Appeared in: Proceedings of the National Conference of Undergraduate Research 2015 (NCUR 2015)
Abstract:
Shoulder rotator cuff surgery is one of the most common orthopedic surgeries performed today, particularly in adults over the age of 65. To restore range-of-motion after this surgery, physical therapy exercises are important, but are often not completed in full due to long recovery times, overly-optimistic patient expectations, and the cost of clinical appointments. We present an application that uses a depth camera (Microsoft Kinect) to aid in shoulder surgery physical therapy exercises by recognizing, measuring, and providing immediate feedback about exercises performed by a patient. This paper evaluates state-of-the-art machine learning algorithms for gesture and motion recognition from skeletal data for use in the aforementioned application. The application and several machine learning algorithms are evaluated using data from clinical trials of pre- and post-operative shoulder rotator cuff patients. Results demonstrate the efficacy of the approach.
BibTeX @inproceedings{macavaney:ncur2015-shoulder, author = {MacAvaney, Sean and Urbain, Jay}, title = {Automatic Recognition of Postoperative Shoulder Surgery Physical Therapy Exercises from Depth Camera Images}, booktitle = {Proceedings of the National Conference of Undergraduate Research}, year = {2015} }