Automated radiology report generation (RRG) aims to produce detailed textual reports from clinical imaging, such as computed tomography (CT) scans, to improve the accuracy and efficiency of diagnosis and provision of management advice. RRG is complicated by two key challenges: (1) inherent complexity in extracting relevant information from imaging data under resource constraints, and (2) difficulty in objectively evaluating discrepancies between model-generated and expert-written reports. To address these challenges, we propose $\mu^2$LLM, a $\underline{\textbf{mu}}$ltiscale $\underline{\textbf{mu}}$ltimodal large language models for RRG tasks. The novel ${\mu}^2$Tokenizer, as an intermediate layer, integrates multi-modal features from the multiscale visual tokenizer and the text tokenizer, then enhances report generation quality through direct preference optimization (DPO), guided by GREEN-RedLlama. Experimental results on four large CT image-report medical datasets demonstrate that our method outperforms existing approaches, highlighting the potential of our fine-tuned $\mu^2$LLMs on limited data for RRG tasks. At the same time, for prompt engineering, we introduce a five-stage, LLM-driven pipeline that converts routine CT reports into paired visual-question-answer triples and citation-linked reasoning narratives, creating a scalable, high-quality supervisory corpus for explainable multimodal radiology LLM. All code, datasets, and models will be publicly available in our official repository. https://github.com/Siyou-Li/u2Tokenizer

Accurate interpretation of multi-view radiographs is crucial for diagnosing fractures, muscular injuries, and other anomalies. While significant advances have been made in AI-based analysis of single images, current methods often struggle to establish robust correspondences between different X-ray views, an essential capability for precise clinical evaluations. In this work, we present a novel self-supervised pipeline that eliminates the need for manual annotation by automatically generating a many-to-many correspondence matrix between synthetic X-ray views. This is achieved using digitally reconstructed radiographs (DRR), which are automatically derived from unannotated CT volumes. Our approach incorporates a transformer-based training phase to accurately predict correspondences across two or more X-ray views. Furthermore, we demonstrate that learning correspondences among synthetic X-ray views can be leveraged as a pretraining strategy to enhance automatic multi-view fracture detection on real data. Extensive evaluations on both synthetic and real X-ray datasets show that incorporating correspondences improves performance in multi-view fracture classification.

Current artificial intelligence models for medical imaging are predominantly single modality and single disease. Attempts to create multimodal and multi-disease models have resulted in inconsistent clinical accuracy. Furthermore, training these models typically requires large, labour-intensive, well-labelled datasets. We developed MerMED-FM, a state-of-the-art multimodal, multi-specialty foundation model trained using self-supervised learning and a memory module. MerMED-FM was trained on 3.3 million medical images from over ten specialties and seven modalities, including computed tomography (CT), chest X-rays (CXR), ultrasound (US), pathology patches, color fundus photography (CFP), optical coherence tomography (OCT) and dermatology images. MerMED-FM was evaluated across multiple diseases and compared against existing foundational models. Strong performance was achieved across all modalities, with AUROCs of 0.988 (OCT); 0.982 (pathology); 0.951 (US); 0.943 (CT); 0.931 (skin); 0.894 (CFP); 0.858 (CXR). MerMED-FM has the potential to be a highly adaptable, versatile, cross-specialty foundation model that enables robust medical imaging interpretation across diverse medical disciplines.

In the medical domain, acquiring large datasets is challenging due to both accessibility issues and stringent privacy regulations. Consequently, data availability and privacy protection are major obstacles to applying machine learning in medical imaging. To address this, our study proposes the Med-LSDM (Latent Semantic Diffusion Model), which operates directly in the 3D domain and leverages de-identified semantic maps to generate synthetic data as a method of privacy preservation and data augmentation. Unlike many existing methods that focus on generating 2D slices, Med-LSDM is designed specifically for 3D semantic image synthesis, making it well-suited for applications requiring full volumetric data. Med-LSDM incorporates a guiding mechanism that controls the 3D image generation process by applying a diffusion model within the latent space of a pre-trained VQ-GAN. By operating in the compressed latent space, the model significantly reduces computational complexity while still preserving critical 3D spatial details. Our approach demonstrates strong performance in 3D semantic medical image synthesis, achieving a 3D-FID score of 0.0054 on the conditional Duke Breast dataset and similar Dice scores (0.70964) to those of real images (0.71496). These results demonstrate that the synthetic data from our model have a small domain gap with real data and are useful for data augmentation.

Whole-heart coronary magnetic resonance angiography (CMRA) enables noninvasive and accurate detection of coronary artery stenosis. Nevertheless, the visual interpretation of CMRA is constrained by the observer's experience, necessitating substantial training. The purposes of this study were to develop a deep learning (DL) algorithm using a deep convolutional neural network to accurately detect significant coronary artery stenosis in CMRA and to investigate the effectiveness of this DL algorithm as a tool for assisting in accurate detection of coronary artery stenosis. Nine hundred and fifty-one coronary segments from 75 patients who underwent both CMRA and invasive coronary angiography (ICA) were studied. Significant stenosis was defined as a reduction in luminal diameter of >50% on quantitative ICA. A DL algorithm was proposed to classify CMRA segments into those with and without significant stenosis. A 4-fold cross-validation method was used to train and test the DL algorithm. An observer study was then conducted using 40 segments with stenosis and 40 segments without stenosis. Three radiology experts and 3 radiology trainees independently rated the likelihood of the presence of stenosis in each coronary segment with a continuous scale from 0 to 1, first without the support of the DL algorithm, then using the DL algorithm. Significant stenosis was observed in 84 (8.8%) of the 951 coronary segments. Using the DL algorithm trained by the 4-fold cross-validation method, the area under the receiver operating characteristic curve (AUC) for the detection of segments with significant coronary artery stenosis was 0.890, with 83.3% sensitivity, 83.6% specificity and 83.6% accuracy. In the observer study, the average AUC of trainees was significantly improved using the DL algorithm (0.898) compared to that without the algorithm (0.821, p<0.001). The average AUC of experts tended to be higher with the DL algorithm (0.897), but not significantly different from that without the algorithm (0.879, p=0.082). We developed a DL algorithm offering high diagnostic accuracy for detecting significant coronary artery stenosis on CMRA. Our proposed DL algorithm appears to be an effective tool for assisting inexperienced observers to accurately detect coronary artery stenosis in whole-heart CMRA.

Biomechanical analysis of the lumbar spine is vital for understanding load distribution and stress patterns under physiological conditions. Traditional finite element analysis (FEA) relies on time-consuming manual segmentation and meshing, leading to long runtimes and inconsistent accuracy. Automating this process improves efficiency and reproducibility. This study introduces an automated FEA methodology for lumbar spine biomechanics, integrating deep learning-based segmentation with computational modeling to streamline workflows from imaging to simulation. Medical imaging data were segmented using deep learning frameworks for vertebrae and intervertebral discs. Segmented structures were transformed into optimized surface meshes via Laplacian smoothing and decimation. Using the Gibbon library and FEBio, FEA models incorporated cortical and cancellous bone, nucleus, annulus, cartilage, and ligaments. Ligament attachments used spherical coordinate-based segmentation; vertebral endplates were extracted via principal component analysis (PCA) for cartilage modeling. Simulations assessed stress, strain, and displacement under axial rotation, extension, flexion, and lateral bending. The automated pipeline cut model preparation time by 97.9%, from over 24 hours to 30 minutes and 49.48 seconds. Biomechanical responses aligned with experimental and traditional FEA data, showing high posterior element loads in extension and flexion, consistent ligament forces, and disc deformations. The approach enhanced reproducibility with minimal manual input. This automated methodology provides an efficient, accurate framework for lumbar spine biomechanics, eliminating manual segmentation challenges. It supports clinical diagnostics, implant design, and rehabilitation, advancing computational and patient-specific spinal studies. Rapid simulations enhance implant optimization, and early detection of degenerative spinal issues, improving personalized treatment and research.

Pneumonia is a respiratory condition characterized by inflammation of the alveolar sacs in the lungs, which disrupts normal oxygen exchange. This disease disproportionately impacts vulnerable populations, including young children (under five years of age) and elderly individuals (over 65 years), primarily due to their compromised immune systems. The mortality rate associated with pneumonia remains alarmingly high, particularly in low-resource settings where healthcare access is limited. Although effective prevention strategies exist, pneumonia continues to claim the lives of approximately one million children each year, earning its reputation as a "silent killer." Globally, an estimated 500 million cases are documented annually, underscoring its widespread public health burden. This study explores the design and evaluation of the CNN-based Computer-Aided Diagnostic (CAD) systems with an aim of carrying out competent as well as resourceful classification and categorization of chest radiographs into binary classes (Normal, Pneumonia). An augmented Kaggle dataset of 18,200 chest radiographs, split between normal and pneumonia cases, was utilized. This study conducts a series of experiments to evaluate lightweight CNN models--ShuffleNet, NASNet-Mobile, and EfficientNet-b0--using transfer learning that achieved accuracy of 90%, 88% and 89%, prompting the task for deep feature extraction from each of the networks and applying feature fusion to further pair it with SVM classifier and XGBoost classifier, achieving an accuracy of 97% and 98% resepectively. The proposed research emphasizes the crucial role of CAD systems in advancing radiological diagnostics, delivering effective solutions to aid radiologists in distinguishing between diagnoses by applying feature fusion, feature selection along with various machine learning algorithms and deep learning architectures.

Lung cancer remains one of the leading causes of cancer mortality, and while low dose CT screening improves mortality, radiological detection is challenging due to the increasing shortage of radiologists. Artificial intelligence can significantly improve the procedure and also decrease the overall workload of the entire healthcare department. Building upon the existing works of application of genetic algorithm this study aims to create a novel algorithm for lung cancer diagnosis with utmost precision. We included a total of 156 CT scans of patients divided into two databases, followed by feature extraction using image statistics, histograms, and 2D transforms (FFT, DCT, WHT). Optimal feature vectors were formed and organized into Excel based knowledge-bases. Genetically trained classifiers like MLP, GFF-NN, MNN and SVM, are then optimized, with experimentations with different combinations of parameters, activation functions, and data partitioning percentages. Evaluation metrics included classification accuracy, Mean Squared Error (MSE), Area under Receiver Operating Characteristics (ROC) curve, and computational efficiency. Computer simulations demonstrated that the MNN (Topology II) classifier, specifically when trained with FFT coefficients and a momentum learning rule, consistently achieved 100% average classification accuracy on the cross-validation dataset for both Data-base I and Data-base II, outperforming MLP-based classifiers. This genetically optimized and trained MNN (Topology II) classifier is therefore recommended as the optimal solution for lung cancer diagnosis from CT scan images.

Accurately and preoperatively predicting tumor response to transarterial chemoembolization (TACE) treatment is crucial for individualized treatment decision-making hepatocellular carcinoma (HCC). In this study, we propose a novel feature fusion network based on the Convolutional Neural Networks Meet Vision Transformers (CMT) architecture, termed CMT-FFNet, to predict TACE efficacy using preoperative multiphase Magnetic Resonance Imaging (MRI) scans. The CMT-FFNet combines local feature extraction with global dependency modeling through attention mechanisms, enabling the extraction of complementary information from multiphase MRI data. Additionally, we introduce an orthogonality loss to optimize the fusion of imaging and clinical features, further enhancing the complementarity of cross-modal features. Moreover, visualization techniques were employed to highlight key regions contributing to model decisions. Extensive experiments were conducted to evaluate the effectiveness of the proposed modules and network architecture. Experimental results demonstrate that our model effectively captures latent correlations among features extracted from multiphase MRI data and multimodal inputs, significantly improving the prediction performance of TACE treatment response in HCC patients.

BackgroundPositron emission tomography (PET)/CT for myocardial perfusion imaging (MPI) provides multiple imaging biomarkers, often evaluated separately. We developed an artificial intelligence (AI) model integrating key clinical PET MPI parameters to improve the diagnosis of obstructive coronary artery disease (CAD). MethodsFrom 17,348 patients undergoing cardiac PET/CT across four sites, we retrospectively enrolled 1,664 subjects who had invasive coronary angiography within 180 days and no prior CAD. Deep learning was used to derive coronary artery calcium score (CAC) from CT attenuation correction maps. XGBoost machine learning model was developed using data from one site to detect CAD, defined as left main stenosis [&ge;]50% or [&ge;]70% in other arteries. The model utilized 10 image-derived parameters from clinical practice: CAC, stress/rest left ventricle ejection fraction, stress myocardial blood flow (MBF), myocardial flow reserve (MFR), ischemic and stress total perfusion deficit (TPD), transient ischemic dilation ratio, rate pressure product, and sex. Generalizability was evaluated in the remaining three sites--chosen to maximize testing power and capture inter-site variability--and model performance was compared with quantitative analyses using the area under the receiver operating characteristic curve (AUC). Patient-specific predictions were explained using shapley additive explanations. ResultsThere was a 61% and 53% CAD prevalence in the training (n=386) and external testing (n=1,278) set, respectively. In the external evaluation, the AI model achieved a higher AUC (0.83 [95% confidence interval (CI): 0.81-0.85]) compared to clinical score by experienced physicians (0.80 [0.77-0.82], p=0.02), ischemic TPD (0.79 [0.77-0.82], p<0.001), MFR (0.75 [0.72-0.78], p<0.001), and CAC (0.69 [0.66-0.72], p<0.001). The models performances were consistent in sex, body mass index, and age groups. The top features driving the prediction were stress/ischemic TPD, CAC, and MFR. ConclusionAI integrating perfusion, flow, and CAC scoring improves PET MPI diagnostic accuracy, offering automated and interpretable predictions for CAD diagnosis.