SAMRI-2: A Memory-based Model for Cartilage and Meniscus Segmentation in 3D MRIs of the Knee Joint
The episode discusses 'SAMRI-2,' a memory-based AI model for segmenting cartilage and meniscus in 3D knee MRI scans. Hosts review its methodology, which uses a Hybrid Shuffling Strategy (HSS) to improve spatial consistency. The model demonstrated superior performance across multiple datasets, showing high accuracy and reduced effort for human annotators.