Automatic evaluation of the Nine Hole Peg Test for subjects with neurological pathologies through artificial vision approaches

According to the World Health Organization, Cardiovascular Diseases are the first cause of death globally and, among these, stroke is the second one. Stroke is a very serious acute pathological condition, responsible for the appearance of numerous neurological deficits caused by insufficient blood supply to an extensive area of the brain. This event causes cognitive and motor deficits that influence the individual's capacity for autonomy and that hinder self-sufficiency, making the person unable to perform daily life tasks in total independence. The degree of disability that derives from it, beyond gravity, makes the rehabilitation treatment necessary. Considering the social and economic impact of stroke, monitoring the progress of outpatients in home-based rehabilitation is paramount, because the risk of re-hospitalization can be reduced. A frequent traditional home-based assessment of the disease is not viable because the costs for healthcare system, but other low-cost solutions based on computer vision approaches can be pursued. In this context, one of the most common clinical tests used for the assessment of the upper limb functions and the fine manual dexterity of post-stroke patients is the Nine Hole Peg Test (NHPT). In this Master Thesis, the post-stroke is considered as a case study, and the focus is on the automated evaluation of the NHPT by a computer vision approach. The movements of the patient’s hand are analysed during the performance of the test, both to evaluate the standard time scores of the test, and to provide further information to clinicians about the individual’s performance. The system for the automated assessment is based on a hardware made of an RGB-D camera and gloves with coloured markers, and a software which implements computer vision algorithms for the capture and the analysis of the upper limp movement. The software evaluates the completion time for the test and characterizes the movements by a set of kinematic parameters. The software outcomes support both the NHPT standard clinical assessment and the evaluation of the efficacy of the rehabilitation treatments in a home environment. This work on NHPT is a part of a more extensive study aimed to the overall assessment of motor impairment in the upper limb, which is applicable not only for post-stroke but also in the context of other neurological diseases. In this thesis work, the performance of the system are experimentally evaluated .Healthy subjects and subjects with mild post stroke side effects are analysed while performing the NHP test .Experimental results show that the system is able to automatically evaluate the standard NHPT scores of both classes of subjects. Furthermore, the kinematic parameters considered are able to capture the differences in the motor performance of the two classes of subjects. These preliminary results support the feasibility for an automated assessment of post-stroke patients at home.