In intensive care units (ICUs), patients with complex clinical conditions require vigilant monitoring and prompt interventions. Chest X-rays (CXRs) are a vital diagnostic tool, providing insights into clinical trajectories, but their irregular acquisition limits their utility. Existing tools for CXR interpretation are constrained by cross-sectional analysis, failing to capture temporal dynamics. To address this, we introduce CXR-TFT, a novel multi-modal framework that integrates temporally sparse CXR imaging and radiology reports with high-frequency clinical data, such as vital signs, laboratory values, and respiratory flow sheets, to predict the trajectory of CXR findings in critically ill patients. CXR-TFT leverages latent embeddings from a vision encoder that are temporally aligned with hourly clinical data through interpolation. A transformer model is then trained to predict CXR embeddings at each hour, conditioned on previous embeddings and clinical measurements. In a retrospective study of 20,000 ICU patients, CXR-TFT demonstrated high accuracy in forecasting abnormal CXR findings up to 12 hours before they became radiographically evident. This predictive capability in clinical data holds significant potential for enhancing the management of time-sensitive conditions like acute respiratory distress syndrome, where early intervention is crucial and diagnoses are often delayed. By providing distinctive temporal resolution in prognostic CXR analysis, CXR-TFT offers actionable 'whole patient' insights that can directly improve clinical outcomes.

Recent advances in medical image segmentation have been driven by deep learning; however, most existing methods remain limited by modality-specific designs and exhibit poor adaptability to dynamic medical imaging scenarios. The Segment Anything Model 2 (SAM2) and its related variants, which introduce a streaming memory mechanism for real-time video segmentation, present new opportunities for prompt-based, generalizable solutions. Nevertheless, adapting these models to medical video scenarios typically requires large-scale datasets for retraining or transfer learning, leading to high computational costs and the risk of catastrophic forgetting. To address these challenges, we propose DD-SAM2, an efficient adaptation framework for SAM2 that incorporates a Depthwise-Dilated Adapter (DD-Adapter) to enhance multi-scale feature extraction with minimal parameter overhead. This design enables effective fine-tuning of SAM2 on medical videos with limited training data. Unlike existing adapter-based methods focused solely on static images, DD-SAM2 fully exploits SAM2's streaming memory for medical video object tracking and segmentation. Comprehensive evaluations on TrackRad2025 (tumor segmentation) and EchoNet-Dynamic (left ventricle tracking) datasets demonstrate superior performance, achieving Dice scores of 0.93 and 0.97, respectively. To the best of our knowledge, this work provides an initial attempt at systematically exploring adapter-based SAM2 fine-tuning for medical video segmentation and tracking. Code, datasets, and models will be publicly available at https://github.com/apple1986/DD-SAM2.

High intra-pancreatic fat deposition (IPFD) plays an important role in diseases of the pancreas. The intricate anatomy of the pancreas and the surrounding structures has historically made IPFD quantification a challenging measurement to make accurately on radiological images. To take on the challenge, automated IPFD quantification methods using artificial intelligence (AI) have recently been deployed. The aim was to benchmark the current knowledge on the use of AI-based models to measure IPFD automatedly. The search was conducted in the MEDLINE, Embase, Scopus, and IEEE Xplore databases. Studies were eligible if they used AI for both segmentation of the pancreas and quantification of IPFD. The ground truth was manual segmentation by radiologists. When possible, data were pooled statistically using a random-effects model. A total of 12 studies (10 cross-sectional and 2 longitudinal) encompassing more than 50 thousand people were included. Eight of the 12 studies used MRI, whereas four studies employed CT. U-Net model and nnU-Net model were the most frequently used AI-based models. The pooled Dice similarity coefficient of AI-based models in quantifying IPFD was 82.3% (95% confidence interval, 73.5 to 91.1%). The clinical application of AI-based models showed the relevance of high IPFD to acute pancreatitis, pancreatic cancer, and type 2 diabetes mellitus. Current AI-based models for IPFD quantification are suboptimal, as the dissimilarity between AI-based and manual quantification of IPFD is not negligible. Future advancements in fully automated measurements of IPFD will accelerate the accumulation of robust, large-scale evidence on the role of high IPFD in pancreatic diseases. KEY POINTS: Question What is the current evidence on the performance and clinical applicability of artificial intelligence-based models for automated quantification of intra-pancreatic fat deposition? Findings The nnU-Net model achieved the highest Dice similarity coefficient among MRI-based studies, whereas the nnTransfer model demonstrated the highest Dice similarity coefficient in CT-based studies. Clinical relevance Standardisation of reporting on artificial intelligence-based models for the quantification of intra-pancreatic fat deposition will be essential to enhancing the clinical applicability and reliability of artificial intelligence in imaging patients with diseases of the pancreas.

Magnetic Resonance Imaging (MRI) enables the acquisition of multiple image contrasts, such as T1-weighted (T1w) and T2-weighted (T2w) scans, each offering distinct diagnostic insights. However, acquiring all desired modalities increases scan time and cost, motivating research into computational methods for cross-modal synthesis. To address this, recent approaches aim to synthesize missing MRI contrasts from those already acquired, reducing acquisition time while preserving diagnostic quality. Image-to-image (I2I) translation provides a promising framework for this task. In this paper, we present a comprehensive benchmark of generative models$\unicode{x2013}$specifically, Generative Adversarial Networks (GANs), diffusion models, and flow matching (FM) techniques$\unicode{x2013}$for T1w-to-T2w 2D MRI I2I translation. All frameworks are implemented with comparable settings and evaluated on three publicly available MRI datasets of healthy adults. Our quantitative and qualitative analyses show that the GAN-based Pix2Pix model outperforms diffusion and FM-based methods in terms of structural fidelity, image quality, and computational efficiency. Consistent with existing literature, these results suggest that flow-based models are prone to overfitting on small datasets and simpler tasks, and may require more data to match or surpass GAN performance. These findings offer practical guidance for deploying I2I translation techniques in real-world MRI workflows and highlight promising directions for future research in cross-modal medical image synthesis. Code and models are publicly available at https://github.com/AndreaMoschetto/medical-I2I-benchmark.

To develop and validate a deep learning model based on arterial phase-enhanced CT for predicting the pathological grading of clear cell renal cell carcinoma (ccRCC). Data from 564 patients diagnosed with ccRCC from five distinct hospitals were retrospectively analyzed. Patients from centers 1 and 2 were randomly divided into a training set (n=283) and an internal test set (n=122). Patients from centers 3, 4, and 5 served as external validation sets 1 (n=60), 2 (n=38), and 3 (n=61), respectively. A 2D model, a 2.5D model (three-slice input), and a radiomics-based multi-layer perceptron (MLP) model were developed. Model performance was evaluated using the area under the curve (AUC), accuracy, and sensitivity. The 2.5D model outperformed the 2D and MLP models. Its AUCs were 0.959 (95% CI: 0.9438-0.9738) for the training set, 0.879 (95% CI: 0.8401-0.9180) for the internal test set, and 0.870 (95% CI: 0.8076-0.9334), 0.862 (95% CI: 0.7581-0.9658), and 0.849 (95% CI: 0.7766-0.9216) for the three external validation sets, respectively. The corresponding accuracy values were 0.895, 0.836, 0.827, 0.825, and 0.839. Compared to the MLP model, the 2.5D model achieved significantly higher AUCs (increases of 0.150 [p<0.05], 0.112 [p<0.05], and 0.088 [p<0.05]) and accuracies (increases of 0.077 [p<0.05], 0.075 [p<0.05], and 0.101 [p<0.05]) in the external validation sets. The 2.5D model based on 2.5D CT image input demonstrated improved predictive performance for the WHO/ISUP grading of ccRCC.

Image-to-image (I2I) translation networks have emerged as promising tools for generating synthetic medical images; however, their clinical reliability and ability to preserve diagnostically relevant features remain underexplored. This study evaluates the performance of state-of-the-art 2D/3D I2I networks for converting ultrasound (US) images to synthetic MRI in prostate cancer (PCa) imaging. The novelty lies in combining radiomics, expert clinical evaluation, and classification performance to comprehensively benchmark these models for potential integration into real-world diagnostic workflows. A dataset of 794 PCa patients was analyzed using ten leading I2I networks to synthesize MRI from US input. Radiomics feature (RF) analysis was performed using Spearman correlation to assess whether high-performing networks (SSIM > 0.85) preserved quantitative imaging biomarkers. A qualitative evaluation by seven experienced physicians assessed the anatomical realism, presence of artifacts, and diagnostic interpretability of synthetic images. Additionally, classification tasks using synthetic images were conducted using two machine learning and one deep learning model to assess the practical diagnostic benefit. Among all networks, 2D-Pix2Pix achieved the highest SSIM (0.855 ± 0.032). RF analysis showed that 76 out of 186 features were preserved post-translation, while the remainder were degraded or lost. Qualitative feedback revealed consistent issues with low-level feature preservation and artifact generation, particularly in lesion-rich regions. These evaluations were conducted to assess whether synthetic MRI retained clinically relevant patterns, supported expert interpretation, and improved diagnostic accuracy. Importantly, classification performance using synthetic MRI significantly exceeded that of US-based input, achieving average accuracy and AUC of ~ 0.93 ± 0.05. Although 2D-Pix2Pix showed the best overall performance in similarity and partial RF preservation, improvements are still required in lesion-level fidelity and artifact suppression. The combination of radiomics, qualitative, and classification analyses offered a holistic view of the current strengths and limitations of I2I models, supporting their potential in clinical applications pending further refinement and validation.

Cardiovascular illnesses continue to be a predominant cause of mortality globally, underscoring the necessity for prompt and precise diagnosis to mitigate consequences and healthcare expenditures. This work presents a complete hybrid methodology that integrates machine learning techniques with medical image analysis to improve the identification of cardiovascular diseases. This research integrates many imaging modalities such as echocardiography, cardiac MRI, and chest radiographs with patient health records, enhancing diagnosis accuracy beyond standard techniques that depend exclusively on numerical clinical data. During the preprocessing phase, essential visual elements are collected from medical pictures utilizing image processing methods and convolutional neural networks (CNNs). These are subsequently integrated with clinical characteristics and input into various machine learning classifiers, including Support Vector Machines (SVM), Random Forest (RF), XGBoost, and Deep Neural Networks (DNNs), to differentiate between healthy persons and patients with cardiovascular illnesses. The proposed method attained a remarkable diagnostic accuracy of up to 96%, exceeding models reliant exclusively on clinical data. This study highlights the capability of integrating artificial intelligence with medical imaging to create a highly accurate and non-invasive diagnostic instrument for cardiovascular disease.

To evaluate the cost-effectiveness of AI-assisted digital mammography (AI-DM) compared to conventional biennial breast cancer digital mammography screening (cDM) with double reading of screening mammograms, and to investigate the change in cost-effectiveness based on four different sub-strategies of AI-DM. A decision-analytic state-transition Markov model was used to analyse the decision of whether to use cDM or AI-DM in breast cancer screening. In this Markov model, one-year cycles were used, and the analysis was performed from a healthcare perspective with a lifetime horizon. In the model, we analysed 1000 hypothetical individuals attending mammography screenings assessed with AI-DM compared with 1000 hypothetical individuals assessed with cDM. The total costs, including both screening-related costs and breast cancer-related costs, were €3,468,967 and €3,528,288 for AI-DM and cDM, respectively. AI-DM resulted in a cost saving of €59,320 compared to cDM. Per 1000 individuals, AI-DM gained 10.8 quality-adjusted life years (QALYs) compared to cDM. Gained QALYs at a lower cost means that the AI-DM screening strategy was dominant compared to cDM. Break-even occurred at the second screening at age 42 years. This analysis showed that AI-assisted mammography for biennial breast cancer screening in a Swedish population of women aged 40-74 years is a cost-saving strategy compared to a conventional strategy using double human screen reading. Further clinical studies are needed, as scenario analyses showed that other strategies, more dependent on AI, are also cost-saving. Question To evaluate the cost-effectiveness of AI-DM in comparison to conventional biennial breast cDM screening. Findings AI-DM is cost-effective, and the break-even point occurred at the second screening at age 42 years. Clinical relevance The implementation of AI is clearly cost-effective as it reduces the total cost for the healthcare system and simultaneously results in a gain in QALYs.

Deep learning models often hallucinate, producing realistic artifacts that are not truly present in the sample. This can have dire consequences for scientific and medical inverse problems, such as MRI and microscopy denoising, where accuracy is more important than perceptual quality. Uncertainty quantification techniques, such as conformal prediction, can pinpoint outliers and provide guarantees for image regression tasks, improving reliability. However, existing methods utilize a linear constant scaling factor to calibrate uncertainty bounds, resulting in larger, less informative bounds. We propose QUTCC, a quantile uncertainty training and calibration technique that enables nonlinear, non-uniform scaling of quantile predictions to enable tighter uncertainty estimates. Using a U-Net architecture with a quantile embedding, QUTCC enables the prediction of the full conditional distribution of quantiles for the imaging task. During calibration, QUTCC generates uncertainty bounds by iteratively querying the network for upper and lower quantiles, progressively refining the bounds to obtain a tighter interval that captures the desired coverage. We evaluate our method on several denoising tasks as well as compressive MRI reconstruction. Our method successfully pinpoints hallucinations in image estimates and consistently achieves tighter uncertainty intervals than prior methods while maintaining the same statistical coverage.

With the widespread application of machine learning (ML) in the diagnosis and treatment of colorectal cancer (CRC), some studies have investigated the use of ML techniques for the diagnosis of KRAS (Kirsten rat sarcoma) mutation. Nevertheless, there is scarce evidence from evidence-based medicine to substantiate its efficacy. Our study was carried out to systematically review the performance of ML models developed using different modeling approaches, in diagnosing KRAS mutations in CRC. We aim to offer evidence-based foundations for the development and enhancement of future intelligent diagnostic tools. PubMed, Cochrane Library, Embase, and Web of Science were systematically retrieved, with the search cutoff date set to December 22, 2024. The encompassed studies are publicly published research papers that use ML to diagnose KRAS gene mutations in CRC. The risk of bias in the encompassed models was evaluated via the PROBAST (Prediction Model Risk of Bias Assessment Tool). A meta-analysis of the model's concordance index (c-index) was performed, and a bivariate mixed-effects model was used to summarize sensitivity and specificity based on diagnostic contingency tables. A total of 43 studies involving 10,888 patients were included. The modeling variables were derived from clinical characteristics, computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography/computed tomography, and pathological histology. In the validation cohort, for the ML model developed based on CT radiomic features, the c-index, sensitivity, and specificity were 0.87 (95% CI 0.84-0.90), 0.85 (95% CI 0.80-0.89), and 0.83 (95% CI 0.73-0.89), respectively. For the model developed using MRI radiomic features, the c-index, sensitivity, and specificity were 0.77 (95% CI 0.71-0.83), 0.78 (95% CI 0.72-0.83), and 0.73 (95% CI 0.63-0.81), respectively. For the ML model developed based on positron emission tomography/computed tomography radiomic features, the c-index, sensitivity, and specificity were 0.84 (95% CI 0.77-0.90), 0.73, and 0.83, respectively. Notably, the deep learning (DL) model based on pathological images demonstrated a c-index, sensitivity, and specificity of 0.96 (95% CI 0.94-0.98), 0.83 (95% CI 0.72-0.91), and 0.87 (95% CI 0.77-0.92), respectively. The DL model MRI-based model showed a c-index of 0.93 (95% CI 0.90-0.96), sensitivity of 0.85 (95% CI 0.75-0.91), and specificity of 0.83 (95% CI 0.77-0.88). ML is highly accurate in diagnosing KRAS mutations in CRC, and DL models based on MRI and pathological images exhibit particularly strong diagnosis accuracy. More broadly applicable DL-based diagnostic tools may be developed in the future. However, the clinical application of DL models remains relatively limited at present. Therefore, future research should focus on increasing sample sizes, improving model architectures, and developing more advanced DL models to facilitate the creation of highly efficient intelligent diagnostic tools for KRAS mutation diagnosis in CRC.