Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2506.01841

Beyond Pixel Agreement: Large Language Models as Clinical Guardrails for Reliable Medical Image Segmentation

Jiaxi Sheng·Leyi Yu·Haoyue Li·Yifan Gao·Xin Gao
View original source ↗Read source document ↗Export metadata ↓

Abstract

Evaluating AI-generated medical image segmentations for clinical acceptability poses a significant challenge, as traditional pixelagreement metrics often fail to capture true diagnostic utility. This paper introduces Hierarchical Clinical Reasoner (HCR), a novel framework that leverages Large Language Models (LLMs) as clinical guardrails for reliable, zero-shot quality assessment. HCR employs a structured, multistage prompting strategy that guides LLMs through a detailed reasoning process, encompassing knowledge recall, visual feature analysis, anatomical inference, and clinical synthesis, to evaluate segmentations. We evaluated HCR on a diverse dataset across six medical imaging tasks. Our results show that HCR, utilizing models like Gemini 2.5 Flash, achieved a classification accuracy of 78.12%, performing comparably to, and in instances exceeding, dedicated vision models such as ResNet50 (72.92% accuracy) that were specifically trained for this task. The HCR framework not only provides accurate quality classifications but also generates interpretable, step-by-step reasoning for its assessments. This work demonstrates the potential of LLMs, when appropriately guided, to serve as sophisticated evaluators, offering a pathway towards more trustworthy and clinically-aligned quality control for AI in medical imaging.

Explore related subjects

eess.IV

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Jiaxi Sheng, Leyi Yu, Haoyue Li, Yifan Gao, Xin Gao. 2025-06-02. Beyond Pixel Agreement: Large Language Models as Clinical Guardrails for Reliable Medical Image Segmentation. https://arxiv.org/abs/2506.01841

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

SCI-Mamba: Unsupervised Learning based Low-Light Image Enhancement for Non-Cooperative Spacecraft

Low-light visual perception acts as the core visual foundation for on-orbit servicing missions targeting non-cooperative spacecraft, supporting autonomous rendezvous, pose estimation, component detection and robotic capture operations. Spaceborne imagery suffers from severe low-light degradation, while the extreme scarcity of paired normal/low-light space samples severely limits the generalization capacity of supervised enhancement algorithms. To address this practical bottleneck, this paper proposes SCI-Mamba, an unsupervised enhancement network for low-light orbital spacecraft observations. The proposed framework unites self-calibrated unsupervised learning, linear-complexity VMamba architecture and Retinex physical priors, delivering a lightweight enhancement pipeline adaptable to resource-limited spaceborne hardware. We construct Space Dark-1.0, a dedicated low-light spacecraft dataset integrating real orbital footage, darkroom hardware-in-the-loop measurements and physically constrained synthetic data covering diverse illumination, motion and attitude conditions. Comprehensive comparisons with CNN-, Transformer- and prevailing Mamba-based approaches verify the advantages of SCI-Mamba in visual authenticity, color fidelity and inference speed. The proposed framework provides a practical low-light enhancement solution for close-proximity non-cooperative space operations. The code is available at https://github.com/bitswh/SCI-Mamba

eess.IV↗
arXiv2026

Recursive Uncertainty-Gated Image Registration for Learning-based Algorithms

Conventional image registration algorithms are robust to domain shifts and achieve low errors, but they are slow and computationally expensive. Deep-learning methods are efficient at inference-time, but face challenges in out-of-domain samples. We propose Recursive Uncertainty-Gated Image Registration (RUGI), an algorithm for iteratively refining deformation fields predicted by learning-based registration models. At each iteration, the registration model predicts an incremental deformation, and a gating map modulates the update. Refinements are hence concentrated in regions that remain difficult to register. We explore two gating strategies: a learned uncertainty-based approach and an image residual error approach. We evaluate RUGI on cardiac MRI and echocardiography datasets and show consistent improvements over single-step inference. Ablation experiments demonstrate that iterative refinement alone improves registration, but informative spatial gating provides a significant additional benefit. The error-gated variant of RUGI can also be applied directly to existing pretrained models; applied to VoxelMorph, TransMorph, and CycleMorph, it yields MSE reductions of 27-37% with no modification to the original training procedure. The improvements in registration performance are reflected in decreased errors in ejection fraction estimation relative to ground truths. These results demonstrate that spatially selective iterative refinement provides an effective strategy to improve registration accuracy at inference-time.

eess.IV↗
arXiv2026

STAMBRIDGE: Spectral-Temporal Amplitude-aware Mid-Feature Bridge for EEG Visual Decoding

Electroencephalography (EEG) visual decoding remains challenging due to the modality gap between low-SNR neural signals and highly structured vision--language spaces, making direct cross-modal alignment unstable. To address this, we propose STAMBRIDGE, a versatile two-stage framework that sequentially tackles feature conditioning and cross-modal alignment. First, we introduce a Spectral-Temporal Amplitude-aware Modulation (STAM) to extract well-conditioned EEG representations. By replacing hard frequency masking with amplitude-derived soft channel weighting and multi-scale temporal convolutions, STAM explicitly preserves frequency-aware transients while reducing the risk of time-domain ringing artifacts. Building upon these robust neural features, we further introduce a model-agnostic Mid-Feature Semantic Bridge (MFSB) that constructs a regularized intermediate space through directed cross-modal interactions, enabling staged distillation and more stable semantic alignment. Experiments on the THINGS-EEG benchmark show competitive 200-way zero-shot retrieval performance, with 34.50\% Top-1 and 65.95\% Top-5 accuracy. In addition, embeddings learned by STAMBRIDGE produce semantically coherent image reconstructions with a diffusion model, demonstrating robust EEG-to-vision semantic alignment. The code is available at: https://github.com/thabeatmjh/STAMBRIDGE.

eess.IV↗