A CNN-based Lower Limb Motion Quality Evaluation Method for Home-based Stroke Rehabilitation

Abstract

Background: With the rapid increase of stroke incidence in recent yearsworldwide, home-based rehabilitation training has become more needed,especially for remote regions or in developing countries where rehabilitationresources are scarce. Studies have demonstrated that home-based rehabilitationfor poststroke patients is essential for reducing the cost as well as for providingefficient rehabilitation. Nevertheless, home-based rehabilitation training requireseffective professional support and timely evaluation. Method: In this paper, a home-based rehabilitation quality evaluation methodfor lower limb training was proposed. The kinematic data of a patient’s lowerlimb during a set of selected training exercises was captured by a wireless bodyarea sensor network (WBASN). The data was then processed by a convolutionalneural network (CNN) based algorithm to classify the rehabilitation training typeand to evaluate the training quality. A series of kinematic features were selectedfor rehabilitation quality scoring. The experiments have been conducted using 26human participants, including 6 healthy participants and 20 stroke patients atdifferent Brunnstrom recovery stages. Results: An accuracy of 95.3% has been achieved for recognizing therehabilitation training types and a statistically significant linear positivecorrelation (R2 = 0.9962) has been obtained between the objective scores andthe Brunnstrom stages evaluated by the clinicians.