Segmentation-Aware Generative Reinforcement Network (GRN) for Tissue Layer Segmentation in 3-D Ultrasound Images for Chronic Low-back Pain (cLBP) Assessment

2025-01-29

AI Summary

To address the high annotation cost and insufficient accuracy/efficiency in multi-layer soft-tissue segmentation of 3D ultrasound images for chronic low back pain assessment, this paper proposes an end-to-end jointly optimized Generative Reinforcement Network (GRN). We introduce the first segmentation-aware generative reinforcement learning framework, incorporating a Segmentation-Guided Enhancement (SGE) mechanism that enables co-training of image enhancement and segmentation. Two variants—sample-efficient (GRN-SEL) and semi-supervised (GRN-SSL)—are further designed. The method integrates generative adversarial networks (GANs), policy-gradient reinforcement learning, segmentation-loss-driven backward guidance, multi-scale 3D U-Net architecture, and semi-supervised consistency regularization. Evaluated on 69 clinical 3D ultrasound volumes for six-tissue segmentation, GRN-SEL+SGE reduces annotation requirements by 70% while improving Dice score by 1.98%. All variants achieve performance comparable to fully supervised baselines, cutting labeling effort by 60–70%.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Computer Vision: Segmentation
• Machine Learning: Reinforcement Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• 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: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

We introduce a novel segmentation-aware joint training framework called generative reinforcement network (GRN) that integrates segmentation loss feedback to optimize both image generation and segmentation performance in a single stage. An image enhancement technique called segmentation-guided enhancement (SGE) is also developed, where the generator produces images tailored specifically for the segmentation model. Two variants of GRN were also developed, including GRN for sample-efficient learning (GRN-SEL) and GRN for semi-supervised learning (GRN-SSL). GRN's performance was evaluated using a dataset of 69 fully annotated 3D ultrasound scans from 29 subjects. The annotations included six anatomical structures: dermis, superficial fat, superficial fascial membrane (SFM), deep fat, deep fascial membrane (DFM), and muscle. Our results show that GRN-SEL with SGE reduces labeling efforts by up to 70% while achieving a 1.98% improvement in the Dice Similarity Coefficient (DSC) compared to models trained on fully labeled datasets. GRN-SEL alone reduces labeling efforts by 60%, GRN-SSL with SGE decreases labeling requirements by 70%, and GRN-SSL alone by 60%, all while maintaining performance comparable to fully supervised models. These findings suggest the effectiveness of the GRN framework in optimizing segmentation performance with significantly less labeled data, offering a scalable and efficient solution for ultrasound image analysis and reducing the burdens associated with data annotation.

Problem

• Chronic Low Back Pain
• 3D Ultrasound Imaging
• Tissue Layer Analysis

Innovation

• Generative Reinforcement Network (GRN)
• Segmentation Guided Enhancement (SGE)
• 3D Ultrasound Imaging