Towards arm tremor diagnosis using Kinect

Many clinical studies have shown that the arm movement of patients with neurological injury is often slow. In this paper, the speed analysis of arm movement is presented, with the aim of evaluating the arm movement automatically using the Kinect camera. The arm movement seems simple at the first glance, but in reality it is very complex neural and biomechanical process that can be used for detecting a neurological disorder. This is a preliminary study investigated three different armmovement speeds fast, medium and slow. Abnormality is associated with slowness in arm movement, while the fastness indicates healthy movement. With a sample size of 27 subjects, the proposed algorithm can classify the three different speed classes (slow, normal, and fast) with overall error 4.94 %. Moreover, the first 40% of the motion can be used to predict the speed type. Additional testing, on real patients with different hand and arm movement impairments is required to verify the effectiveness of the proposed approach.