Reducing radiation dose from PET imaging is essential to minimize cancer risks; however, it often leads to increased noise and degraded image quality, compromising diagnostic reliability. Recent advances in deep learning have shown promising results in addressing these limitations through effective denoising. However, existing networks trained on specific noise levels often fail to generalize across diverse acquisition conditions. Moreover, training multiple models for different noise levels is impractical due to data and computational constraints. This study aimed to develop a supervised Swin Transformer-based unified noise-aware (ST-UNN) network that handles diverse noise levels and reconstructs high-quality images in low-dose PET imaging. We present a Swin Transformer-based Noise-Aware Network (ST-UNN), which incorporates multiple sub-networks, each designed to address specific noise levels ranging from 1 % to 10 %. An adaptive weighting mechanism dynamically integrates the outputs of these sub-networks to achieve effective denoising. The model was trained and evaluated using PET/CT dataset encompassing the entire head and malignant lesions in the head and neck region. Performance was assessed using a combination of structural and statistical metrics, including the Structural Similarity Index (SSIM), Peak Signal-to-Noise Ratio (PSNR), Standardized Uptake Value (SUV) mean bias, SUV_max bias, and Root Mean Square Error (RMSE). This comprehensive evaluation ensured reliable results for both global and localized regions within PET images. The ST-UNN consistently outperformed conventional networks, particularly in ultra-low-dose scenarios. At 1 % count level, it achieved a PSNR of 34.77, RMSE of 0.05, and SSIM of 0.97, notably surpassing the baseline networks. It also achieved the lowest SUV_mean bias (0.08) and RMSE lesion (0.12) at this level. Across all count levels, ST-UNN maintained high performance and low error, demonstrating strong generalization and diagnostic integrity. ST-UNN offers a scalable, transformer-based solution for low-dose PET imaging. By dynamically integrating sub-networks, it effectively addresses noise variability and provides superior image quality, thereby advancing the capabilities of low-dose and dynamic PET imaging.

Manually analyzing a series of MRI images to obtain information about the heart's motion is a time-consuming and labor-intensive task. Recently, many AI-driven tools have been used to automatically analyze cardiac MRI. However, it is still unknown whether the results generated by these tools are consistent. The aim of the present study was to investigate the agreement of AI-powered automated tools for measuring cine MRI-derived cardiac function and motion indices. Cine MRI datasets of 23 healthy volunteers (10 males, 32.7 ± 11.3 years) were processed using heart deformation analysis (HDA, Trufistrain) and Circle CVI 42. The left and right ventricular (LV/RV) end-diastolic volume (LVEDV and RVEDV), end-systolic volume (LVESV and RVESV), stroke volume (LVSV and RVSV), cardiac output (LVCO and RVCO), ejection fraction (LVEF and RVEF), LV mass (LVM), LV global strain, strain rate, displacement, and velocity were calculated without interventions. Agreements and discrepancies of indices acquired with the two tools were evaluated from various aspects using t-tests, Pearson correlation coefficient (r), interclass correlation coefficient (ICC), and coefficient of variation (CoV). Systematic biases for measuring cardiac function and motion indices were observed. In global cardiac function indices, LVEF (56.9% ± 6.4 vs. 57.8% ± 5.7, p = 0.433, r = 0.609, ICC = 0.757, CoV = 6.7%) and LVM (82.7 g ± 21.6 vs. 82.6 g ± 18.7, p = 0.988, r = 0.923, ICC = 0.956, CoV = 11.7%) acquired with HDA and Circle seemed to be exchangeable. Among cardiac motion indices, circumferential strain rate demonstrated good agreements between two tools (97 ± 14.6 vs. 97.8 ± 13.6, p = 0.598, r = 0.89, ICC = 0.943, CoV = 5.1%). Cine MRI-derived cardiac function and motion indices obtained using different AI-powered image processing tools are related but may also differ. Such variations should be considered when evaluating results sourced from different studies.

Brain lesions are abnormalities or injuries in brain tissue that are often detectable using magnetic resonance imaging (MRI), which reveals structural changes in the affected areas. This broad definition of brain lesions includes areas of the brain that are irreversibly damaged, as well as areas of brain tissue that are deformed as a result of lesion growth or swelling. Despite the importance of differentiating between damaged and deformed tissue, existing lesion segmentation methods overlook this distinction, labeling both of them as a single anomaly. In this work, we introduce a diffusion model-based framework for analyzing and reversing the brain lesion process. Our pipeline first segments abnormal regions in the brain, then estimates and reverses tissue deformations by restoring displaced tissue to its original position, isolating the core lesion area representing the initial damage. Finally, we inpaint the core lesion area to arrive at an estimation of the pre-lesion healthy brain. This proposed framework reverses a forward lesion growth process model that is well-established in biomechanical studies that model brain lesions. Our results demonstrate improved accuracy in lesion segmentation, characterization, and brain labeling compared to traditional methods, offering a robust tool for clinical and research applications in brain lesion analysis. Since pre-lesion healthy versions of abnormal brains are not available in any public dataset for validation of the reverse process, we simulate a forward model to synthesize multiple lesioned brain images.

Manual annotation of medical images is a labor-intensive and time-consuming process, posing a significant bottleneck in the development and deployment of robust medical imaging AI systems. This paper introduces a novel Human-AI collaborative framework for medical image segmentation that substantially reduces the annotation burden by eliminating the need for explicit manual pixel-level labeling. The core innovation lies in a preference learning paradigm, where human experts provide minimal, intuitive feedback -- simply indicating whether an AI-generated segmentation is better or worse than a previous version. The framework comprises four key components: (1) an adaptable foundation model (FM) for feature extraction, (2) label propagation based on feature similarity, (3) a clicking agent that learns from human better-or-worse feedback to decide where to click and with which label, and (4) a multi-round segmentation learning procedure that trains a state-of-the-art segmentation network using pseudo-labels generated by the clicking agent and FM-based label propagation. Experiments on three public datasets demonstrate that the proposed approach achieves competitive segmentation performance using only binary preference feedback, without requiring experts to directly manually annotate the images.

Accurate prediction of stroke risk at an early stage is essential for timely intervention and prevention, especially given the serious health consequences and economic burden that strokes can cause. In this study, we proposed a class-balanced and data-augmented (CBDA-ResNet50) deep learning model to improve the prediction accuracy of the well-known ResNet50 architecture for stroke risk. Our approach uses advanced techniques such as class balancing and data augmentation to address common challenges in medical imaging datasets, such as class imbalance and limited training examples. In most cases, these problems lead to biased or less reliable predictions. To address these issues, the proposed model assures that the predictions are still accurate even when some stroke risk factors are absent in the data. The performance of CBDA-ResNet50 improves by using the Adam optimizer and the ReduceLROnPlateau scheduler to adjust the learning rate. The application of weighted cross entropy removes the imbalance between classes and significantly improves the results. It achieves an accuracy of 97.87% and a balanced accuracy of 98.27%, better than many of the previous best models. This shows that we can make more reliable predictions by combining modern deep-learning models with advanced data-processing techniques. CBDA-ResNet50 has the potential to be a model for early stroke prevention, aiming to improve patient outcomes and reduce healthcare costs.

Artificial intelligence (AI) is increasingly used in radiology, but its development in pediatric imaging remains limited, particularly for emergent conditions. Ileocolic intussusception is an important cause of acute abdominal pain in infants and toddlers and requires timely diagnosis to prevent complications such as bowel ischemia or perforation. While ultrasonography is the diagnostic standard due to its high sensitivity and specificity, its accessibility may be limited, especially outside tertiary centers. Abdominal radiographs (AXRs), despite their limited sensitivity, are often the first-line imaging modality in clinical practice. In this context, AI could support early screening and triage by analyzing AXRs and identifying patients who require further ultrasonography evaluation. This study aimed to upgrade and externally validate an AI model for screening ileocolic intussusception using pediatric AXRs with multicenter data and to assess the diagnostic performance of the model in comparison with radiologists of varying experience levels with and without AI assistance. This retrospective study included pediatric patients (≤5 years) who underwent both AXRs and ultrasonography for suspected intussusception. Based on the preliminary study from hospital A, the AI model was retrained using data from hospital B and validated with external datasets from hospitals C and D. Diagnostic performance of the upgraded AI model was evaluated using sensitivity, specificity, and the area under the receiver operating characteristic curve (AUC). A reader study was conducted with 3 radiologists, including 2 trainees and 1 pediatric radiologist, to evaluate diagnostic performance with and without AI assistance. Based on the previously developed AI model trained on 746 patients from hospital A, an additional 431 patients from hospital B (including 143 intussusception cases) were used for further training to develop an upgraded AI model. External validation was conducted using data from hospital C (n=68; 19 intussusception cases) and hospital D (n=90; 30 intussusception cases). The upgraded AI model achieved a sensitivity of 81.7% (95% CI 68.6%-90%) and a specificity of 81.7% (95% CI 73.3%-87.8%), with an AUC of 86.2% (95% CI 79.2%-92.1%) in the external validation set. Without AI assistance, radiologists showed lower performance (overall AUC 64%; sensitivity 49.7%; specificity 77.1%). With AI assistance, radiologists' specificity improved to 93% (difference +15.9%; P<.001), and AUC increased to 79.2% (difference +15.2%; P=.05). The least experienced reader showed the largest improvement in specificity (+37.6%; P<.001) and AUC (+14.7%; P=.08). The upgraded AI model improved diagnostic performance for screening ileocolic intussusception on pediatric AXRs. It effectively enhanced the specificity and overall accuracy of radiologists, particularly those with less experience in pediatric radiology. A user-friendly software platform was introduced to support broader clinical validation and underscores the potential of AI as a screening and triage tool in pediatric emergency settings.

The deployment of AI models in clinical practice faces a critical challenge: models achieving expert-level performance on benchmarks can fail catastrophically when confronted with real-world variations in medical imaging. Minor shifts in scanner hardware, lighting or demographics can erode accuracy, but currently reliability auditing to identify such catastrophic failure cases before deployment is a bespoke and time-consuming process. Practitioners lack accessible and interpretable tools to expose and repair hidden failure modes. Here we introduce ModelAuditor, a self-reflective agent that converses with users, selects task-specific metrics, and simulates context-dependent, clinically relevant distribution shifts. ModelAuditor then generates interpretable reports explaining how much performance likely degrades during deployment, discussing specific likely failure modes and identifying root causes and mitigation strategies. Our comprehensive evaluation across three real-world clinical scenarios - inter-institutional variation in histopathology, demographic shifts in dermatology, and equipment heterogeneity in chest radiography - demonstrates that ModelAuditor is able correctly identify context-specific failure modes of state-of-the-art models such as the established SIIM-ISIC melanoma classifier. Its targeted recommendations recover 15-25% of performance lost under real-world distribution shift, substantially outperforming both baseline models and state-of-the-art augmentation methods. These improvements are achieved through a multi-agent architecture and execute on consumer hardware in under 10 minutes, costing less than US$0.50 per audit.

Aims: To develop a deep-learning (DL) framework that will allow fully automated longitudinal and circumferential co-registration of intravascular ultrasound (IVUS) and optical coherence tomography (OCT) images. Methods and results: Data from 230 patients (714 vessels) with acute coronary syndrome that underwent near-infrared spectroscopy (NIRS)-IVUS and OCT imaging in their non-culprit vessels were included in the present analysis. The lumen borders annotated by expert analysts in 61,655 NIRS-IVUS and 62,334 OCT frames, and the side branches and calcific tissue identified in 10,000 NIRS-IVUS frames and 10,000 OCT frames, were used to train DL solutions for the automated extraction of these features. The trained DL solutions were used to process NIRS-IVUS and OCT images and their output was used by a dynamic time warping algorithm to co-register longitudinally the NIRS-IVUS and OCT images, while the circumferential registration of the IVUS and OCT was optimized through dynamic programming. On a test set of 77 vessels from 22 patients, the DL method showed high concordance with the expert analysts for the longitudinal and circumferential co-registration of the two imaging sets (concordance correlation coefficient >0.99 for the longitudinal and >0.90 for the circumferential co-registration). The Williams Index was 0.96 for longitudinal and 0.97 for circumferential co-registration, indicating a comparable performance to the analysts. The time needed for the DL pipeline to process imaging data from a vessel was <90s. Conclusion: The fully automated, DL-based framework introduced in this study for the co-registration of IVUS and OCT is fast and provides estimations that compare favorably to the expert analysts. These features renders it useful in research in the analysis of large-scale data collected in studies that incorporate multimodality imaging to characterize plaque composition.

Low-dose computed tomography (LDCT) reduces radiation exposure but often degrades image quality, potentially compromising diagnostic accuracy. Existing deep learning-based denoising methods focus primarily on pixel-level mappings, overlooking the potential benefits of high-level semantic guidance. Recent advances in vision-language models (VLMs) suggest that language can serve as a powerful tool for capturing structured semantic information, offering new opportunities to improve LDCT reconstruction. In this paper, we introduce LangMamba, a Language-driven Mamba framework for LDCT denoising that leverages VLM-derived representations to enhance supervision from normal-dose CT (NDCT). LangMamba follows a two-stage learning strategy. First, we pre-train a Language-guided AutoEncoder (LangAE) that leverages frozen VLMs to map NDCT images into a semantic space enriched with anatomical information. Second, we synergize LangAE with two key components to guide LDCT denoising: Semantic-Enhanced Efficient Denoiser (SEED), which enhances NDCT-relevant local semantic while capturing global features with efficient Mamba mechanism, and Language-engaged Dual-space Alignment (LangDA) Loss, which ensures that denoised images align with NDCT in both perceptual and semantic spaces. Extensive experiments on two public datasets demonstrate that LangMamba outperforms conventional state-of-the-art methods, significantly improving detail preservation and visual fidelity. Remarkably, LangAE exhibits strong generalizability to unseen datasets, thereby reducing training costs. Furthermore, LangDA loss improves explainability by integrating language-guided insights into image reconstruction and offers a plug-and-play fashion. Our findings shed new light on the potential of language as a supervisory signal to advance LDCT denoising. The code is publicly available on https://github.com/hao1635/LangMamba.

Objective: Latent diffusion models (LDMs) could mitigate data scarcity challenges affecting machine learning development for medical image interpretation. The recent CCELLA LDM improved prostate cancer detection performance using synthetic MRI for classifier training but was limited to the axial T2-weighted (AxT2) sequence, did not investigate inter-institutional domain shift, and prioritized radiology over histopathology outcomes. We propose CCELLA++ to address these limitations and improve clinical utility. Methods: CCELLA++ expands CCELLA for simultaneous biparametric prostate MRI (bpMRI) generation, including the AxT2, high b-value diffusion series (HighB) and apparent diffusion coefficient map (ADC). Domain adaptation was investigated by pretraining classifiers on real or LDM-generated synthetic data from an internal institution, followed with fine-tuning on progressively smaller fractions of an out-of-distribution, external dataset. Results: CCELLA++ improved 3D FID for HighB and ADC but not AxT2 (0.013, 0.012, 0.063 respectively) sequences compared to CCELLA (0.060). Classifier pretraining with CCELLA++ bpMRI outperformed real bpMRI in AP and AUC for all domain adaptation scenarios. CCELLA++ pretraining achieved highest classifier performance below 50% (n=665) external dataset volume. Conclusion: Synthetic bpMRI generated by our method can improve downstream classifier generalization and performance beyond real bpMRI or CCELLA-generated AxT2-only images. Future work should seek to quantify medical image sample quality, balance multi-sequence LDM training, and condition the LDM with additional information. Significance: The proposed CCELLA++ LDM can generate synthetic bpMRI that outperforms real data for domain adaptation with a limited target institution dataset. Our code is available at https://github.com/grabkeem/CCELLA-plus-plus