Machine Learning Based Automatic Implant Size Prediction for Total Knee Arthroplasty Using Bone Dimensions

Predicting suitable implant sizes from 3D radiographic images of joint anatomy can be accomplished using templating methods. Automatic templating, which eliminates the need for manual intervention, is especially valuable for speeding up the creation of computer- or robot-assisted surgical plans. In our previous work, automatic templating for total knee arthroplasty was achieved through automatic bone segmentation, followed by matching a set of anatomical landmarks with corresponding points on candidate implants of various sizes. This paper introduces a novel approach that eliminates the reliance on point correspondences and matching, instead leveraging bone dimensions for implant size prediction. An experimental analysis on a dataset of 3261 knee cases demonstrates that the proposed method improves the performance of implant size prediction.