Flow Matching for Medical Image Synthesis: Bridging the Gap Between Speed and Quality

2025-03-01

AI Summary

To address the contradiction between scarce high-quality medical imaging data and slow generation speed in medical image synthesis, this paper proposes an optimal transport (OT)-guided flow matching generative framework. We pioneer the integration of OT-based geodesic paths into continuous-time flow matching, enabling flatter and more stable distribution alignment. A novel conditional embedding mechanism and a cross-dimensional convolutional architecture are introduced to jointly support 2D/3D modalities, multi-modal imaging (MRI/CT/X-ray), and multi-condition inputs (e.g., semantic labels or segmentation masks), while extending naturally to image enhancement tasks. Experiments demonstrate a 3–5× acceleration in inference speed and a 12% reduction in FID. Furthermore, the synthesized images yield consistent improvements of 2.1–4.3% in downstream classification and segmentation performance. The code, pre-trained models, and synthetic datasets are publicly released.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Deep learning models have emerged as a powerful tool for various medical applications. However, their success depends on large, high-quality datasets that are challenging to obtain due to privacy concerns and costly annotation. Generative models, such as diffusion models, offer a potential solution by synthesizing medical images, but their practical adoption is hindered by long inference times. In this paper, we propose the use of an optimal transport flow matching approach to accelerate image generation. By introducing a straighter mapping between the source and target distribution, our method significantly reduces inference time while preserving and further enhancing the quality of the outputs. Furthermore, this approach is highly adaptable, supporting various medical imaging modalities, conditioning mechanisms (such as class labels and masks), and different spatial dimensions, including 2D and 3D. Beyond image generation, it can also be applied to related tasks such as image enhancement. Our results demonstrate the efficiency and versatility of this framework, making it a promising advancement for medical imaging applications. Code with checkpoints and a synthetic dataset (beneficial for classification and segmentation) is now available on: https://github.com/milad1378yz/MOTFM.

Problem

• Accelerates medical image synthesis while maintaining quality
• Addresses slow inference times in generative models
• Supports diverse medical imaging modalities and tasks

Innovation

• Optimal transport flow matching accelerates image generation.
• Straighter mapping reduces inference time, enhances quality.
• Adaptable to various medical imaging modalities and tasks.