← All Projects 
Medical Imaging · Deep Learning
MRI Segmentation & Localization for Osteonecrosis
Published
A deep-learning pipeline for segmenting femur, pelvis, and osteonecrotic lesions in MRI and registering them to intraoperative imaging to enable visualization during intervention. Includes 2D and 3D U-Net architectures.
Contributions
- Designed 2D/3D U-Net segmentation models for femur, pelvis, and necrotic lesion segmentation.
- Developed the registration pipeline to align MRI segmentations to intraoperative imaging.
- Evaluated segmentation and registration accuracy on clinical MRI data.