OncoReg: Medical Image Registration for Oncological Challenges
The OncoReg Challenge has been introduced to address privacy concerns in medical image registration for cancer research. This challenge utilizes a two-phase framework, starting with public datasets and progressing to private datasets within secure hospital networks. Building on the Learn2Reg Challenge, OncoReg focuses on aligning interventional cone-beam computed tomography with planning fan-beam CT images for radiotherapy. Analysis of the challenge entries indicates that feature extraction is a critical component for successful registration, with combined deep learning and classical approaches proving most effective. AI
IMPACT Introduces a novel framework for AI model development in medical imaging that prioritizes patient privacy.