Introduction: Three-dimensional (3D) gait analysis using optoelectronic stereophotogrammetry and force platforms is widely employed to assess joint loading and movement patterns [1]. However, marker-based motion capture systems are susceptible to measurement errors, particularly soft tissue artefacts (STA) arising from skin movement relative to the underlying bone [2]. These errors are even more pronounced in osteoarthritis (OA) research, where nearly 90% of participants have a body mass index (BMI) of 30 or higher, classifying them as obese [3, 4]. Accurate tracking of anterior pelvic anatomical landmarks is therefore crucial, especially in individuals with higher BMI. This study evaluated the impact of two methods for anterior pelvic landmark tracking on gait kinematics across different BMI
groups.
Methods: Twenty participants were evenly divided into two groups based on their BMI, classified as ”low BMI” (19–27.5 kg/m2) or ”high BMI” (>27.5 kg/m2). Each participant completed a 3D gait analysis on an instrumented treadmill using a modified CGM2.4 marker set, which included additional sacrum markers and pointer-based ASIS tracking. Two methods for anterior pelvic landmark tracking were compared: skin-mounted ASIS markers (ASIS𝑠) and pointer-based virtual ASIS tracking (ASIS𝑣). Their effects on the estimation of ASIS location, pelvic orientation, and hip joint centre (HJC) were
evaluated. Furthermore, intra-rater reliability of the pointer technique was assessed using intraclass correlation coefficients (ICC) and standard error of measurement (SEM), and the relationship between
BMI and pointing errors was analysed. The effect of using multiple pointer measurements on reliability was also examined.
Results: Data from twenty participants showed no statistically significant differences between ASIS𝑠 and ASIS𝑣 within the low BMI group. In contrast, the high BMI group exhibited statistically significant
and clinically relevant differences across all directions in ASIS trajectories, pelvic obliquity, and HJC estimation. A positive correlation was found between BMI and the mean absolute error in ASIS
dentification, with the largest deviations occurring in the anterior-posterior direction. Intra-rater reliability was generally high (ICC > 0.90); however, the medial-lateral direction within the high BMI group showed notably lower reliability, with ICC improving from 0.29 to 0.77 and SEM decreasing from 11.92 mm to 6.79 mm after averaging four pointer measurements.
Conclusion: Within the low BMI group, both ASIS𝑠 and ASIS𝑣 could be used for gait kinematics assessment. However, within the high BMI group, the two methods could not be used interchangeably due to statistically significant differences in ASIS tracking, pelvic orientation, and HJC estimation. ASIS𝑣 appeared to better estimate pelvic kinematics, as it was less affected by STA associated with skinmounted markers. Nevertheless, the accuracy and reliability of ASIS𝑣 require further validation using imaging techniques such as fluoroscopy, enabling direct measurement and potential correction of STA.

BACKGROUND: Research on the role of patellar biomechanics in patellofemoral (PF) joint pathologies requires accurate and reliable measurement of the patella relative to the femur (patellar tracking) but current methods have limitations. In vivo measurement of PF kinematics is done using medical imaging techniques but these are expensive, involve radiation exposure or only allow slow, supine motions. A widely used alternative is optical motion capture combined with musculoskeletal modeling. However, due to the large motion of the skin relative to the patella, the typical approach using anatomical calibrated skin-markers is unsuitable. As the shape of the patella is grossly visible underneath the skin, placement of a grid of skin-mounted markers on the skin covering the patella might offer an alternative for patellar tracking.
OBJECTIVE: The objective of this study was to develop a patellar tracking framework designed to measure in vivo PF translations using a grid of skin-mounted markers and to compare it to in vivo measures derived from dynamic CT images and musculoskeletal model estimates.
MATERIALS AND METHODS: Two participants performed a supine knee flexion-extension motion with a grid of 42 retroreflective markers placed on the right knee during four-dimensional computed tomography (4D-CT) measurement and optical motion capture. A patellar identification algorithm (PIA) was developed to estimate the PF translations in the medial – lateral (ML), superior- inferior (SI) and anterior – posterior (AP) directions. PF translations were determined by four different methods: CT bone geometry-based (CT_BONES), CT marker grid-based (CT_GRID), optical motion capture marker grid-based (OMC_GRID) and by musculoskeletal modeling (OMC_MODEL). To evaluate the feasibility of our method we first compared CT_BONES to CT_GRID. Next, results obtained by different measurement types (CT_GRID and OMC_GRID) and results obtained by two different marker-based methods (OMC_GRID and OMC_MODEL) were compared. To compare ranges of motions, patellar excursions relative to the femur were evaluated and compared for all knee angles (KA_ALL) and for knee angles > 20° (KA_ENGAGED).
RESULTS: Differences in patellar excursions given by CT_BONES and CT_GRID during KA_ENGAGED were less than 4 mm and RMSEs less than 3 mm. These differences were less than 5 mm between CT_GRID and OMC_GRID (RMSE: 5 mm) and less than 16 mm between OMC_GRID and OMC_MODEL (RMSE: 10 mm) for knee angles lower than 20°. Larger differences in patellar excursions and RMSEs were seen during KA_ALL.
CONCLUSION: PF translations could be measured using a marker grid with an accuracy within 3 mm for knee angles larger than 20° (i.e. during patellar engagement). However, our method performed worse for knee angles smaller than 20° flexion (an accuracy within 12 mm). The findings of this study provide valuable baseline knowledge for non-invasive, marker-based patellar tracking. Future studies should include larger sample sizes to allow for proper validation.