COMPREHENSIVE BONE STRUCTURE SEGMENTATION AND 3D RECONSTRUCTION FROM MRI WITH MASK R-CNN

Accurate segmentation and three-dimensional (3D) reconstruction of the femur, tibia, and patella are vital for biomechanical modeling, orthopedic surgery, and personalized treatment planning. Traditional segmentation methods face challenges due to the complex anatomy and low contrast in MRI scans. This study introduces a deep learning–based framework using mask R-CNN for automated segmentation and high-fidelity 3D reconstruction of these bones from MRI images. The approach involves preprocessing MRI data, training a mask R-CNN model, automating bone segmentation, and rendering precise 3D models. The framework uses a backbone network for feature extraction and a region proposal network for bone localization. These segmented masks are then processed to create accurate 3D models that are biomechanically viable for further analysis. The mask R-CNN framework achieved a dice similarity coefficient (DSC) of [Formula: see text] and intersection over union (IoU) of [Formula: see text], outperforming traditional methods such as U-Net, fully convolutional networks (FCNs), and active shape models (ASMs). This system offers significant potential for clinical applications, improving surgical planning, diagnosis, and personalized treatments. Future work will focus on expanding this framework to other anatomical regions and incorporating multimodal imaging for enhanced clinical use.