سامانه همایش‌ها TSTA | ثبت‌نام و ارسال مقاله کنفرانس‌ها و کنگره‌های ملی

صفحه اصلی / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

Gait Retraining of Musculoskeletal Patients Using Deep Learning Techniques

نویسندگان :

Kourosh Alimadadi¹ Masoud Shariat Panahi² Morad Karimpour³ Hadi Ghattan Kashani⁴

1- دانشگاه تهران 2- دانشگاه تهران 3- دانشگاه تهران 4- دانشگاه تهران

کلمات کلیدی :

Skeletal abnormalities،Gait retraining،Depth image processing،Deep learning

چکیده :

Reconstructive surgeries and the use of bone and joint prostheses are among the most common treatments for lower-limb skeletal disorders. Patients undergoing such treatments often face challenges in regaining a proper gait pattern. To overcome these difficulties, they are typically required to participate in rehabilitation programs where correct walking strategies are taught by instructors. However, these programs demand specialized equipment and experienced trainers, which are both costly and time-consuming, and not easily accessible to all patients. In this paper, we propose a novel method for gait extraction, analysis, and correction in musculoskeletal patients using depth-sensing camera (RGBD) and deep learning techniques. The three-dimensional coordinates of anatomical key points in the patient’s lower limbs are automatically extracted from gait videos, and the corresponding motion patterns are represented through spatial variation graphs of these key points. Subsequently, inverse kinematics analysis of the motion pattern is performed to derive variations in anatomical indicators (distances and angles) across the main gait phases, including Toe-off, Mid-Stance, Mid Swing, and Heel-Strike. By comparing these indicators with those of healthy individuals, the system evaluates the extent and nature of gait deviations that require correction. Finally, the proposed framework provides recommendations for adjusting the patient’s gait and aligning it more closely with healthy walking patterns. Results from multiple case studies demonstrate that the proposed approach can significantly improve gait performance in the post-surgery phase and substantially reduce musculoskeletal complications caused by improper walking.