Latent Space Synergy: Text-Guided Data Augmentation for Direct Diffusion Biomedical Segmentation

2025-07-21

AI Summary

Medical image segmentation suffers from severe scarcity of annotated data—particularly for polyp detection, which demands expert domain knowledge. To address this, we propose a text-guided latent-space diffusion framework that synthesizes clinically realistic polyp images in a single step within the latent space, leveraging text-conditioned latent variable estimation and direct latent modeling. This approach avoids distribution shift while preserving both generation diversity and inference efficiency. Our method integrates latent diffusion models, text-guided inpainting, and an end-to-end segmentation network for effective data augmentation. Evaluated on CVC-ClinicDB, it achieves 96.0% Dice and 92.9% IoU scores, with inference speed accelerated by a factor of T, enabling real-time deployment in resource-constrained clinical settings. The core innovations lie in text-driven single-step latent synthesis and an unbiased latent variable estimation mechanism.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Medical image segmentation suffers from data scarcity, particularly in polyp detection where annotation requires specialized expertise. We present SynDiff, a framework combining text-guided synthetic data generation with efficient diffusion-based segmentation. Our approach employs latent diffusion models to generate clinically realistic synthetic polyps through text-conditioned inpainting, augmenting limited training data with semantically diverse samples. Unlike traditional diffusion methods requiring iterative denoising, we introduce direct latent estimation enabling single-step inference with T x computational speedup. On CVC-ClinicDB, SynDiff achieves 96.0% Dice and 92.9% IoU while maintaining real-time capability suitable for clinical deployment. The framework demonstrates that controlled synthetic augmentation improves segmentation robustness without distribution shift. SynDiff bridges the gap between data-hungry deep learning models and clinical constraints, offering an efficient solution for deployment in resourcelimited medical settings.

Problem

• Addresses data scarcity in medical image segmentation
• Generates synthetic polyps using text-guided diffusion models
• Enables real-time segmentation with computational efficiency

Innovation

• Text-guided synthetic data generation for augmentation
• Direct latent estimation enabling single-step inference
• Latent diffusion models for clinically realistic polyps