Development of methods and software for diagnostics of human motor functions based on video fixation data

Abstract

Diagnosis of diseases of the musculoskeletal system is an urgent problem of medicine. The purpose of this study is to create a hardware-software complex to determine the symmetry of the body, mobility in joints and limbs. Such data contributes not only to the diagnosis of the disease, but also helps to monitor the results of treatment, allowing to adjust therapy to achieve the best results. Despite the fact that this paper focuses on the application in medical practice, this approach is also applicable to athletes, as in this field there is also a need to determine the quality of movement and correctness of exercise performance, monitoring the effectiveness of training.
The exercises were recorded on a digital video camera, and the video data were processed by a program complex. The program detects human contours on the video sequence, finding reference points and lines on each frame. Tracking the changes from frame to frame, allows you to assess the quality of movement. The program allows you to visually verify the presence or absence of symmetry and mobility problems, as well as to compare the results of one subject throughout the rehabilitation process. Object recognition in an image is performed using a pre-trained neural model for image segmentation. This model supports flexible cues and outputs segmentation masks in real time when prompted to interactively use Segment Anything (SAM).
This program is the first step in creating a motion quality assessment suite that includes analysis of both static and dynamic tests.

Speaker

Victoria Vladimirovna Smirnova
LeonRod LLC, Kazan, Russia
Russia

Discussion

Ask question