Training robust learning algorithms across different medical imaging modalities is challenging due to the large domain gap. Unsupervised domain adaptation (UDA) mitigates this problem by using annotated images from the source domain and unlabeled images from the target domain to train the deep models. Existing approaches often rely on GAN-based style transfer, but these methods struggle to capture cross-domain mappings in regions with high variability. In this paper, we propose a unified framework, B\'ezier Meets Diffusion, for cross-domain image generation. First, we introduce a B\'ezier-curve-based style transfer strategy that effectively reduces the domain gap between source and target domains. The transferred source images enable the training of a more robust segmentation model across domains. Thereafter, using pseudo-labels generated by this segmentation model on the target domain, we train a conditional diffusion model (CDM) to synthesize high-quality, labeled target-domain images. To mitigate the impact of noisy pseudo-labels, we further develop an uncertainty-guided score matching method that improves the robustness of CDM training. Extensive experiments on public datasets demonstrate that our approach generates realistic labeled images, significantly augmenting the target domain and improving segmentation performance.

Semantic segmentation is a fundamental task in medical image analysis, aiding medical decision-making by helping radiologists distinguish objects in an image. Research in this field has been driven by deep learning applications, which have the potential to scale these systems even in the presence of noise and artifacts. However, these systems are not yet perfected. We argue that performance can be improved by incorporating common medical knowledge into the segmentation model's loss function. To this end, we introduce Logic Tensor Networks (LTNs) to encode medical background knowledge using first-order logic (FOL) rules. The encoded rules span from constraints on the shape of the produced segmentation, to relationships between different segmented areas. We apply LTNs in an end-to-end framework with a SwinUNETR for semantic segmentation. We evaluate our method on the task of segmenting the hippocampus in brain MRI scans. Our experiments show that LTNs improve the baseline segmentation performance, especially when training data is scarce. Despite being in its preliminary stages, we argue that neurosymbolic methods are general enough to be adapted and applied to other medical semantic segmentation tasks.

Lightweight 3D medical image segmentation remains constrained by a fundamental "efficiency / robustness conflict", particularly when processing complex anatomical structures and heterogeneous modalities. In this paper, we study how to redesign the framework based on the characteristics of high-dimensional 3D images, and explore data synergy to overcome the fragile representation of lightweight methods. Our approach, VeloxSeg, begins with a deployable and extensible dual-stream CNN-Transformer architecture composed of Paired Window Attention (PWA) and Johnson-Lindenstrauss lemma-guided convolution (JLC). For each 3D image, we invoke a "glance-and-focus" principle, where PWA rapidly retrieves multi-scale information, and JLC ensures robust local feature extraction with minimal parameters, significantly enhancing the model's ability to operate with low computational budget. Followed by an extension of the dual-stream architecture that incorporates modal interaction into the multi-scale image-retrieval process, VeloxSeg efficiently models heterogeneous modalities. Finally, Spatially Decoupled Knowledge Transfer (SDKT) via Gram matrices injects the texture prior extracted by a self-supervised network into the segmentation network, yielding stronger representations than baselines at no extra inference cost. Experimental results on multimodal benchmarks show that VeloxSeg achieves a 26% Dice improvement, alongside increasing GPU throughput by 11x and CPU by 48x. Codes are available at https://github.com/JinPLu/VeloxSeg.

Recent advances in vision-language models (VLMs) have achieved remarkable performance on standard medical benchmarks, yet their true clinical reasoning ability remains unclear. Existing datasets predominantly emphasize classification accuracy, creating an evaluation illusion in which models appear proficient while still failing at high-stakes diagnostic reasoning. We introduce Neural-MedBench, a compact yet reasoning-intensive benchmark specifically designed to probe the limits of multimodal clinical reasoning in neurology. Neural-MedBench integrates multi-sequence MRI scans, structured electronic health records, and clinical notes, and encompasses three core task families: differential diagnosis, lesion recognition, and rationale generation. To ensure reliable evaluation, we develop a hybrid scoring pipeline that combines LLM-based graders, clinician validation, and semantic similarity metrics. Through systematic evaluation of state-of-the-art VLMs, including GPT-4o, Claude-4, and MedGemma, we observe a sharp performance drop compared to conventional datasets. Error analysis shows that reasoning failures, rather than perceptual errors, dominate model shortcomings. Our findings highlight the necessity of a Two-Axis Evaluation Framework: breadth-oriented large datasets for statistical generalization, and depth-oriented, compact benchmarks such as Neural-MedBench for reasoning fidelity. We release Neural-MedBench at https://neuromedbench.github.io/ as an open and extensible diagnostic testbed, which guides the expansion of future benchmarks and enables rigorous yet cost-effective assessment of clinically trustworthy AI.

Multimodal large language models (MLLMs) have shown remarkable potential in various domains, yet their application in the medical field is hindered by several challenges. General-purpose MLLMs often lack the specialized knowledge required for medical tasks, leading to uncertain or hallucinatory responses. Knowledge distillation from advanced models struggles to capture domain-specific expertise in radiology and pharmacology. Additionally, the computational cost of continual pretraining with large-scale medical data poses significant efficiency challenges. To address these issues, we propose InfiMed-Foundation-1.7B and InfiMed-Foundation-4B, two medical-specific MLLMs designed to deliver state-of-the-art performance in medical applications. We combined high-quality general-purpose and medical multimodal data and proposed a novel five-dimensional quality assessment framework to curate high-quality multimodal medical datasets. We employ low-to-high image resolution and multimodal sequence packing to enhance training efficiency, enabling the integration of extensive medical data. Furthermore, a three-stage supervised fine-tuning process ensures effective knowledge extraction for complex medical tasks. Evaluated on the MedEvalKit framework, InfiMed-Foundation-1.7B outperforms Qwen2.5VL-3B, while InfiMed-Foundation-4B surpasses HuatuoGPT-V-7B and MedGemma-27B-IT, demonstrating superior performance in medical visual question answering and diagnostic tasks. By addressing key challenges in data quality, training efficiency, and domain-specific knowledge extraction, our work paves the way for more reliable and effective AI-driven solutions in healthcare. InfiMed-Foundation-4B model is available at \href{https://huggingface.co/InfiX-ai/InfiMed-Foundation-4B}{InfiMed-Foundation-4B}.

Clinical dataset labels are rarely certain as annotators disagree and confidence is not uniform across cases. Typical aggregation procedures, such as majority voting, obscure this variability. In simple experiments on medical imaging benchmarks, accounting for the confidence in binary labels significantly impacts model rankings. We therefore argue that machine-learning evaluations should explicitly account for annotation uncertainty using probabilistic metrics that directly operate on distributions. These metrics can be applied independently of the annotations' generating process, whether modeled by simple counting, subjective confidence ratings, or probabilistic response models. They are also computationally lightweight, as closed-form expressions have linear-time implementations once examples are sorted by model score. We thus urge the community to release raw annotations for datasets and to adopt uncertainty-aware evaluation so that performance estimates may better reflect clinical data.

Identifying objective neuroimaging biomarkers to forecast Alzheimer's disease (AD) progression is crucial for timely intervention. However, this task remains challenging due to the complex dysfunctions in the spatio-temporal characteristics of underlying brain networks, which are often overlooked by existing methods. To address these limitations, we develop an interpretable spatio-temporal graph neural network framework to predict future AD progression, leveraging dual Stochastic Differential Equations (SDEs) to model the irregularly-sampled longitudinal functional magnetic resonance imaging (fMRI) data. We validate our approach on two independent cohorts, including the Open Access Series of Imaging Studies (OASIS-3) and the Alzheimer's Disease Neuroimaging Initiative (ADNI). Our framework effectively learns sparse regional and connective importance probabilities, enabling the identification of key brain circuit abnormalities associated with disease progression. Notably, we detect the parahippocampal cortex, prefrontal cortex, and parietal lobule as salient regions, with significant disruptions in the ventral attention, dorsal attention, and default mode networks. These abnormalities correlate strongly with longitudinal AD-related clinical symptoms. Moreover, our interpretability strategy reveals both established and novel neural systems-level and sex-specific biomarkers, offering new insights into the neurobiological mechanisms underlying AD progression. Our findings highlight the potential of spatio-temporal graph-based learning for early, individualized prediction of AD progression, even in the context of irregularly-sampled longitudinal imaging data.

In clinical applications, the utility of segmentation models is often based on the accuracy of derived downstream metrics such as organ size, rather than by the pixel-level accuracy of the segmentation masks themselves. Thus, uncertainty quantification for such metrics is crucial for decision-making. Conformal prediction (CP) is a popular framework to derive such principled uncertainty guarantees, but applying CP naively to the final scalar metric is inefficient because it treats the complex, non-linear segmentation-to-metric pipeline as a black box. We introduce COMPASS, a practical framework that generates efficient, metric-based CP intervals for image segmentation models by leveraging the inductive biases of their underlying deep neural networks. COMPASS performs calibration directly in the model's representation space by perturbing intermediate features along low-dimensional subspaces maximally sensitive to the target metric. We prove that COMPASS achieves valid marginal coverage under exchangeability and nestedness assumptions. Empirically, we demonstrate that COMPASS produces significantly tighter intervals than traditional CP baselines on four medical image segmentation tasks for area estimation of skin lesions and anatomical structures. Furthermore, we show that leveraging learned internal features to estimate importance weights allows COMPASS to also recover target coverage under covariate shifts. COMPASS paves the way for practical, metric-based uncertainty quantification for medical image segmentation.

Cone-beam computed tomography (CBCT) is widely used for image-guided radiotherapy (IGRT). It provides real time visualization at low cost and dose. However, photon scattering and beam hindrance cause artifacts in CBCT. These include inaccurate Hounsfield Units (HU), reducing reliability for dose calculation, and adaptive planning. By contrast, computed tomography (CT) offers better image quality and accurate HU calibration but is usually acquired offline and fails to capture intra-treatment anatomical changes. Thus, accurate CBCT-to-CT synthesis is needed to close the imaging-quality gap in adaptive radiotherapy workflows. To cater to this, we propose a novel diffusion-based conditional generative model, coined EqDiff-CT, to synthesize high-quality CT images from CBCT. EqDiff-CT employs a denoising diffusion probabilistic model (DDPM) to iteratively inject noise and learn latent representations that enable reconstruction of anatomically consistent CT images. A group-equivariant conditional U-Net backbone, implemented with e2cnn steerable layers, enforces rotational equivariance (cyclic C4 symmetry), helping preserve fine structural details while minimizing noise and artifacts. The system was trained and validated on the SynthRAD2025 dataset, comprising CBCT-CT scans across multiple head-and-neck anatomical sites, and we compared it with advanced methods such as CycleGAN and DDPM. EqDiff-CT provided substantial gains in structural fidelity, HU accuracy and quantitative metrics. Visual findings further confirm the improved recovery, sharper soft tissue boundaries, and realistic bone reconstructions. The findings suggest that the diffusion model has offered a robust and generalizable framework for CBCT improvements. The proposed solution helps in improving the image quality as well as the clinical confidence in the CBCT-guided treatment planning and dose calculations.

Accurate classification of Human epidermal growth factor receptor 2 (HER2) expression is crucial for guiding treatment in breast cancer, especially with emerging therapies like trastuzumab deruxtecan (T-DXd) for HER2-low patients. Current gold-standard methods relying on invasive biopsy and immunohistochemistry suffer from sampling bias and interobserver variability, highlighting the need for reliable non-invasive alternatives. We developed an artificial intelligence framework that integrates a pretrained foundation model with a task-specific classifier to predict HER2 expression categories (HER2-zero, HER2-low, HER2-positive) directly from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). The model was trained and validated using multicenter datasets. Model interpretability was assessed through feature visualization using t-SNE and UMAP dimensionality reduction techniques, complemented by SHAP analysis for post-hoc interpretation of critical predictive imaging features. The developed model demonstrated robust performance across datasets, achieving micro-average AUCs of 0.821 (95% CI 0.795–0.846) and 0.835 (95% CI 0.797–0.864), and macro-average AUCs of 0.833 (95% CI 0.818–0.847) and 0.857 (95% CI 0.837–0.872) in external validation. Subgroup analysis demonstrated strong discriminative power in distinguishing HER2 categories, particularly HER2-zero and HER2-low cases. Visualization techniques revealed distinct, biologically plausible clustering patterns corresponding to HER2 expression categories. This study presents a reproducible, non-invasive solution for comprehensive HER2 phenotyping using DCE-MRI, addressing fundamental limitations of biopsy-dependent assessment. Our approach enables accurate identification of HER2-low patients who may benefit from novel therapies like T-DXd. This framework represents a significant advancement in precision oncology, with potential to transform diagnostic workflows and guide targeted therapy selection in breast cancer care. The online version contains supplementary material available at 10.1186/s13058-025-02118-2.