Parkinsons freezing of gait prediction improved using IMU supervised video analysis
Freezing of Gait Prediction Under Spatial Occlusion: An IMU-Supervised Cross-Modal Distillation Approach
Computer Vision and Pattern Recognition
Summary
People with Parkinson's disease can have problems with walking, especially freezing of gait (FOG), where they temporarily cannot move their feet. Detecting these freezes helps assess their condition, but video methods struggle when legs are hidden during turns, and wearable sensors are hard to use outside clinics. The authors combine data from wearable sensors and video to improve detection, teaching video models using sensor data to handle cases when the camera view is blocked. Their method works well on data of patients doing full turns, predicting freezing accurately without needing sensors worn all the time.
What this means in practice
- •For physical therapists: Provide accurate gait impairment monitoring in Parkinson's patients using video without requiring wearable sensors during clinical assessments.
- •For medical device developers: Create more practical FOG detection tools that combine video and sensor data to offer non-invasive patient monitoring solutions.$Commercial implications: Enables development of sensor-free gait monitoring products for patients and clinics improving usability and comfort.
Authors
Chandan Biswas, Aryan Singh, Anabik Pal
Abstract
Parkinson's disease is a progressive neurodegenerative disorder characterised by gradual deterioration of movement control. Automated freezing-of-gait (FOG) detection supports the objective assessment of gait-related motor impairment. Two common approaches are used for FOG prediction: (i) analysing video recordings of the patient's movements and (ii) analysing data collected using inertial measurement unit (IMU) wearable sensors attached to the patient's lower limbs. Video-based approaches may suffer detection errors during continuous turning-in-place tasks because the lower limbs undergo substantial geometric self-occlusion, degrading pose-estimation accuracy. IMU-based approaches are generally less affected by visual occlusion; however, they are difficult to deploy outside clinical or laboratory settings, as the sensors must be attached securely and remain in place throughout the assessment. Motivated by this, we propose a cross-modal subspace distillation framework to mitigate the limitations of unimodal FOG detection by combining IMU accuracy with video-based practicality. We extract invariant latent topologies from a pre-trained kinematic oracle to structurally supervise a non-encoded visual architecture during training. To resolve periods of severe spatial occlusion, a dual-stream visual model probabilistically fuses skeletal graph nodes and continuous spatial pixels, dynamically shifting reliance to uninterrupted pixel boundaries as joint tracking confidence drops. Evaluated against a public, multi-modal sequence dataset of Parkinson's individuals executing continuous $360^\circ$ turns, empirical results demonstrate that applying sensory boundary topologies strictly mitigates tracking evaluation entropy. Our constrained optimisation confirms that highly precise FOG prediction bounds can be achieved over zero-wearable inference environments.