src/MedSAM/comparisons/SAM/README.md

Introduction

The pre-trained SAM model was used as a baseline in this study. SAM has three model types: ViT-H, ViT-L, and ViT-B. We used ViT-B model since it has a good trade off between segmentation accuracy and performance based on their ablation study (Fig.13).

1. Download the checkpoint.

Note: We assuem that the preprocessing steps have been done.

2. For 2D images, run

python infer_SAM_2D_npz.py -i input_path -o output_path -m model_path

3. For 3D images, run

python infer_SAM_3D_npz.py -i input_path -o output_path -m model_path