Kalman filtering of accelerometer and electromyography (EMG) data in pathological tremor sensing system

Currently there is a lack of objective clinical diagnosis and classification of tremor is difficult when it is subtle. Thus in previous work, a sensing system has been developed to quantify pathological tremor in human upper limb. In this paper, a Kalman filter algorithm to fuse information from accelerometers and surface electromyography is proposed. As the ground truth, an optical motion tracking system will be utilized. Then two sensor fusion algorithms based on Kalman filter are formulated to estimate the joint angle of the limb from the reading of accelerometers and surface EMG. Initial results using tremor data from two Parkinson's disease patients show promising future in this sensor fusion. The sensing system and the algorithms proposed are useful for actively compensating the tremor and helping the clinicians in tremor diagnostics.