SynthFM: Training Modality-agnostic Foundation Models for Medical Image Segmentation without Real Medical Data

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sengupta, Sourya, Chakrabarty, Satrajit, Ravi, Keerthi Sravan, Avinash, Gopal, Soni, Ravi
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916685503528960
author Sengupta, Sourya
Chakrabarty, Satrajit
Ravi, Keerthi Sravan
Avinash, Gopal
Soni, Ravi
author_facet Sengupta, Sourya
Chakrabarty, Satrajit
Ravi, Keerthi Sravan
Avinash, Gopal
Soni, Ravi
contents Foundation models like the Segment Anything Model (SAM) excel in zero-shot segmentation for natural images but struggle with medical image segmentation due to differences in texture, contrast, and noise. Annotating medical images is costly and requires domain expertise, limiting large-scale annotated data availability. To address this, we propose SynthFM, a synthetic data generation framework that mimics the complexities of medical images, enabling foundation models to adapt without real medical data. Using SAM's pretrained encoder and training the decoder from scratch on SynthFM's dataset, we evaluated our method on 11 anatomical structures across 9 datasets (CT, MRI, and Ultrasound). SynthFM outperformed zero-shot baselines like SAM and MedSAM, achieving superior results under different prompt settings and on out-of-distribution datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2504_08177
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SynthFM: Training Modality-agnostic Foundation Models for Medical Image Segmentation without Real Medical Data
Sengupta, Sourya
Chakrabarty, Satrajit
Ravi, Keerthi Sravan
Avinash, Gopal
Soni, Ravi
Image and Video Processing
Artificial Intelligence
Computer Vision and Pattern Recognition
Foundation models like the Segment Anything Model (SAM) excel in zero-shot segmentation for natural images but struggle with medical image segmentation due to differences in texture, contrast, and noise. Annotating medical images is costly and requires domain expertise, limiting large-scale annotated data availability. To address this, we propose SynthFM, a synthetic data generation framework that mimics the complexities of medical images, enabling foundation models to adapt without real medical data. Using SAM's pretrained encoder and training the decoder from scratch on SynthFM's dataset, we evaluated our method on 11 anatomical structures across 9 datasets (CT, MRI, and Ultrasound). SynthFM outperformed zero-shot baselines like SAM and MedSAM, achieving superior results under different prompt settings and on out-of-distribution datasets.
title SynthFM: Training Modality-agnostic Foundation Models for Medical Image Segmentation without Real Medical Data
topic Image and Video Processing
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.08177