TITLE:
Examining Gait Patterns after Total Knee Arthroplasty Using Parameterization and Principal Component Analysis
AUTHORS:
Kevin E. Roy, Victoria L. Chester, Chris A. McGibbon
KEYWORDS:
Total Knee Arthroplasty (TKA); Principal Component Analysis (PCA); Parameterization; Gait Analysis
JOURNAL NAME:
Open Journal of Orthopedics,
Vol.3 No.2,
June
17,
2013
ABSTRACT:
The use of parameterization in
assessing gait waveforms has been widely accepted, although it is recognized
that this approach excludes the majority of information contained in the
waveform. Waveform analysis techniques,
such as principal component analysis (PCA), have gained popularity in
recent years as a more
effective approach to extracting important information from human movement
waveforms, but are more challenging to interpret. Few studies have compared
these two different approaches to determine which yields the most relevant
information. This study compared the kinematic patterns during gait of six total knee arthroplasty (TKA)
subjects (10 TKA knees), to a group of 10 age-matched asymptomatic control subjects (19 control knees). An eight-camera Vicon M-cam system
was used to track movement and compute joint angles. Group differences in parameterization (max and min peaks) values
and principal component scores were tested using one-way ANOVA and
Kruskal-Wallis tests. Using parameterization, the TKA group was characterized by reduced hip extension,
increased hip flexion, increased anterior pelvic tilt, increased trunk tilt,
and reduced sagittal ankle angles compared to the control group. Waveform
analysis, by means of PCA, showed-magnitude
shifts in sagittal ankle waveforms between
groups, rather than solely reporting differences in peaks. Waveform analysis also indicated a significant shift in the magnitude
of the entire waveform for hip angles, pelvic tilt, and trunk tilt, indicating
no change in range of motion between groups, but rather a change in the way in
which range of motion is
achieved at the hip. This study has identified several gait variables that were
significantly different between the TKA and control groups. Our results suggest
that waveform analysis is effective at identifying magnitude shifts as sources
of variability between groups, which would not necessarily be analyzed using
conventional parameterization techniques unless one knew a priori where the
variability would exist.