This paper introduces a new Single Domain Generalization (SDG) method called ConDiSR for medical image classification, using channel-wise contrastive disentanglement and reconstruction-based style regularization. The method is evaluated on multicenter histopathology image classification, achieving a 1% improvement in average accuracy compared to state-of-the-art SDG baselines. Code is available at https://github.com/BioMedIA-MBZUAI/ConDiSR.
MBZUAI faculty, researchers, and students presented eight academic papers at the 25th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2022) in Singapore. Seven of the accepted papers feature a master’s or doctoral student as first author. The papers are the outcome of two MBZUAI faculty led labs – BioMedical Image Analysis (BioMedIA) lab and SPriNT-AI. Why it matters: This highlights MBZUAI's growing prominence in medical image analysis and AI, showcasing the university's commitment to producing high-quality research and fostering young talent in the field.