Using machine learning techniques, we this project proposes to find an automated process for determining ligament insertion sites on subject specific knees. The main goal is to refine and improve the current processes of 3D modelling a subject specific knee and aid clinical orthopaedic processes. This project will determine a more accurate automated process for detecting ligament insertion sites within the human knee geometry. We are starting with ligament insertion sites on subject specific femur bones with pre-segmented geometry and insertion sites, however in the long term, we would like to be able to detect ligament insertion sites for the knee based solely on imaging data (Ie, MRI, CT, etc.)
Automated Determination of Ligament Insertion Sites - using Machine Learning
*Current methods of ligament insertion sites
*Uses for determining ligament insertion sites
*What we propose to do and how it differs
-- AIS aims to refine the current 3D knee modelling field by improving how we determine ligament insertion sites for subject specific knee models from imaging data. Being able to determine ligament insertion sites in a more accurate and automated manner benefits the biomedical community as a whole. It will cut down manual processing times, increasing 3D modelling accuracy, decreasing human error for segmenting, aid in determining consistent ligament insertion sites across users, and help inform clinical evaluation of subject specific knees (such as ligament morphology in subjects and ACL injuries). Therefore, our proposed framework will build on the existing literature but extend methods by applying machine learning for automated processing and recreation of tibia and femur ligament insertion sites.
*Current Stage Summary