Abstract
The Jebsen–Taylor Hand Function Test (JTHFT) is a standardized and objective assessment tool for evaluating hand dysfunction in various conditions (including stroke and spinal cord injury). The JTHFT has been widely used successfully in clinical settings; however, the instrument score does not reflect mechanical movement information of the upper extremities. Therefore, we developed a software to quantitatively evaluate the JTHFT. The subject’s motion was recorded using a webcam and a deep learning-based three-dimensional motion-extraction technique. Several elements were calculated from vectors between two points. Consequently, eight factors were analyzed: distance to move beans, velocity of the hand moving beans, time to move beans, time to put beans in a spoon, time to put beans in a cup, angle of the dorsum of the hand, trajectory of the hand, and total task time. The average task time was 6.82 s (standard deviation, 0.09), which within the normal range for adults. The cross-product angle of the dorsum of the hand was confirmed to be 0.01 units, depending on how tilted the hand was. In summary, we identified new quantitative assessment tools for occupational therapy; the extracted index results allowed for quantitative evaluation of the aspect that was previously impossible to judge.