MedSAM: Segment Anything in Medical Images

Kalbe Digital Lab

Overview

MedSAM, a foundation model for universal medical image segmentation. MedSAM is adapted from the SAM model on an unprecedented scale, with more than one million medical image-mask pairs.
Reference: https://arxiv.org/abs/2204.05798

Dataset

The model is trained using a diverse and large-scale medical image segmentation dataset with 1,090,486 medical image-mask pairs, covering 15 imaging modalities, over 30 cancer types, and a multitude of imaging protocols.

  • Target: Capturing a broad spectrum of anatomies and lesions across different modalities.
  • Task: Segmentation
  • Modality: Computed Tomography (CT), Magnetic Resonance Imaging (MRI), Endoscopy, Ultrasound, Pathology, Fundus, Dermoscopy, Mammography, and Optical Coherence Tomography (OCT).

Model Architecture

Parameterized Hypercomplex ResNets-18 Variants (PHYResNet).

model-architecture

Demo

Please select the example below or upload 2 pairs of mammography exam result.