We propose a cascaded 3D diffusion model framework to synthesize high-fidelity 3D PET/CT volumes directly from demographic variables, addressing the growing need for realistic digital twins in oncologic imaging, virtual trials, and AI-driven data augmentation. Unlike deterministic phantoms, which rely on predefined anatomical and metabolic templates, our method employs a two-stage generative process. An initial score-based diffusion model synthesizes low-resolution PET/CT volumes from demographic variables alone, providing global anatomical structures and approximate metabolic activity. This is followed by a super-resolution residual diffusion model that refines spatial resolution. Our framework was trained on 18-F FDG PET/CT scans from the AutoPET dataset and evaluated using organ-wise volume and standardized uptake value (SUV) distributions, comparing synthetic and real data between demographic subgroups. The organ-wise comparison demonstrated strong concordance between synthetic and real images. In particular, most deviations in metabolic uptake values remained within 3-5% of the ground truth in subgroup analysis. These findings highlight the potential of cascaded 3D diffusion models to generate anatomically and metabolically accurate PET/CT images, offering a robust alternative to traditional phantoms and enabling scalable, population-informed synthetic imaging for clinical and research applications.

Limited DXA access hinders osteoporosis screening. This proof-of-concept study proposes using widely available knee X-rays for opportunistic Bone Mineral Density (BMD) estimation via deep learning, emphasizing robust uncertainty quantification essential for clinical use. An EfficientNet model was trained on the OAI dataset to predict BMD from bilateral knee radiographs. Two Test-Time Augmentation (TTA) methods were compared: traditional averaging and a multi-sample approach. Crucially, Split Conformal Prediction was implemented to provide statistically rigorous, patient-specific prediction intervals with guaranteed coverage. Results showed a Pearson correlation of 0.68 (traditional TTA). While traditional TTA yielded better point predictions, the multi-sample approach produced slightly tighter confidence intervals (90%, 95%, 99%) while maintaining coverage. The framework appropriately expressed higher uncertainty for challenging cases. Although anatomical mismatch between knee X-rays and standard DXA limits immediate clinical use, this method establishes a foundation for trustworthy AI-assisted BMD screening using routine radiographs, potentially improving early osteoporosis detection.

Lung cancer is the leading cause of cancer mortality worldwide, and non-invasive methods for detecting key mutations and staging are essential for improving patient outcomes. Here, we compare the performance of two machine learning models - FMCIB+XGBoost, a supervised model with domain-specific pretraining, and Dinov2+ABMIL, a self-supervised model with attention-based multiple-instance learning - on 3D lung nodule data from the Stanford Radiogenomics and Lung-CT-PT-Dx cohorts. In the task of KRAS and EGFR mutation detection, FMCIB+XGBoost consistently outperformed Dinov2+ABMIL, achieving accuracies of 0.846 and 0.883 for KRAS and EGFR mutations, respectively. In cancer staging, Dinov2+ABMIL demonstrated competitive generalization, achieving an accuracy of 0.797 for T-stage prediction in the Lung-CT-PT-Dx cohort, suggesting SSL's adaptability across diverse datasets. Our results emphasize the clinical utility of supervised models in mutation detection and highlight the potential of SSL to improve staging generalization, while identifying areas for enhancement in mutation sensitivity.

In this work, we investigate the performance across multiple classification models to classify chest X-ray images into four categories of COVID-19, pneumonia, tuberculosis (TB), and normal cases. We leveraged transfer learning techniques with state-of-the-art pre-trained Convolutional Neural Networks (CNNs) models. We fine-tuned these pre-trained architectures on a labeled medical x-ray images. The initial results are promising with high accuracy and strong performance in key classification metrics such as precision, recall, and F1 score. We applied Gradient-weighted Class Activation Mapping (Grad-CAM) for model interpretability to provide visual explanations for classification decisions, improving trust and transparency in clinical applications.

Occupational health assessment is critical for detecting respiratory issues caused by harmful exposures, such as cement dust. Quantitative computed tomography (QCT) imaging provides detailed insights into lung structure and function, enhancing the diagnosis of lung diseases. However, its high dimensionality poses challenges for traditional machine learning methods. In this study, Kolmogorov-Arnold networks (KANs) were used for the binary classification of QCT imaging data to assess respiratory conditions associated with cement dust exposure. The dataset comprised QCT images from 609 individuals, including 311 subjects exposed to cement dust and 298 healthy controls. We derived 141 QCT-based variables and employed KANs with two hidden layers of 15 and 8 neurons. The network parameters, including grid intervals, polynomial order, learning rate, and penalty strengths, were carefully fine-tuned. The performance of the model was assessed through various metrics, including accuracy, precision, recall, F1 score, specificity, and the Matthews Correlation Coefficient (MCC). A five-fold cross-validation was employed to enhance the robustness of the evaluation. SHAP analysis was applied to interpret the sensitive QCT features. The KAN model demonstrated consistently high performance across all metrics, with an average accuracy of 98.03 %, precision of 97.35 %, recall of 98.70 %, F1 score of 98.01 %, and specificity of 97.40 %. The MCC value further confirmed the robustness of the model in managing imbalanced datasets. The comparative analysis demonstrated that the KAN model outperformed traditional methods and other deep learning approaches, such as TabPFN, ANN, FT-Transformer, VGG19, MobileNets, ResNet101, XGBoost, SVM, random forest, and decision tree. SHAP analysis highlighted structural and functional lung features, such as airway geometry, wall thickness, and lung volume, as key predictors. KANs significantly improved the classification of QCT imaging data, enhancing early detection of cement dust-induced respiratory conditions. SHAP analysis supported model interpretability, enhancing its potential for clinical translation in occupational health assessments.

Planning an osteotomy or arthroplasty surgery on a lower limb requires prior classification/identification of its deformities. The detection of skeletal landmarks and the calculation of angles required to identify the deformities are traditionally done manually, with measurement accuracy relying considerably on the experience of the individual doing the measurements. We propose a novel, image pyramid-based approach to skeletal landmark detection. The proposed approach uses a Convolutional Neural Network (CNN) that receives the raw X-ray image as input and produces the coordinates of the landmarks. The landmark estimations are modified iteratively via the error feedback method to come closer to the target. Our clinically produced full-leg X-Rays dataset is made publically available and used to train and test the network. Angular quantities are calculated based on detected landmarks. Angles are then classified as lower than normal, normal or higher than normal according to predefined ranges for a normal condition. The performance of our approach is evaluated at several levels: landmark coordinates accuracy, angles' measurement accuracy, and classification accuracy. The average absolute error (difference between automatically and manually determined coordinates) for landmarks was 0.79 ± 0.57 mm on test data, and the average absolute error (difference between automatically and manually calculated angles) for angles was 0.45 ± 0.42°. Results from multiple case studies involving high-resolution images show that the proposed approach outperforms previous deep learning-based approaches in terms of accuracy and computational cost. It also enables the automatic detection of the lower limb misalignments in full-leg x-ray images.

Ultrasound imaging is pivotal in clinical diagnostics due to its affordability, portability, safety, real-time capability, and non-invasive nature. It is widely utilized for examining various organs, such as the breast, thyroid, ovary, cardiac, and more. However, the manual interpretation and annotation of ultrasound images are time-consuming and prone to variability among physicians. While single-task artificial intelligence (AI) solutions have been explored, they are not ideal for scaling AI applications in medical imaging. Foundation models, although a trending solution, often struggle with real-world medical datasets due to factors such as noise, variability, and the incapability of flexibly aligning prior knowledge with task adaptation. To address these limitations, we propose an orchestration learning framework named PerceptGuide for general-purpose ultrasound classification and segmentation. Our framework incorporates a novel orchestration mechanism based on prompted hyper-perception, which adapts to the diverse inductive biases required by different ultrasound datasets. Unlike self-supervised pre-trained models, which require extensive fine-tuning, our approach leverages supervised pre-training to directly capture task-relevant features, providing a stronger foundation for multi-task and multi-organ ultrasound imaging. To support this research, we compiled a large-scale Multi-task, Multi-organ public ultrasound dataset (M²-US), featuring images from 9 organs and 16 datasets, encompassing both classification and segmentation tasks. Our approach employs four specific prompts-Object, Task, Input, and Position-to guide the model, ensuring task-specific adaptability. Additionally, a downstream synchronization training stage is introduced to fine-tune the model for new data, significantly improving generalization capabilities and enabling real-world applications. Experimental results demonstrate the robustness and versatility of our framework in handling multi-task and multi-organ ultrasound image processing, outperforming both specialist models and existing general AI solutions. Compared to specialist models, our method improves segmentation from 82.26% to 86.45%, classification from 71.30% to 79.08%, while also significantly reducing model parameters.

Machine learning, particularly through advanced imaging techniques such as three-dimensional Magnetic Resonance Imaging (MRI), has significantly improved medical diagnostics. This is especially critical for diagnosing complex conditions like Alzheimer's disease. Our study introduces Triamese-ViT, an innovative Tri-structure of Vision Transformers (ViTs) that incorporates a built-in interpretability function, it has structure-aware explainability that allows for the identification and visualization of key features or regions contributing to the prediction, integrates information from three perspectives to enhance brain age estimation. This method not only increases accuracy but also improves interoperability with existing techniques. When evaluated, Triamese-ViT demonstrated superior performance and produced insightful attention maps. We applied these attention maps to the analysis of natural aging and the diagnosis of Autism Spectrum Disorder (ASD). The results aligned with those from occlusion analysis, identifying the Cingulum, Rolandic Operculum, Thalamus, and Vermis as important regions in normal aging, and highlighting the Thalamus and Caudate Nucleus as key regions for ASD diagnosis.

Traditional computed tomography (CT) methods for 3D reconstruction face resolution limitations and require time-consuming post-processing workflows. While deep learning techniques improve the accuracy of segmentation, traditional voxel-based segmentation and surface reconstruction pipelines tend to introduce artifacts such as disconnected regions, topological inconsistencies, and stepped distortions. To overcome these challenges, we propose ToPoMesh, an end-to-end 3D mesh reconstruction deep learning framework for direct reconstruction of high-fidelity surface meshes from CT volume data. To address the existing problems, our approach introduces three core innovations: (1) accurate local and global shape modeling by preserving and enhancing local feature information through residual connectivity and self-attention mechanisms in graph convolutional networks; (2) an adaptive variant density (Avd) mesh de-pooling strategy, which dynamically optimizes the vertex distribution; (3) a topology modification module that iteratively prunes the error surfaces and boundary smoothing via variable regularity terms to obtain finer mesh surfaces. Experiments on the LiTS, MSD pancreas tumor, MSD hippocampus, and MSD spleen datasets demonstrate that ToPoMesh outperforms state-of-the-art methods. Quantitative evaluations demonstrate a 57.4% reduction in Chamfer distance (liver) and a 0.47% improvement in F-score compared to end-to-end 3D reconstruction methods, while qualitative results confirm enhanced fidelity for thin structures and complex anatomical topologies versus segmentation frameworks. Importantly, our method eliminates the need for manual post-processing, realizes the ability to reconstruct 3D meshes from images, and can provide precise guidance for surgical planning and diagnosis.

In deep multi-instance learning, the number of applicable instances depends on the data set. In histopathology images, deep learning multi-instance learners usually assume there are hundreds to thousands instances in a bag. However, when the number of instances in a bag increases to 256 in brain hematoma CT, learning becomes extremely difficult. In this paper, we address this drawback. To overcome this problem, we propose using a pre-trained model with self-supervised learning for the multi-instance learner as a downstream task. With this method, even when the original target task suffers from the spurious correlation problem, we show improvements of 5% to 13% in accuracy and 40% to 55% in the F1 measure for the hypodensity marker classification of brain hematoma CT.