SAMRI: Segment Anything Model for MRI
This work addresses the need for accurate and efficient MRI segmentation for clinical decision-making, representing an incremental adaptation of an existing model to a specific domain.
The paper tackled the problem of poor generalization in MRI segmentation due to variable contrast and protocols by fine-tuning the Segment Anything Model (SAM) specifically for MRI, achieving a mean Dice score of 0.87 and reducing training time by 94% and parameters by 96% compared to full retraining.
Accurate magnetic resonance imaging (MRI) segmentation is crucial for clinical decision-making, but remains labor-intensive when performed manually. Convolutional neural network (CNN)-based methods can be accurate and efficient, but often generalize poorly to MRI's variable contrast, intensity inhomogeneity, and protocols. Although the transformer-based Segment Anything Model (SAM) has demonstrated remarkable generalizability in natural images, existing adaptations often treat MRI as another imaging modality, overlooking these modality-specific challenges. We present SAMRI, an MRI-specialized SAM trained and validated on 1.1 million labeled MR slices spanning whole-body organs and pathologies. We demonstrate that SAM can be effectively adapted to MRI by simply fine-tuning its mask decoder using a two-stage strategy, reducing training time by 94% and trainable parameters by 96% versus full-model retraining. Across diverse MRI segmentation tasks, SAMRI achieves a mean Dice of 0.87, delivering state-of-the-art accuracy across anatomical regions and robust generalization on unseen structures, particularly small and clinically important structures.