Background and Objective: Prototype-based methods improve interpretability by learning fine-grained part-prototypes; however, their visualization in the input pixel space is not always consistent with human-understandable biomarkers. In addition, well-known prototype-based approaches typically learn extremely granular prototypes that are less interpretable in medical imaging, where both the presence and extent of biomarkers and lesions are critical. Methods: To address these challenges, we propose PiPViT (Patch-based Visual Interpretable Prototypes), an inherently interpretable prototypical model for image recognition. Leveraging a vision transformer (ViT), PiPViT captures long-range dependencies among patches to learn robust, human-interpretable prototypes that approximate lesion extent only using image-level labels. Additionally, PiPViT benefits from contrastive learning and multi-resolution input processing, which enables effective localization of biomarkers across scales. Results: We evaluated PiPViT on retinal OCT image classification across four datasets, where it achieved competitive quantitative performance compared to state-of-the-art methods while delivering more meaningful explanations. Moreover, quantitative evaluation on a hold-out test set confirms that the learned prototypes are semantically and clinically relevant. We believe PiPViT can transparently explain its decisions and assist clinicians in understanding diagnostic outcomes. Github page: https://github.com/marziehoghbaie/PiPViT

Medical image segmentation is crucial for clinical diagnosis, yet existing models are limited by their reliance on explicit human instructions and lack the active reasoning capabilities to understand complex clinical questions. While recent advancements in multimodal large language models (MLLMs) have improved medical question-answering (QA) tasks, most methods struggle to generate precise segmentation masks, limiting their application in automatic medical diagnosis. In this paper, we introduce medical image reasoning segmentation, a novel task that aims to generate segmentation masks based on complex and implicit medical instructions. To address this, we propose MedSeg-R, an end-to-end framework that leverages the reasoning abilities of MLLMs to interpret clinical questions while also capable of producing corresponding precise segmentation masks for medical images. It is built on two core components: 1) a global context understanding module that interprets images and comprehends complex medical instructions to generate multi-modal intermediate tokens, and 2) a pixel-level grounding module that decodes these tokens to produce precise segmentation masks and textual responses. Furthermore, we introduce MedSeg-QA, a large-scale dataset tailored for the medical image reasoning segmentation task. It includes over 10,000 image-mask pairs and multi-turn conversations, automatically annotated using large language models and refined through physician reviews. Experiments show MedSeg-R's superior performance across several benchmarks, achieving high segmentation accuracy and enabling interpretable textual analysis of medical images.

In medical imaging, challenges are competitions that aim to provide a fair comparison of different methodologic solutions to a common problem. Challenges typically focus on addressing real-world problems, such as segmentation, detection, and prediction tasks, using various types of medical images and associated data. Here, we describe the organization and results of such a challenge to compare machine-learning models for predicting survival in patients with diffuse large B-cell lymphoma using a baseline ¹⁸F-FDG PET/CT radiomics dataset. <b>Methods:</b> This challenge aimed to predict progression-free survival (PFS) in patients with diffuse large B-cell lymphoma, either as a binary outcome (shorter than 2 y versus longer than 2 y) or as a continuous outcome (survival in months). All participants were provided with a radiomic training dataset, including the ground truth survival for designing a predictive model and a radiomic test dataset without ground truth. Figures of merit (FOMs) used to assess model performance were the root-mean-square error for continuous outcomes and the C-index for 1-, 2-, and 3-y PFS binary outcomes. The challenge was endorsed and initiated by the Society of Nuclear Medicine and Molecular Imaging AI Task Force. <b>Results:</b> Nineteen models for predicting PFS as a continuous outcome from 15 teams were received. Among those models, external validation identified 6 models showing similar performance to that of a simple general linear reference model using SUV and total metabolic tumor volumes (TMTV) only. Twelve models for predicting binary outcomes were submitted by 9 teams. External validation showed that 1 model had higher, but nonsignificant, C-index values compared with values obtained by a simple logistic regression model using SUV and TMTV. <b>Conclusion:</b> Some of the radiomic-based machine-learning models developed by participants showed better FOMs than did simple linear or logistic regression models based on SUV and TMTV only, although the differences in observed FOMs were nonsignificant. This suggests that, for the challenge dataset, there was limited or no value seen from the addition of sophisticated radiomic features and use of machine learning when developing models for outcome prediction.

Medical image segmentation remains a challenging task due to the intricate nature of anatomical structures and the wide range of target sizes. In this paper, we propose a novel U -shaped segmentation network that integrates CNN and Transformer architectures to address these challenges. Specifically, our network architecture consists of three main components. In the encoder, we integrate an attention-guided multi-scale feature extraction module with a dual-path downsampling block to learn hierarchical features. The decoder employs an advanced feature aggregation and fusion module that effectively models inter-dependencies across different hierarchical levels. For the bottleneck, we explore multi-scale feature activation and multi-layer context Transformer modules to facilitate high-level semantic feature learning and global context modeling. Additionally, we implement a multi-resolution input-output strategy throughout the network to enrich feature representations and ensure fine-grained segmentation outputs across different scales. The experimental results on diverse multi-modal medical image datasets (ultrasound, gastrointestinal polyp, MR, and CT images) demonstrate that our approach can achieve superior performance over state-of-the-art methods in both quantitative measurements and qualitative assessments. The code is available at https://github.com/zsj0577/MSAGHNet.

Three-dimensional ultrasound (US) aims to provide sonographers with the spatial relationships of anatomical structures, playing a crucial role in clinical diagnosis. Recently, deep-learning-based freehand 3-D US has made significant advancements. It reconstructs volumes by estimating transformations between images without external tracking. However, image-only reconstruction poses difficulties in reducing cumulative drift and further improving reconstruction accuracy, particularly in scenarios involving complex motion trajectories. In this context, we propose an enhanced motion network (MoNetV2) to enhance the accuracy and generalizability of reconstruction under diverse scanning velocities and tactics. First, we propose a sensor-based temporal and multibranch structure (TMS) that fuses image and motion information from a velocity perspective to improve image-only reconstruction accuracy. Second, we devise an online multilevel consistency constraint (MCC) that exploits the inherent consistency of scans to handle various scanning velocities and tactics. This constraint exploits scan-level velocity consistency (SVC), path-level appearance consistency (PAC), and patch-level motion consistency (PMC) to supervise interframe transformation estimation. Third, we distill an online multimodal self-supervised strategy (MSS) that leverages the correlation between network estimation and motion information to further reduce cumulative errors. Extensive experiments clearly demonstrate that MoNetV2 surpasses existing methods in both reconstruction quality and generalizability performance across three large datasets.

As a cornerstone of patient care, clinical decision-making significantly influences patient outcomes and can be enhanced by large language models (LLMs). Although LLMs have demonstrated remarkable performance, their application to visual question answering in medical imaging, particularly for reasoning-based diagnosis, remains largely unexplored. Furthermore, supervised fine-tuning for reasoning tasks is largely impractical due to limited data availability and high annotation costs. In this work, we introduce a zero-shot framework for reliable medical image diagnosis that enhances the reasoning capabilities of LLMs in clinical settings through test-time scaling. Given a medical image and a textual prompt, a vision-language model processes a medical image along with a corresponding textual prompt to generate multiple descriptions or interpretations of visual features. These interpretations are then fed to an LLM, where a test-time scaling strategy consolidates multiple candidate outputs into a reliable final diagnosis. We evaluate our approach across various medical imaging modalities -- including radiology, ophthalmology, and histopathology -- and demonstrate that the proposed test-time scaling strategy enhances diagnostic accuracy for both our and baseline methods. Additionally, we provide an empirical analysis showing that the proposed approach, which allows unbiased prompting in the first stage, improves the reliability of LLM-generated diagnoses and enhances classification accuracy.

Different medical imaging modalities capture diagnostic information at varying spatial resolutions, from coarse global patterns to fine-grained localized structures. However, most existing vision-language frameworks in the medical domain apply a uniform strategy for local feature extraction, overlooking the modality-specific demands. In this work, we present MedMoE, a modular and extensible vision-language processing framework that dynamically adapts visual representation based on the diagnostic context. MedMoE incorporates a Mixture-of-Experts (MoE) module conditioned on the report type, which routes multi-scale image features through specialized expert branches trained to capture modality-specific visual semantics. These experts operate over feature pyramids derived from a Swin Transformer backbone, enabling spatially adaptive attention to clinically relevant regions. This framework produces localized visual representations aligned with textual descriptions, without requiring modality-specific supervision at inference. Empirical results on diverse medical benchmarks demonstrate that MedMoE improves alignment and retrieval performance across imaging modalities, underscoring the value of modality-specialized visual representations in clinical vision-language systems.

Safe deployment of machine learning (ML) models in safety-critical domains such as medical imaging requires detecting inputs with characteristics not seen during training, known as out-of-distribution (OOD) detection, to prevent unreliable predictions. Effective OOD detection after deployment could benefit from access to the training data, enabling direct comparison between test samples and the training data distribution to identify differences. State-of-the-art OOD detection methods, however, either discard training data after deployment or assume that test samples and training data are centrally stored together, an assumption that rarely holds in real-world settings. This is because shipping training data with the deployed model is usually impossible due to the size of training databases, as well as proprietary or privacy constraints. We introduce the Isolation Network, an OOD detection framework that quantifies the difficulty of separating a target test sample from the training data by solving a binary classification task. We then propose Decentralized Isolation Networks (DIsoN), which enables the comparison of training and test data when data-sharing is impossible, by exchanging only model parameters between the remote computational nodes of training and deployment. We further extend DIsoN with class-conditioning, comparing a target sample solely with training data of its predicted class. We evaluate DIsoN on four medical imaging datasets (dermatology, chest X-ray, breast ultrasound, histopathology) across 12 OOD detection tasks. DIsoN performs favorably against existing methods while respecting data-privacy. This decentralized OOD detection framework opens the way for a new type of service that ML developers could provide along with their models: providing remote, secure utilization of their training data for OOD detection services. Code will be available upon acceptance at: *****

Medical ultrasonography is an essential imaging technique for examining superficial organs and tissues, including lymph nodes, breast, and thyroid. It employs high-frequency ultrasound waves to generate detailed images of the internal structures of the human body. However, manually contouring regions of interest in these images is a labor-intensive task that demands expertise and often results in inconsistent interpretations among individuals. Vision-language foundation models, which have excelled in various computer vision applications, present new opportunities for enhancing ultrasound image analysis. Yet, their performance is hindered by the significant differences between natural and medical imaging domains. This research seeks to overcome these challenges by developing domain adaptation methods for vision-language foundation models. In this study, we explore the fine-tuning pipeline for vision-language foundation models by utilizing large language model as text refiner with special-designed adaptation strategies and task-driven heads. Our approach has been extensively evaluated on six ultrasound datasets and two tasks: segmentation and classification. The experimental results show that our method can effectively improve the performance of vision-language foundation models for ultrasound image analysis, and outperform the existing state-of-the-art vision-language and pure foundation models. The source code of this study is available at \href{https://github.com/jinggqu/NextGen-UIA}{GitHub}.

Different medical imaging modalities capture diagnostic information at varying spatial resolutions, from coarse global patterns to fine-grained localized structures. However, most existing vision-language frameworks in the medical domain apply a uniform strategy for local feature extraction, overlooking the modality-specific demands. In this work, we present MedMoE, a modular and extensible vision-language processing framework that dynamically adapts visual representation based on the diagnostic context. MedMoE incorporates a Mixture-of-Experts (MoE) module conditioned on the report type, which routes multi-scale image features through specialized expert branches trained to capture modality-specific visual semantics. These experts operate over feature pyramids derived from a Swin Transformer backbone, enabling spatially adaptive attention to clinically relevant regions. This framework produces localized visual representations aligned with textual descriptions, without requiring modality-specific supervision at inference. Empirical results on diverse medical benchmarks demonstrate that MedMoE improves alignment and retrieval performance across imaging modalities, underscoring the value of modality-specialized visual representations in clinical vision-language systems.