Chest X-ray (CXR) imaging is one of the most widely used diagnostic modalities in clinical practice, encompassing a broad spectrum of diagnostic tasks. Recent advancements have seen the extensive application of reasoning-based multimodal large language models (MLLMs) in medical imaging to enhance diagnostic efficiency and interpretability. However, existing multimodal models predominantly rely on "one-time" diagnostic approaches, lacking verifiable supervision of the reasoning process. This leads to challenges in multi-task CXR diagnosis, including lengthy reasoning, sparse rewards, and frequent hallucinations. To address these issues, we propose CX-Mind, the first generative model to achieve interleaved "think-answer" reasoning for CXR tasks, driven by curriculum-based reinforcement learning and verifiable process rewards (CuRL-VPR). Specifically, we constructed an instruction-tuning dataset, CX-Set, comprising 708,473 images and 2,619,148 samples, and generated 42,828 high-quality interleaved reasoning data points supervised by clinical reports. Optimization was conducted in two stages under the Group Relative Policy Optimization framework: initially stabilizing basic reasoning with closed-domain tasks, followed by transfer to open-domain diagnostics, incorporating rule-based conditional process rewards to bypass the need for pretrained reward models. Extensive experimental results demonstrate that CX-Mind significantly outperforms existing medical and general-domain MLLMs in visual understanding, text generation, and spatiotemporal alignment, achieving an average performance improvement of 25.1% over comparable CXR-specific models. On real-world clinical dataset (Rui-CXR), CX-Mind achieves a mean recall@1 across 14 diseases that substantially surpasses the second-best results, with multi-center expert evaluations further confirming its clinical utility across multiple dimensions.

Mitochondrial morphology and structural changes are closely associated with metabolic dysfunction and disease progression. However, the structural complexity of mitochondria presents a major challenge for accurate segmentation and analysis. Most existing methods focus on delineating entire mitochondria but lack the capability to resolve fine internal features, particularly cristae. In this study, we introduce MitoStructSeg, a deep learning-based framework for mitochondrial structure segmentation and quantitative analysis. The core of MitoStructSeg is AMM-Seg, a novel model that integrates domain adaptation to improve cross-sample generalization, dual-channel feature fusion to enhance structural detail extraction, and continuity learning to preserve spatial coherence. This architecture enables accurate segmentation of both mitochondrial membranes and intricately folded cristae. MitoStructSeg further incorporates a quantitative analysis module that extracts key morphological metrics, including surface area, volume, and cristae density, allowing comprehensive and scalable assessment of mitochondrial morphology. The effectiveness of our approach has been validated on both human myocardial tissue and mouse kidney tissue, demonstrating its robustness in accurately segmenting mitochondria with diverse morphologies. In addition, we provide an open source, user-friendly tool to ensure practical usability.

Artificial intelligence and large language models (LLMs)-particularly GPT-4 and GPT-4o-have demonstrated high correct-answer rates in medical examinations. GPT-4o has enhanced diagnostic capabilities, advanced image processing, and updated knowledge. Japanese surgeons face critical challenges, including a declining workforce, regional health care disparities, and work-hour-related challenges. Nonetheless, although LLMs could be beneficial in surgical education, no studies have yet assessed GPT-4o's surgical knowledge or its performance in the field of surgery. This study aims to evaluate the potential of GPT-4 and GPT-4o in surgical education by using them to take the Japan Surgical Board Examination (JSBE), which includes both textual questions and medical images-such as surgical and computed tomography scans-to comprehensively assess their surgical knowledge. We used 297 multiple-choice questions from the 2021-2023 JSBEs. The questions were in Japanese, and 104 of them included images. First, the GPT-4 and GPT-4o responses to only the textual questions were collected via OpenAI's application programming interface to evaluate their correct-answer rate. Subsequently, the correct-answer rate of their responses to questions that included images was assessed by inputting both text and images. The overall correct-answer rates of GPT-4o and GPT-4 for the text-only questions were 78% (231/297) and 55% (163/297), respectively, with GPT-4o outperforming GPT-4 by 23% (P=<.01). By contrast, there was no significant improvement in the correct-answer rate for questions that included images compared with the results for the text-only questions. GPT-4o outperformed GPT-4 on the JSBE. However, the results of the LLMs were lower than those of the examinees. Despite the capabilities of LLMs, image recognition remains a challenge for them, and their clinical application requires caution owing to the potential inaccuracy of their results.

Most artificial intelligence (AI) models for thyroid nodules are designed to screen for malignancy to guide further interventions; however, these models have not yet been fully implemented in clinical practice. This study aimed to evaluate AI in real clinical settings for identifying potentially benign thyroid nodules initially deemed to be at risk for malignancy by radiologists, reducing unnecessary fine needle aspiration (FNA) and optimizing management. We retrospectively collected a validation cohort of thyroid nodules that had undergone FNA. These nodules were initially assessed as "suspicious for malignancy" by radiologists based on ultrasound features, following standard clinical practice, which prompted further FNA procedures. Ultrasound images of these nodules were re-evaluated using a deep learning-based AI system, and its diagnostic performance was assessed in terms of correct identification of benign nodules and error identification of malignant nodules. Performance metrics such as sensitivity, specificity, and the area under the receiver operating characteristic curve were calculated. In addition, a separate comparison cohort was retrospectively assembled to compare the AI system's ability to correctly identify benign thyroid nodules with that of radiologists. The validation cohort comprised 4572 thyroid nodules (benign: n=3134, 68.5%; malignant: n=1438, 31.5%). AI correctly identified 2719 (86.8% among benign nodules) and reduced unnecessary FNAs from 68.5% (3134/4572) to 9.1% (415/4572). However, 123 malignant nodules (8.6% of malignant cases) were mistakenly identified as benign, with the majority of these being of low or intermediate suspicion. In the comparison cohort, AI successfully identified 81.4% (96/118) of benign nodules. It outperformed junior and senior radiologists, who identified only 40% and 55%, respectively. The area under the curve (AUC) for the AI model was 0.88 (95% CI 0.85-0.91), demonstrating a superior AUC compared with that of the junior radiologists (AUC=0.43, 95% CI 0.36-0.50; P=.002) and senior radiologists (AUC=0.63, 95% CI 0.55-0.70; P=.003). Compared with radiologists, AI can better serve as a "goalkeeper" in reducing unnecessary FNAs by identifying benign nodules that are initially assessed as malignant by radiologists. However, active surveillance is still necessary for all these nodules since a very small number of low-aggressiveness malignant nodules may be mistakenly identified.

<i>"Just Accepted" papers have undergone full peer review and have been accepted for publication in <i>Radiology: Artificial Intelligence</i>. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content.</i> Purpose To develop and optimize a federated learning (FL) framework across multiple clients for biparametric MRI prostate segmentation and clinically significant prostate cancer (csPCa) detection. Materials and Methods A retrospective study was conducted using Flower FL to train a nnU-Net-based architecture for MRI prostate segmentation and csPCa detection, using data collected from January 2010 to August 2021. Model development included training and optimizing local epochs, federated rounds, and aggregation strategies for FL-based prostate segmentation on T2-weighted MRIs (four clients, 1294 patients) and csPCa detection using biparametric MRIs (three clients, 1440 patients). Performance was evaluated on independent test sets using the Dice score for segmentation and the Prostate Imaging: Cancer Artificial Intelligence (PI-CAI) score, defined as the average of the area under the receiver operating characteristic curve and average precision, for csPCa detection. <i>P</i> values for performance differences were calculated using permutation testing. Results The FL configurations were independently optimized for both tasks, showing improved performance at 1 epoch 300 rounds using FedMedian for prostate segmentation and 5 epochs 200 rounds using FedAdagrad, for csPCa detection. Compared with the average performance of the clients, the optimized FL model significantly improved performance in prostate segmentation (Dice score increase from 0.73 ± 0.06 to 0.88 ± 0.03; <i>P</i> ≤ .01) and csPCa detection (PI-CAI score increase from 0.63 ± 0.07 to 0.74 ± 0.06; <i>P</i> ≤ .01) on the independent test set. The optimized FL model showed higher lesion detection performance compared with the FL-baseline model (PICAI score increase from 0.72 ± 0.06 to 0.74 ± 0.06; <i>P</i> ≤ .01), but no evidence of a difference was observed for prostate segmentation (Dice scores, 0.87 ± 0.03 vs 0.88 ± 03; <i>P</i> > .05). Conclusion FL enhanced the performance and generalizability of MRI prostate segmentation and csPCa detection compared with local models, and optimizing its configuration further improved lesion detection performance. ©RSNA, 2025.

Accurate monitoring of tumor progression is crucial for optimizing outcomes in neurofibromatosis type 2-related schwannomatosis. Standard 2D linear analysis on magnetic resonance imaging is less accurate than 3D volumetric analysis, but since 3D volumetric analysis is time-consuming, it is not widely used. To shorten the time required for 3D volumetric analysis, our lab has been developing an automated artificial intelligence-driven 3D volumetric tool. The objective of the study was to survey and interview clinicians treating neurofibromatosis type 2-related schwannomatosis to understand their views on current 2D analysis and to gather insights for the design of an artificial intelligence-driven 3D volumetric analysis tool. Interviews examined for the following themes: (1) shortcomings of the currently used linear analysis, (2) utility of 3D visualizations, (3) features of an interactive 3D modeling software, and (4) lack of a gold standard to assess the accuracy of 3D volumetric analysis. A Likert scale questionnaire was used to survey clinicians' levels of agreement with 25 statements related to 2D and 3D tumor analyses. A total of 14 clinicians completed a survey, and 12 clinicians were interviewed. Specialties ranged across neurosurgery, neuroradiology, neurology, oncology, and pediatrics. Overall, clinicians expressed concerns with current linear techniques, with clinicians agreeing that linear measurements can be variable with the possibility of two different clinicians calculating 2 different tumor sizes (mean 4.64, SD 0.49) and that volumetric measurements would be more helpful for determining clearer thresholds of tumor growth (mean 4.50, SD 0.52). For statements discussing the capabilities of a 3D volumetric analysis and visualization software, clinicians expressed strong interest in being able to visualize tumors with respect to critical brain structures (mean 4.36, SD 0.74) and in forecasting tumor growth (mean 4.77, SD 0.44). Clinicians were overall in favor of the adoption of 3D volumetric analysis techniques for measuring vestibular schwannoma tumors but expressed concerns regarding the novelty and inexperience surrounding these techniques. However, clinicians felt that the ability to visualize tumors with reference to critical structures, to overlay structures, to interact with 3D models, and to visualize areas of slow versus rapid growth in 3D would be valuable contributions to clinical practice. Overall, clinicians provided valuable insights for designing a 3D volumetric analysis tool for vestibular schwannoma tumor growth. These findings may also apply to other central nervous system tumors, offering broader utility in tumor growth assessments.

Large vessel occlusion (LVO) accounts for a third of all ischemic strokes. Artificial intelligence (AI) has shown good accuracy in identifying LVOs on computed tomography angiograms (CTA). We sought to analyze whether AI-adjudicated CTA improves workflow times and clinical outcomes in patients with confirmed LVOs. We systematically searched PubMed, Embase, and Web of Science for studies comparing initial radiological assessment assisted by AI softwares versus standard assessment of patients with acute LVO strokes. Results were pooled using a random-effects model as mean differences for continuous outcomes or odds ratio (OR) for dichotomous outcomes, along with 95% confidence intervals (CI). We included 9 studies comprising 1,270 patients, of whom 671 (52.8%) had AI-assisted radiological assessment. AI consistently improved treatment times when compared to standard assessment, as evidenced by a mean reduction of 20.55 minutes in door-to-groin time (95% CI -36.69 to -4.42 minutes; p<0.01) and a reduction of 14.99 minutes in CTA to reperfusion (95% CI -28.45 to -1.53 minutes; p=0.03). Functional independence, defined as a modified Rankin scale 0-2, occurred at similar rates in the AI-supported group and with the standard workflow (OR 1.27; 95% CI 0.92 to 1.76; p=0.14), as did mortality (OR 0.71; 95% CI 0.27 to 1.88; p=0.49). The incorporation of AI softwares for LVO detection in acute ischemic stroke enhanced workflow efficiency and was associated with decreased time to treatment. However, AI did not improve clinical outcomes as compared with standard assessment.

<i>"Just Accepted" papers have undergone full peer review and have been accepted for publication in <i>Radiology: Artificial Intelligence</i>. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content.</i> Purpose To examine common patterns among different computer-aided diagnosis (CAD) models for Alzheimer's disease (AD) using structural MRI data and to characterize the clinical and imaging features associated with their misclassifications. Materials and Methods This retrospective study utilized 3258 baseline structural MRIs from five multisite datasets and two multidisease datasets collected between September 2005 and December 2019. The 3D Nested Hierarchical Transformer (3DNesT) model and other CAD techniques were utilized for AD classification using 10-fold cross-validation and cross-dataset validation. Subgroup analysis of CAD-misclassified individuals compared clinical/neuroimaging biomarkers using independent <i>t</i> tests with Bonferroni correction. Results This study included 1391 patients with AD (mean age, 72.1 ± 9.2 years, 757 female), 205 with other neurodegenerative diseases (mean age, 64.9 ± 9.9 years, 117 male), and 1662 healthy controls (mean age, 70.6 ± 7.6 years, 935 female). The 3DNesT model achieved 90.1 ± 2.3% crossvalidation accuracy and 82.2%, 90.1%, and 91.6% in three external datasets. Further analysis suggested that false negative (FN) subgroup (<i>n</i> = 223) exhibited minimal atrophy and better cognitive performance than true positive (TP) subgroup (MMSE, FN, 21.4 ± 4.4; TP, 19.7 ± 5.7; <i>P_FWE</i> < 0.001), despite displaying similar levels of amyloid beta (FN, 705.9 ± 353.9; TP, 665.7 ± 305.8; <i>P_FWE</i> = 0.47), Tau (FN, 352.4 ± 166.8; TP, 371.0 ± 141.8; <i>P_FWE</i> = 0.47) burden. Conclusion FN subgroup exhibited atypical structural MRI patterns and clinical measures, fundamentally limiting the diagnostic performance of CAD models based solely on structural MRI. ©RSNA, 2025.

Feature matching is a cornerstone task in computer vision, essential for applications such as image retrieval, stereo matching, 3D reconstruction, and SLAM. This survey comprehensively reviews modality-based feature matching, exploring traditional handcrafted methods and emphasizing contemporary deep learning approaches across various modalities, including RGB images, depth images, 3D point clouds, LiDAR scans, medical images, and vision-language interactions. Traditional methods, leveraging detectors like Harris corners and descriptors such as SIFT and ORB, demonstrate robustness under moderate intra-modality variations but struggle with significant modality gaps. Contemporary deep learning-based methods, exemplified by detector-free strategies like CNN-based SuperPoint and transformer-based LoFTR, substantially improve robustness and adaptability across modalities. We highlight modality-aware advancements, such as geometric and depth-specific descriptors for depth images, sparse and dense learning methods for 3D point clouds, attention-enhanced neural networks for LiDAR scans, and specialized solutions like the MIND descriptor for complex medical image matching. Cross-modal applications, particularly in medical image registration and vision-language tasks, underscore the evolution of feature matching to handle increasingly diverse data interactions.

Diffuse midline glioma (DMG) is a rare, aggressive, and fatal tumor that largely occurs in the pediatric population. To improve outcomes, it is important to characterize DMGs, which can be performed via magnetic resonance imaging (MRI) assessment. Recently, artificial intelligence (AI) and advanced imaging have demonstrated their potential to improve the evaluation of various brain tumors, gleaning more information from imaging data than is possible without these methods. This narrative review compiles the existing literature on the intersection of MRI-based AI use and DMG tumors. The applications of AI in DMG revolve around classification and diagnosis, segmentation, radiogenomics, and prognosis/survival prediction. Currently published articles have utilized a wide spectrum of AI algorithms, from traditional machine learning and radiomics to neural networks. Challenges include the lack of cohorts of DMG patients with publicly available, multi-institutional, multimodal imaging and genomics datasets as well as the overall rarity of the disease. As an adjunct to AI, advanced MRI techniques, including diffusion-weighted imaging, perfusion-weighted imaging, and Magnetic Resonance Spectroscopy (MRS), as well as positron emission tomography (PET), provide additional insights into DMGs. Establishing AI models in conjunction with advanced imaging modalities has the potential to push clinical practice toward precision medicine.