. These discrepancies represent a difference of about 2 mm in depth.

Head level
Six dual-fields dose distributions for the different energies (4 and 6 MeV) were calculated using MC and measured with film methods. In all cases, the six pairs of dual fields were arranged with angles of incidence 60 degrees apart. For comparison of MC and film the dose to entire skin was normalized at the calibration point which is near to the skin surface at the navel level, at a depth of2 and 3 mm for 4 and 6 MeV respectively. Also for both methods the dose distributions are higher in the areas that have sharp protrusions due to the larger degree of overlap and penetration of beams in those areas (e.g. nose and ear areas). There is a decrease in dose in the areas shielded by protrusions.
In figure 5.53, the film dose distribution shows a discontinuity in isodose lines near the back of the head which is due to an artefact created during the film processing.
In general the film and MC dose distributions agree well. The dose distribution for the higher energy (6 MeV) clearly shows a greater depth of penetration in both the above figures. The dose distribution shows some hot spots in the direction of 60 degree beam incidence which results from the overlapping of the fields. Also we observe that the dose distribution penetrate small areas of the posterior part of the left lung as shown for both methods in particular in case of 6 MeV, that is due to larger penetration of this energy.

Thorax level
The interesting feature here is that the film method tends to follow MC method in the isodose lines in that area of the lung and this is quite apparent at the higher energy as shown in figure 5.55.

Statistical uncertainty analysis
The Monte Carlo calculation method is subject to statistical uncertainties due to the stochastic nature of the radiation transport process. The number of histories that is needed to reduce the uncertainties of a given number of histories by half is equal to 4 times the initial number of histories (see section 4.3.1). The statistical uncertainties in these MC dose distributions were about 2 percent. This was achieved by running a large enough number of histories. The statistical errors also depend on the actual voxel size of the CT phantom, which was determined by the scan setting of the CT scanner. These voxel sizes were 1.5x1.5x10 mnr' for the whole phantom.
For the film dose distribution the statistical errors were 3, 2, 3, and 6 percent for the head, thorax, navel and pelvis respectively. The overall statistical error for the film data was calculated as follows: the average value to the dose at each point of three films was determined as well as the standard deviation at each point. A global standard deviation was then calculated as the average of the point standard deviations. This global standard deviation was indicated as the error bars in the percentage depth dose curves for film in the previous figures (section 5.8.9.1).

CONCLUSION AND RECOMMENDATIONS
The main purpose of this study was to commission and optimize a total skin electron therapy (TSET) technique for the treatment of mycosis fungiodes with the Elekta Precise linear accelerator using the high dose rate electron (HDRE) mode that is installed on the machine. This was done through an extensive set of measurements and a large number of Monte Carlo simulations. It is generally accepted that MC simulation is the most accurate way to obtain detailed information about any radiation beam and its accompanying dose distributions. In this research the Monte Carlo beam models were validated by matching the simulated beam data at different scoring planes with corresponding measurements.
From the previous chapters several conclusions can be made: 1) The HDRE mode is a useful facility providing reasonable dose output and field size, and acceptably low levels of x-ray contamination. The dose uniformity and dose rate at the treatment plane (at 350 cm SSD) meets the recommendation (AAPM No. 23) concerning beam flatness and x-ray contamination (less than 2%), and it considerably reduces the treatment time of TSET while retaining proper functioning of all accelerator dosimetry systems. Moreover the machine can be easily set up for treatment without additional technical support. Nevertheless the TSET with HDRE mode requires careful calibration and a well controlled procedure to carry out the treatment.
2) Use of a dual field technique produces acceptable beam uniformity over an area large enough to allow total skin electron therapy in the existing treatment room. This uniformity was achieved through optimization of the angles of incidence of the two beams forming the composite dual field in the vertical direction.
3) The absolute calibration of absorbed dose to the patient requires the measurement of the ratio "skin dose to calibration point dose". The value of this ratio (also called overlap factor) was found to vary between 2.4 and 2.9 for the different dosimeters used in this study. To assess the effective treatment depth and the degree of bremsstrahlung contamination, it is necessary to measure the percentage depth dose curves for the complete treatment using all twelve fields on a cylindrical phantom.
From the results it can be concluded that the depth of maximum dose (d max ) is approximately 2 and 3 mm and the therapeutic range (Rso) is approximately 7 and 9 mm for the 4 and 6 MeV beam energies respectively. Therefore the depth doses (in cylindrical phantoms) from multiple beams can indicate to the clinician the effective treatment depth. 4) Monte Carlo simulation of a linear accelerator can provide information such as fluence, energy fluence, energy spectra and angular distributions of the radiation beam which is almost impossible to measure. The BEAMDP code allows this information to be acquired from the phase space files produced by BEAM at any scoring plane. This detailed information about the radiotherapy beam increased our understanding of the clinical beam characteristics especially for the TSET technique.
Some investigators have used this information to improve accelerator design and improve accuracy of dosimetry (Pavón et aI, 2003 andSung et aI, 2005).

5)
The results have proven that the MC method can accurately reproduce the measurements in non-standard conditions such as the TSET technique by acquiring the large beam profiles in a 400x200x24 crrr' phantom at the patient treatment plane.
This confirms the suitability of the DOSXYZ Monte Carlo code for the calculation of the beam data in a 3D water phantom at extended SSD. The agreement between the MC calculated data and corresponding data obtained by measurements are remarkably good. Results agree within less than 2% except for the areas near the field edges at the treatment plane.
6) The Monte Carlo method can also be used to calculate dose distributions throughout a 3D volume constructed from CT slices. Dose distributions were generated with the DOSXYZ code at different levels in a CT based model of the Rando phantom from the twelve fields incident at the patient treatment plane. The results agree well with the film measurements done at the corresponding levels in the phantom, with relatively small discrepancies at a few positions. This confirms the validity of the assumptions that dose distributions can be determined with acceptable accuracy anywhere in the 3D volume of a suitable patient model.

7)
A large number of histories are required to achieve adequate statistical uncertainty in the dose distributions. Because of this the Monte Carlo simulation time was relatively long using the available computing facilities. For example a complete treatment using six dual fields took about 72 hours on a 2.41 GHz Pentium IV Pc.
8) The dose distributions in phantom were found to comply with the guidelines described in the AAPM TG-23 protocol, showing the suitability of this technique for treatment of mycosis fungiodes. This also shows the advantage of using MC as a treatment commissioning and optimization tool for TSET.

RECOMMENDATIONS:
1) The final delivered treatment dose will include the effect of the overlap factor. This relates the true average skin dose (due to beam overlap) to the calibration dose. It must be understood that this factor plays a very important role in the final patient treatment dose. Therefore the actual patient dose should be checked during treatment using TLD.
2) A detailed treatment protocol should be established for the clinical implementation of the TSET technique, dealing with: i) The equipment used in this technique, ii) The calibration and preparation of equipment immediately prior to TSET treatment and iii) The measurements of the actual dose delivered to the patient during the TSET procedure.
3) Although the MC calculations currently take a long time, the patient population requiring TSET is relatively small and one can use MC techniques to assess the dose distribution in this technique with current PC computer facilities.
4 The EGS4/BEAM code package running on a Windows based platform was used for the MC simulation. Percentage depth-dose curves and beam profiles were calculated and measured experimentally for the 40x40 cm 2 nominal field at both 100 cm SSD and at the patient surface at the treatment plane (SSD 350 cm) for a single beam. The accuracy of the simulated beam was validated by the good correspondence (within less than 2%) between measured beam characteristic parameters (Rso, dmax, Rp) and Monte Carlo calculated results. To obtain a uniform profile vertically, two vertical angles of incidence were used. The angle between the two beams that gave best uniformity was considered the optimum angle. The patient is to be placed on a rotating platform perpendicular to the beam and rotated through 60 degree increments to obtain six horizontal directions of beam incidence. The doses expected in the patient were measured with Kodak EDR2 films positioned at different levels between slices of a Rando phantom. TLDs were placed on the surface to relate the film measurements to dose. The delivered doses in the treatment plane were compared to simulated data that was obtained from the MC simulation.
The penetration depth of the dose distribution varied over various scanning directions between 2-3 mm and 3-4 mm for 4 and 6 MeV respectively. This information is useful when treatment of lesions of different thickness are being considered. The composite percentage depth dose of all six dual fields for both 4 and 6 Me V yielded an 80 % dose at -7 mm and -9 mm depth, respectively. Good dose uniformity was achieved for both energies and it was about ± 5% for 4 MeVand about ± 3% for 6 MeVover a range of -100 to +100 cm. The bremsstrahlung contamination was 0.9 and 1.3 %.
Generally there was good agreement between the dose distribution calculated with MC and measured with films, thus validating our MC calculations. The dose distributions in phantom were found to comply with the guidelines described in the AAPM TG-23 protocol, showing the suitability of this technique for treatments of the skin diseases.
The HDRE is a useful operational mode providing reasonable output, field size, and Xray contamination. Use of a dual field technique produces reasonable beam uniformity over an area large enough to allow total skin electron therapy in a conventional treatment room. Monte Carlo techniques provided a guiding principle to assist the verification of the beam characterization of a TSET technique. The absolute calibration of dose to the patient required the measurement of the ratio "skin dose to calibration point dose"; this was achieved by measurements with a parallel plate ionization chamber and TLDs.