November 8, 2025Open Access

MedSAM2: Segment Anything in 3D Medical Images and Videos

View Full Paper

Ask AI

Bookmark

Authors

Discussion

This system improves segmentation for lesions in 3D images, demonstrating reduced manual costs in healthcare settings.

MedSAM2 reduces manual costs by over 85% for 3D image segmentation, enhancing efficiency in precision medicine.
The model is trained on a large dataset with 455,000 image-mask pairs and 76,000 video frames, leading to improved outcomes.
Observational analysis involved the annotation of 5,000 CT lesions and 3,984 liver MRI lesions, ensuring diverse coverage.
Integration into user-friendly platforms supports large-scale adaptation in research and healthcare environments.

Ma et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark