The Medical SAM (Segment Anything Model) repository is a fork of the original SAM repository with modifications to support object segmentation in medical imaging using DICOM files. The SAM model is a state-of-the-art object segmentation model that predicts object masks given prompts that indicate the desired object. This implementation uses SAM to efficiently produce high-quality masks from prompts for medical imaging tasks using DICOM files. It allows the user to provide box prompt via the SamPredictor class to predict masks for a given medical DICOM file.
This repository provides a demo notebook demonstrating how to use SAM for medical imaging. The notebook includes:
- Lung CT: Examples of segmenting lung structures from CT DICOM files.
- Axial CBCT View: Examples of segmenting regions of interest in axial CBCT slices.
-
Lung CT Segmentation Example
-
Axial CBCT Segmentation Example
A special shoutout to Mohammed El Amine Mokhtari for his great tutorials on medical imaging and computer vision. His work has been an inspiration and a valuable resource for the community!