Chest radiography frequently serves as baseline imaging for most lung diseases¹. Deep learning has great potential for automating the interpretation of chest radiography². However, existing chest radiographic deep learning models are limited in diagnostic scope, generalizability, adaptability, robustness and extensibility. To overcome these limitations, we have developed Ark⁺, a foundation model applied to chest radiography and pretrained by cyclically accruing and reusing the knowledge from heterogeneous expert labels in numerous datasets. Ark⁺ excels in diagnosing thoracic diseases. It expands the diagnostic scope and addresses potential misdiagnosis. It can adapt to evolving diagnostic needs and respond to novel diseases. It can learn rare conditions from a few samples and transfer to new diagnostic settings without training. It tolerates data biases and long-tailed distributions, and it supports federated learning to preserve privacy. All codes and pretrained models have been released, so that Ark⁺ is open for fine-tuning, local adaptation and improvement. It is extensible to several modalities. Thus, it is a foundation model for medical imaging. The exceptional capabilities of Ark⁺ stem from our insight: aggregating various datasets diversifies the patient populations and accrues knowledge from many experts to yield unprecedented performance while reducing annotation costs³. The development of Ark⁺ reveals that open models trained by accruing and reusing knowledge from heterogeneous expert annotations with a multitude of public (big or small) datasets can surpass the performance of proprietary models trained on large data. We hope that our findings will inspire more researchers to share code and datasets or federate privacy-preserving data to create open foundation models with diverse, global expertise and patient populations, thus accelerating open science and democratizing AI for medicine.

The rapid advancements in natural language processing, particularly the development of large language models (LLMs), have opened new avenues for managing complex clinical text data. However, the inherent complexity and specificity of medical texts present significant challenges for the practical application of prompt engineering in diagnostic tasks. This paper explores LLMs with new prompt engineering technology to enhance model interpretability and improve the prediction performance of pulmonary disease based on a traditional deep learning model. A retrospective dataset including 2965 chest CT radiology reports was constructed. The reports were from 4 cohorts, namely, healthy individuals and patients with pulmonary tuberculosis, lung cancer, and pneumonia. Then, a novel prompt engineering strategy that integrates feature summarization (F-Sum), chain of thought (CoT) reasoning, and a hybrid retrieval-augmented generation (RAG) framework was proposed. A feature summarization approach, leveraging term frequency-inverse document frequency (TF-IDF) and K-means clustering, was used to extract and distill key radiological findings related to 3 diseases. Simultaneously, the hybrid RAG framework combined dense and sparse vector representations to enhance LLMs' comprehension of disease-related text. In total, 3 state-of-the-art LLMs, GLM-4-Plus, GLM-4-air (Zhipu AI), and GPT-4o (OpenAI), were integrated with the prompt strategy to evaluate the efficiency in recognizing pneumonia, tuberculosis, and lung cancer. The traditional deep learning model, BERT (Bidirectional Encoder Representations from Transformers), was also compared to assess the superiority of LLMs. Finally, the proposed method was tested on an external validation dataset consisted of 343 chest computed tomography (CT) report from another hospital. Compared with BERT-based prediction model and various other prompt engineering techniques, our method with GLM-4-Plus achieved the best performance on test dataset, attaining an F1-score of 0.89 and accuracy of 0.89. On the external validation dataset, F1-score (0.86) and accuracy (0.92) of the proposed method with GPT-4o were the highest. Compared to the popular strategy with manually selected typical samples (few-shot) and CoT designed by doctors (F1-score=0.83 and accuracy=0.83), the proposed method that summarized disease characteristics (F-Sum) based on LLM and automatically generated CoT performed better (F1-score=0.89 and accuracy=0.90). Although the BERT-based model got similar results on the test dataset (F1-score=0.85 and accuracy=0.88), its predictive performance significantly decreased on the external validation set (F1-score=0.48 and accuracy=0.78). These findings highlight the potential of LLMs to revolutionize pulmonary disease prediction, particularly in resource-constrained settings, by surpassing traditional models in both accuracy and flexibility. The proposed prompt engineering strategy not only improves predictive performance but also enhances the adaptability of LLMs in complex medical contexts, offering a promising tool for advancing disease diagnosis and clinical decision-making.

<b>Background:</b> Studies of artificial intelligence (AI) for lung nodule detection on CT have primarily been conducted in investigational settings and/or focused on lung cancer screening. <b>Objective:</b> To evaluate the impact of AI assistance on radiologists' diagnostic performance for detecting lung metastases on chest CT in patients with colorectal cancer (CRC) in real-world clinical practice and to assess the clinical utility of AI assistance in this setting. <b>Methods:</b> This retrospective study included patients with CRC who underwent chest CT as surveillance for lung metastasis from May 2020 to December 2020 (conventional interpretation) or May 2022 to December 2022 (AI-assisted interpretation). Between periods, the institution implemented a commercial AI lung nodule detection system. During the second period, radiologists interpreted examinations concurrently with AI-generated reports, using clinical judgment regarding whether to report AI-detected nodules. The reference standard for metastasis incorporated pathologic and clinical follow-up criteria. Diagnostic performance (sensitivity, specificity, accuracy), and clinical utility (diagnostic yield, false-referral rate, management changes after positive reports) were compared between groups based on clinical radiology reports. Net benefit was estimated using decision curve analysis equation. Standalone AI interpretation was evaluated. <b>Results:</b> The conventional interpretation group included 647 patients (mean age, 64±11 years; 394 men, 253 women; metastasis prevalence, 4.3%); AI-assisted interpretation group included 663 patients (mean age, 63±12 years; 381 men, 282 women; metastasis prevalence, 4.4%). The AI-assisted interpretation group compared with the conventional interpretation group showed higher sensitivity (72.4% vs 32.1%; p=.008), accuracy (98.5% vs 96.0%; p=.005), and frequency of management changes (55.2% vs 25.0%, p=.02), without significant difference in specificity (99.7% vs 98.9%; p=.11), diagnostic yield (3.2% vs 1.4%, p=.30) or false-referral rate (0.3% vs 1.1%, p=.10). AI-assisted interpretation had positive estimated net benefit across outcome ratios. Standalone AI correctly detected metastasis in 24 of 29 patients but had 381 false-positive detections in 634 patients without metastasis; only one AI false-positive was reported as positive by interpretating radiologists. <b>Conclusion:</b> AI assistance yielded increased sensitivity, accuracy, and frequency of management changes, without significantly changed specificity. False-positive AI results minimally impacted radiologists' interpretations. <b>Clinical Impact:</b> The findings support clinical utility of AI assistance for CRC metastasis surveillance.

Pragmatics plays a crucial role in effectively conveying messages across various social communication contexts. This aspect is frequently highlighted in the challenges experienced by children diagnosed with autism spectrum disorder (ASD). Notably, there remains a paucity of research investigating how the structural connectome (SC) predicts pragmatic language abilities within this population. Using diffusion tensor imaging (DTI) and deterministic tractography, we constructed the whole-brain white matter structural network (WMSN) in a cohort comprising 92 children with ASD and 52 typically developing (TD) preschoolers, matched for age and gender. We employed network-based statistic (NBS)-Predict, a novel methodology that integrates machine learning (ML) with NBS, to identify dysconnected subnetworks associated with ASD, and then to predict pragmatic language abilities based on the SC derived from the whole-brain WMSN in the ASD group. Initially, NBS-Predict identified a subnetwork characterized by 42 reduced connections across 37 brain regions (p = 0.01), achieving a highest classification accuracy of 79.4% (95% CI: 0.791 ~ 0.796). The dysconnected regions were predominantly localized within the brain's frontotemporal and subcortical areas, with the right superior medial frontal gyrus (SFGmed.R) emerging as the region exhibiting the most extensive disconnection. Moreover, NBS-Predict demonstrated that the optimal correlation coefficient between the predicted pragmatic language scores and the actual measured scores was 0.220 (95% CI: 0.174 ~ 0.265). This analysis revealed a significant association between the pragmatic language abilities of the ASD cohort and the white matter connections linking the SFGmed.R with the bilateral anterior cingulate gyrus (ACG). In summary, our findings suggest that the subnetworks displaying the most significant abnormal connections were concentrated in the frontotemporal and subcortical regions among the ASD group. Furthermore, the observed abnormalities in the white matter connection pathways between the SFGmed.R and ACG may underlie the neurobiological basis for pragmatic language deficits in preschool children with ASD.

Breast ultrasound is essential for evaluating breast nodules, with Breast Imaging Reporting and Data System (BI-RADS) providing standardized classification. However, interobserver variability among radiologists can affect diagnostic accuracy. Large language models (LLMs) like ChatGPT-4 have shown potential in medical imaging interpretation. This study explores its feasibility in improving BI-RADS classification consistency and malignancy prediction compared to radiologists. This study aims to evaluate the feasibility of using LLMs, particularly ChatGPT-4, to assess the consistency and diagnostic accuracy of standardized breast ultrasound imaging reports, using pathology as the reference standard. This retrospective study analyzed breast nodule ultrasound data from 671 female patients (mean 45.82, SD 9.20 years; range 26-75 years) who underwent biopsy or surgical excision at our hospital between June 2019 and June 2024. ChatGPT-4 was used to interpret BI-RADS classifications and predict benign versus malignant nodules. The study compared the model's performance to that of two senior radiologists (≥15 years of experience) and two junior radiologists (<5 years of experience) using key diagnostic metrics, including accuracy, sensitivity, specificity, area under the receiver operating characteristic curve, P values, and odds ratios with 95% CIs. Two diagnostic models were evaluated: (1) image interpretation model, where ChatGPT-4 classified nodules based on BI-RADS features, and (2) image-to-text-LLM model, where radiologists provided textual descriptions, and ChatGPT-4 determined malignancy probability based on keywords. Radiologists were blinded to pathological outcomes, and BI-RADS classifications were finalized through consensus. ChatGPT-4 achieved an overall BI-RADS classification accuracy of 96.87%, outperforming junior radiologists (617/671, 91.95% and 604/671, 90.01%, P<.01). For malignancy prediction, ChatGPT-4 achieved an area under the receiver operating characteristic curve of 0.82 (95% CI 0.79-0.85), an accuracy of 80.63% (541/671 cases), a sensitivity of 90.56% (259/286 cases), and a specificity of 73.51% (283/385 cases). The image interpretation model demonstrated performance comparable to senior radiologists, while the image-to-text-LLM model further improved diagnostic accuracy for all radiologists, increasing their sensitivity and specificity significantly (P<.001). Statistical analyses, including the McNemar test and DeLong test, confirmed that ChatGPT-4 outperformed junior radiologists (P<.01) and showed noninferiority compared to senior radiologists (P>.05). Pathological diagnoses served as the reference standard, ensuring robust evaluation reliability. Integrating ChatGPT-4 into an image-to-text-LLM workflow improves BI-RADS classification accuracy and supports radiologists in breast ultrasound diagnostics. These results demonstrate its potential as a decision-support tool to enhance diagnostic consistency and reduce variability.

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.

Renal cell carcinoma (RCC) represents the most prevalent malignant neoplasm of the kidney, with a rising global incidence. Tumor nuclear grade is a crucial prognostic factor, guiding treatment decisions, but current histopathological grading via biopsy is invasive and prone to sampling errors. This study aims to assess the diagnostic performance and quality of CT-based radiomics for preoperatively predicting RCC nuclear grade. A comprehensive search was conducted across PubMed, Scopus, Embase, and Web of Science to identify relevant studies up until 19 April 2025. Quality was assessed using the QUADAS-2 and METRICS tools. A bivariate random-effects meta-analysis was performed to evaluate model performance, including sensitivity, specificity, and Area Under the Curve (AUC). Results from separate validation cohorts were pooled, and clinical and combined models were analyzed separately in distinct analyses. A total of 26 studies comprising 1993 individuals in 10 external and 16 internal validation cohorts were included. Meta-analysis of radiomics models showed pooled AUC of 0.88, sensitivity of 0.78, and specificity of 0.82. Clinical and combined (clinical-radiomics) models showed AUCs of 0.73 and 0.86, respectively. QUADAS-2 revealed significant risk of bias in the Index Test and Flow and Timing domains. METRICS scores ranged from 49.7 to 88.4%, with an average of 66.65%, indicating overall good quality, though gaps in some aspects of study methodologies were identified. This study suggests that radiomics models show great potential and diagnostic accuracy for non-invasive preoperative nuclear grading of RCC. However, challenges related to generalizability and clinical applicability remain, as further research with standardized methodologies, external validation, and larger cohorts is needed to enhance their reliability and integration into routine clinical practice.

Early diagnosis of Alzheimer's Disease (AD), especially at the mild cognitive impairment (MCI) stage, is vital yet hindered by subjective assessments and the high cost of multimodal imaging modalities. Although deep learning methods offer automated alternatives, their energy inefficiency and computational demands limit real-world deployment, particularly in resource-constrained settings. As a brain-inspired paradigm, spiking neural networks (SNNs) are inherently well-suited for modeling the sparse, event-driven patterns of neural degeneration in AD, offering a promising foundation for interpretable and low-power medical diagnostics. However, existing SNNs often suffer from weak expressiveness and unstable training, which restrict their effectiveness in complex medical tasks. To address these limitations, we propose FasterSNN, a hybrid neural architecture that integrates biologically inspired LIF neurons with region-adaptive convolution and multi-scale spiking attention. This design enables sparse, efficient processing of 3D MRI while preserving diagnostic accuracy. Experiments on benchmark datasets demonstrate that FasterSNN achieves competitive performance with substantially improved efficiency and stability, supporting its potential for practical AD screening. Our source code is available at https://github.com/wuchangw/FasterSNN.

Magnetic resonance imaging of fetal and neonatal brains reveals rapid neurodevelopment marked by substantial anatomical changes unfolding within days. Studying this critical stage of the developing human brain, therefore, requires accurate brain models-referred to as atlases-of high spatial and temporal resolution. To meet these demands, established traditional atlases and recently proposed deep learning-based methods rely on large and comprehensive datasets. This poses a major challenge for studying brains in the presence of pathologies for which data remains scarce. We address this limitation with CINeMA (Conditional Implicit Neural Multi-Modal Atlas), a novel framework for creating high-resolution, spatio-temporal, multimodal brain atlases, suitable for low-data settings. Unlike established methods, CINeMA operates in latent space, avoiding compute-intensive image registration and reducing atlas construction times from days to minutes. Furthermore, it enables flexible conditioning on anatomical features including GA, birth age, and pathologies like ventriculomegaly (VM) and agenesis of the corpus callosum (ACC). CINeMA supports downstream tasks such as tissue segmentation and age prediction whereas its generative properties enable synthetic data creation and anatomically informed data augmentation. Surpassing state-of-the-art methods in accuracy, efficiency, and versatility, CINeMA represents a powerful tool for advancing brain research. We release the code and atlases at https://github.com/m-dannecker/CINeMA.