Summary
Our abstract emphasizes the need for an AI-driven single camera system in clinical settings that can help determine movement parameters to streamline the rehabilitation process and return-to-sport following an ACL injury and surgery.
Abstract
Purpose
Rehabilitation progression and return-to-sport guidelines vary greatly and are mostly based on expert opinion and subjective criteria. Laboratory based assessment of joint velocities and forces are resource dependent and not practical in the clinical setting. Video motion capture systems on smartphone devices offer novel biomechanical assessment capability. Our purpose is to validate an AI-driven single-camera system for analyzing human motion, compare it to a multi-camera system, and to train an artificial neural network. We hypothesize that an AI-driven single-camera system has the potential to accurately determine important movement parameters that could ultimately streamline rehabilitation progression and return-to-sport following ACL injury and surgery.
Methods
A sample of healthy controls, non-surgical, and post-surgical knee patients are currently enrolling in this pilot study. Inclusion criteria include those over the age of 18, deemed safe for participation by their treating physician, and willing to give consent. We analyze subjects performing movements of increasing complexity during a standard test battery using three iPhone cameras and two motion analysis systems (OpenCap and SMA4Athletes). OpenCap is a previously validated multi-camera motion capture system and SMA4Athletes is a recently introduced single camera motion capture system for the smartphone. The systems will compute positions of anatomic landmarks, joint angles, and joint forces during the test battery. Both systems will capture identical movements simultaneously. Results will be analyzed statistically.
Results
The study has enrolled 90/100 (90%) of target subjects to date. Our cohort consists of 50% females and has an average age of 41.3 years. All participants have completed the easy test battery (100%), 53% the intermediate test battery, and 22% completed the advanced test battery. In pilot analysis between OpenCap and SMA4Athletes for single leg squat, statistically significant correlations between the three- and one-camera system were found for pelvic tilt (p=0.04), hip adduction angle (p=0.04), trunk lean (p=0.04), and flexion depth (p<0.01). Data analysis for the other movements and kinematic parameters is ongoing.
Future directions: Our multi-disciplinary team intends to (1) train the neural network of SMA4Athletes and validate the single camera system against the multicamera video motion analysis (2) correlate biomechanical parameters with progression into specific rehabilitation phases and (3) correlate biomechanical parameters with return to sport functional assessment. The ultimate goal is to provide a rapid, inexpensive, portable, and objective assessment tool to facilitate rehabilitation and return to sport following ACL injury and surgery.