Segment Anything in Medical Images

TLDR

Medical image segmentation is essential for diagnosis, treatment planning, and monitoring, yet current methods rely on customized models with limited generality across tasks. The study introduces MedSAM, a foundation model designed for universal medical image segmentation. MedSAM is trained on a curated dataset of over one million medical images. MedSAM, trained on over one million images, outperforms state‑of‑the‑art foundation and specialist models, accurately extracts tumor‑burden biomarkers, and delivers efficient segmentation across diverse tasks, promising faster diagnostics and personalized treatment.

Abstract

Medical image segmentation is a critical component in clinical practice, facilitating accurate diagnosis, treatment planning, and disease monitoring. However, current methods predominantly rely on customized models, which exhibit limited generality across diverse tasks. In this study, we present MedSAM, the inaugural foundation model designed for universal medical image segmentation. Harnessing the power of a meticulously curated dataset comprising over one million images, MedSAM not only outperforms existing state-of-the-art segmentation foundation models, but also exhibits comparable or even superior performance to specialist models. Moreover, MedSAM enables the precise extraction of essential biomarkers for tumor burden quantification. By delivering accurate and efficient segmentation across a wide spectrum of tasks, MedSAM holds significant potential to expedite the evolution of diagnostic tools and the personalization of treatment plans.