patient-specific intensity normalization

This study investigates whether general-purpose vision models can serve as viable alternatives to specialized architectures for 2D medical image segmentation. Under a unified training and evaluation protocol, the authors systematically compare eleven medical-specific models against multiple general vision models across three heterogeneous medical imaging datasets, complemented by Grad-CAM analyses to assess interpretability. The results demonstrate that, in most cases, general-purpose models achieve segmentation accuracy on par with or superior to specialized counterparts while effectively capturing clinically relevant anatomical structures. These findings challenge the prevailing assumption that domain-specific architectural design is indispensable for medical image segmentation and, for the first time, establish the feasibility and promise of general vision models on multimodal, heterogeneous medical data.

To address the underutilization of patient metadata (e.g., age, disease duration) in chronic wound segmentation—particularly diabetic foot ulcers (DFUs)—this work introduces the first framework modeling structured metadata as Gaussian random fields (GRFs) and embedding them into a multimodal image segmentation network. We propose a novel paradigm: “metadata-grouped training + distance-transform-weighted fusion”, wherein dedicated models are trained on subsets of metadata, and their predictions are dynamically ensemble-weighted based on patient feature similarity. Evaluated on the DFU2022 dataset, our method achieves an IoU of 0.4890 (+0.0220) and a Dice score of 0.6137 (+0.0229), demonstrating significant improvement in personalized lesion localization accuracy. All code is publicly available.

In whole-body medical image segmentation, segmented anatomical structures often exhibit geometric inaccuracies, compromising clinical utility. Method: This paper proposes a lightweight, plug-and-play post-processing framework—ShapeKit—that rectifies segmentation outputs during inference without model retraining. ShapeKit integrates explicit geometric constraints with morphological analysis to enforce anatomical plausibility, operating independently of the underlying segmentation model. Contribution/Results: By avoiding architectural modifications or costly retraining, ShapeKit significantly lowers deployment barriers. Evaluated on a multi-organ segmentation benchmark, it achieves an average Dice score improvement of over 8%, substantially exceeding the typical <3% gain from model-level enhancements. This demonstrates that shape-centric post-processing is both effective and practical for improving anatomical fidelity in medical image segmentation.

Evaluating anatomical segmentation models in the absence of ground-truth annotations remains a critical challenge in medical AI. Method: This paper proposes the first anatomy-term-driven unsupervised collaborative evaluation framework. It harmonizes and standardizes multi-model segmentation outputs at the anatomical-structure level via the JSON-Seg schema, enabling automated, cross-model and cross-structure comparison. The framework integrates a 3D Slicer plugin and OHIF Viewer for interactive visualization and provides Summary Plot analytics. Contribution/Results: Evaluated on the NLST CT dataset across 31 anatomical structures using six leading open-source models (e.g., TotalSegmentator, MOOSE), the framework successfully identified high inter-model consistency in lung segmentation and systematic failures in vertebral and rib segmentation—demonstrating its validity and practical utility for ground-truth-free model assessment.

To address the high computational complexity (O(N²)) and inadequate long-range dependency modeling of Transformers in medical image analysis, this work systematically reviews the application of the Mamba architecture across medical image classification, segmentation, and restoration tasks. We comprehensively categorize and analyze three classes of medical Mamba variants: pure Mamba, U-Net-enhanced Mamba, and hybrid models integrating CNNs, Transformers, or GNNs. Key innovations include optimized scanning strategies, multimodal alignment mechanisms, lightweight sequence modeling, and domain-specific adaptation to medical datasets. Experimental results demonstrate that Mamba achieves linear time complexity, reduces GPU memory consumption by 37% on average, accelerates inference by 2.1×, and matches or surpasses state-of-the-art accuracy across multiple benchmarks. This study elucidates the clinical potential and adaptation principles of state space models for efficient, scalable, and multimodal intelligent diagnosis and treatment.

To address critical challenges in 3D reconstruction from sparse medical slices (e.g., ultrasound, MRI)—including structural discontinuity, fine-detail loss, and high computational cost—this paper proposes the first joint modeling framework integrating 3D Gaussian splatting with learnable triplane feature encoding. The method introduces sparsity-aware geometric regularization and a multimodal collaborative rendering mechanism, significantly improving anatomical continuity, semantic consistency, and geometric plausibility. Evaluated on multimodal US/MRI datasets, our approach achieves state-of-the-art reconstruction quality, with an average PSNR gain of 5.8 dB and 3.2× faster inference speed compared to prior methods. Moreover, it demonstrates strong generalization across diverse anatomical regions and clinical scanning protocols, underscoring its practical applicability in real-world medical imaging workflows.

This study addresses the challenge of intensity inconsistency in MRI scans arising from differences in scanners and acquisition protocols by systematically evaluating seven intensity normalization methods—including Z-score, Nyúl histogram matching, CLAHE, and Gaussian mixture models—on the cross-domain generalization performance of a 3D U-Net for meniscus segmentation in knee MRI. Leveraging the IWOAI 2019 dataset for training and testing on both internal and external datasets such as SKM-TEA, this work presents the first comprehensive comparison of normalization strategies in 3D knee MRI segmentation. Results indicate that Z-score, Nyúl matching, and CLAHE exhibit relatively robust performance; however, the gains conferred by normalization are modest and insufficient to offset the substantial performance degradation caused by inter-dataset distribution shifts, underscoring the necessity of integrating additional robustness mechanisms to enhance generalization.

本文提出一种结合CLAHE的混合预处理方法，以提高MRI脑肿瘤边界检测准确性，通过优化局部对比度和自动化阈值选择来解决传统算法在噪声、复杂结构下的局限。

This work proposes the first systematic framework to enhance the interpretability of Transformer-based deep learning models in computational pathology by integrating class visualizations (CVs) and activation atlases (AAs) to analyze morphological features learned across multi-granular tissue and cancer classification tasks. The approach is rigorously validated through expert pathologist assessments and quantitative metrics, including Fleiss’ Kappa for inter-rater agreement, as well as perceptual and distributional similarity measures. Findings reveal that CVs remain interpretable in tissues with pronounced morphological distinctions, while AAs uncover a hierarchical, dependency-aware representational structure. Critically, strong correlations between expert consensus and atlas separability demonstrate that model representations effectively capture the inherent complexity of pathological patterns, thereby establishing a novel expert-in-the-loop paradigm for evaluating interpretability in medical AI.

This study addresses the core challenges in radiomics—feature instability, poor reproducibility, and limited clinical translation—by systematically evaluating how methodological choices across the end-to-end pipeline (including image acquisition, preprocessing, feature engineering, modeling, and evaluation) impact model robustness and generalizability. It is the first to comprehensively uncover the interdependencies among pipeline components and underscores the critical need for rigorous validation protocols to prevent data leakage and assessment bias. By integrating feature selection, dimensionality reduction, classical machine learning, and deep learning—and further exploring emerging paradigms such as hybrid AI, multimodal fusion, and federated learning—the work identifies key determinants of reliability and highlights persistent challenges related to standardization, domain shift, and clinical deployment, offering a systematic roadmap to enhance the quality and clinical applicability of radiomics research.