[PDF] Medical SAM 2: Segment medical images as video via Segment Anything Model 2 | Semantic Scholar (2025)

[PDF] Semantic Reader

A transfer learning pipeline is developed and it is demonstrated SAM2 can be quickly adapted to medical domain by fine-tuning and implemented as a 3D slicer plugin and Gradio API for efficient 3D image and video segmentation.

This paper explores the zero-shot capabilities of SAM 2, the next-generation Meta SAM model trained on videos, for 3D medical image segmentation and proposes a practical pipeline for using SAM 2 in 3D medical image segmentation and presents key findings highlighting its efficiency and potential for further optimization.

A new paradigm toward the universal medical image segmentation, termed 'One-Prompt Segmentation,' which combines the strengths of one-shot and interactive methods and can adeptly handle the unseen task in a single forward pass.

A data engine is built, which improves model and data via user interaction, to collect the largest video segmentation dataset to date, and a simple transformer architecture with streaming memory for real-time video processing is presented.

A fully-automatic framework is proposed and two models for segmentation of 45 Organs at Risk (OARs) and two Gross Tumor Volumes (GTVs) are developed and took second place for each of the tasks in the validation phase of the SegRap 2023 challenge.

The most comprehensive studies on applying SAM to medical 2D images and fine-tune the encoder and decoder of the original SAM to obtain a well-performed SAM-Med2D, leading to the most comprehensive fine-tuning strategies to date.

Overall KiTS21 facilitated a significant advancement in the state of the art in kidney tumor segmentation, and provides useful insights that are applicable to the field of semantic segmentation as a whole.

The ATLAS dataset is the first public dataset providing CE-MRI of HCC with annotations and should greatly facilitate the development of automated tools designed to optimise the delineation process, which is essential for treatment planning in liver cancer patients.

The proposed SAMed is built upon the large-scale image segmentation model, Segment Anything Model (SAM), to explore the new research paradigm of customizing large- scale models for medical images segmentation.

This paper proposes the Medical SAM Adapter (Med-SA), which incorporates domain-specific medical knowledge into the segmentation model using a light yet effective adaptation technique, and outperforms several state-of-the-art medical image segmentation methods, while updating only 2\% of the parameters.

SegGPT is presented, a generalist model for segmenting everything in context that accommodates different kinds of segmentation data by transforming them into the same format of images and shows strong capabilities in segmenting in-domain and out-of-domain targets.

The Segment Anything Model (SAM) is introduced: a new task, model, and dataset for image segmentation, and its zero-shot performance is impressive – often competitive with or even superior to prior fully supervised results.

...

...