Conventional marker based optical motion capture (mocap) methods for estimating the position and orientation (pose) of anatomical segments use assumptions that anatomical segments are rigid bodies and the position of tracking markers is invariant relative to bones. Soft tissue artefact (STA) is the error in pose estimation due to markers secured to soft tissue that moves relative to bones. STA is a major source of pose estimation error and is most prevalent when markers are placed over joints. Mocap and bi-plane videoradiography data were recorded synchronously while three individuals walked on a treadmill. For all three, pose of the thigh and shank, and movement of markers relative to the bones, were determined from the videoradiography data (DSX, C-Motion). Independently, pose of thighs and shanks was estimated using mocap data (Visual3D, C-Motion). Our measures of error in the mocap pose estimation were the relative thigh and shank translations. X-ray data from two subjects were used to generate a regression model for the antero/posterior movement of the lateral knee marker against internal/external hip rotation. The mocap translation errors of the third subject, attributed to STA of the knee marker, were 15.6mm and 32.0mm respectively. The pose of the third subject was then estimated using a probabilistic algorithm incorporating our regression model. Mocap translation errors were reduced to 10.6mm (thigh) and 4.4mm (shank). The results from these data suggest that errors in pose estimation due to STA may possibly be reduced via the application of algorithms based on probabilistic inference to mocap data

Access to health care, including physiotherapy, is increasingly occurring through virtual formats. At-home adherence to physical therapy programs is often poor and few tools exist to objectively measure low back physiotherapy exercise participation without the direct supervision of a medical professional. The aim of this study was to develop and evaluate the potential for performing automatic, unsupervised video-based monitoring of at-home low back physiotherapy exercises using a single mobile phone camera. 24 healthy adult subjects performed seven exercises based on the McKenzie low back physiotherapy program while being filmed with two smartphone cameras. Joint locations were automatically extracted using an open-source pose estimation framework. Engineered features were extracted from the joint location time series and used to train a support vector machine classifier (SVC). A convolutional neural network (CNN) was trained directly on the joint location time series data to classify exercises based on a recording from a single camera. The models were evaluated using a 5-fold cross validation approach, stratified by subject, with the class-balanced accuracy used as the performance metric. Optimal performance was achieved when using a total of 12 pose estimation landmarks from the upper and lower body, with the SVC model achieving a classification accuracy of 96±4% and the CNN model an accuracy of 97±2%. This study demonstrates the feasibility of using a smartphone camera and a supervised machine learning model to effectively assess at-home low back physiotherapy adherence. This approach could provide a low-cost, scalable method for tracking adherence to physical therapy exercise programs in a variety of settings