TumorGen: Boundary-Aware Tumor-Mask Synthesis with Rectified Flow Matching

2025-05-30

AI Summary

Existing tumor synthesis methods suffer from three key limitations: (1) predefined templates constrain morphological diversity; (2) two-stage denoising incurs high computational overhead; and (3) binary masks fail to model gradual tumor boundary transitions. To address these, we propose the first boundary-aware, single-stage 3D tumor mask synthesis framework, comprising: (i) boundary-aware pseudo-mask generation; (ii) spatially constrained vector field estimation; and (iii) a Rectified Flow Matching (RFM)-guided VAE refinement module. Notably, this work introduces RFM—the first application of rectified flow matching to medical image synthesis—enabling progressive, differentiable modeling of tumor boundaries. Evaluated on multicenter datasets, our method significantly outperforms state-of-the-art approaches, reducing sampling steps by over 60%, while generating pathologically realistic tumor masks with high fidelity. This advances scalable, annotation-efficient tumor synthesis and supports high-fidelity AI-assisted cancer diagnosis.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Matrix & Tensor Methods
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• 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

Abstract

Tumor data synthesis offers a promising solution to the shortage of annotated medical datasets. However, current approaches either limit tumor diversity by using predefined masks or employ computationally expensive two-stage processes with multiple denoising steps, causing computational inefficiency. Additionally, these methods typically rely on binary masks that fail to capture the gradual transitions characteristic of tumor boundaries. We present TumorGen, a novel Boundary-Aware Tumor-Mask Synthesis with Rectified Flow Matching for efficient 3D tumor synthesis with three key components: a Boundary-Aware Pseudo Mask Generation module that replaces strict binary masks with flexible bounding boxes; a Spatial-Constraint Vector Field Estimator that simultaneously synthesizes tumor latents and masks using rectified flow matching to ensure computational efficiency; and a VAE-guided mask refiner that enhances boundary realism. TumorGen significantly improves computational efficiency by requiring fewer sampling steps while maintaining pathological accuracy through coarse and fine-grained spatial constraints. Experimental results demonstrate TumorGen's superior performance over existing tumor synthesis methods in both efficiency and realism, offering a valuable contribution to AI-driven cancer diagnostics.

Problem

• Addresses shortage of annotated medical tumor datasets
• Improves computational inefficiency in tumor synthesis
• Enhances realism of tumor boundary transitions

Innovation

• Boundary-Aware Pseudo Mask Generation module
• Spatial-Constraint Vector Field Estimator
• VAE-guided mask refiner