E-BayesSAM

This repository provides the official implementation of our MICCAI 2025 accepted paper:
E-BayesSAM: Efficient Bayesian Adaptation of SAM with Self-Optimizing KAN-Based Interpretation for Uncertainty-Aware Ultrasonic Segmentation.

---

🔧 Getting Started

Inference

To perform segmentation on a sample image, run:

python inference_demo.py -p data_demo/images/amos_0004_75.png

Inference with Uncertainty Visualization

To visualize both the segmentation and its uncertainty map, run:

python inference_demo.py -p data_demo/images/amos_0004_75.png -u True

---

🙏 Acknowledgements

We sincerely thank the authors of SAM-Med2D and MedSAM for their open-source contributions, which laid the foundation for our development.