Optical-Tracking-based Motion Capture Pillow Sensor

Accuracy in radiotherapy treatments for brain, and head and neck cancers is critical for safe ablation of tumours. The Motion Capture Pillow (MCP) was designed to enhance real-time head tracking accuracy during radiotherapy, whilst ensuring minimal occlusions and ferromagnetic interferences, alongside improving patient comfort. The contributions include enhancing the MCP’s compatibility with radiotherapy equipment, alongside optimising the marker density to improve its tracking accuracy. It strengthens MCP’s abilities using Optical Flow-based tracking algorithms and sensor fusion to improve robustness. Validation for performance and feasibility was obtained through participant and clinical-oriented participatory design studies.

Human-scaled mannequins were used for data collection. A mannequin-robot arm platform using Robot Operating System provided the ground truth via the robot’s end-effector. The imaging and tracking tools were investigated using this platform to compare cameras, image processing, and tracking algorithms. The fibrescope with greyscale imaging using the Kanade-Lukas Tomasi tracking algorithm showed strong correlations, establishing the methodology. The MCP’s marker density was optimised for accuracy using three marker spacings—5, 10, and 15 mm. Here, 10 mm was the default parameter from previous work. The spacing with 5 mm showed the lowest Mean Absolute Error of 1.34° for roll motion and 4.99° for pitch motion, post artefact-filtration.

The MCP with a gyroscope using a Kalman filtering-based fusion model reduced the RMSE relative to the standalone MCP by 70% and 61% for the pitch and roll motions from the mannequin-based experiments, and 42% and 70% for the pitch and roll motions from the participant-based experiments. The ground truth was obtained via a passive marker-based optoelectronic tracking system. Dissimilarities relating to the dynamics of the varying hardware set-ups for experimental environments were observed. This affected air pressure variations in the MCP’s cavity, requiring further investigation to minimise the measurement errors arising from experimental discrepancies and pressure variations.