Medical image analysis plays a critical role in brain tumor detection, but training deep learning models often requires large, labeled datasets, which can be time-consuming and costly. This study explores a comparative analysis of machine learning and deep learning models for brain tumor classification, focusing on whether deep learning models are necessary for small medical datasets and whether self-supervised learning can reduce annotation costs. The primary goal is to evaluate trade-offs between traditional machine learning and deep learning, including self-supervised models under small medical image data. The secondary goal is to assess model robustness, transferability, and generalization through evaluation of unseen data within- and cross-domains. Four models were compared: (1) support vector machine (SVM) with histogram of oriented gradients (HOG) features, (2) a convolutional neural network based on ResNet18, (3) a transformer-based model using vision transformer (ViT-B/16), and (4) a self-supervised learning approach using Simple Contrastive Learning of Visual Representations (SimCLR). These models were selected to represent diverse paradigms. SVM+HOG represents traditional feature engineering with low computational cost, ResNet18 serves as a well-established convolutional neural network with strong baseline performance, ViT-B/16 leverages self-attention to capture long-range spatial features, and SimCLR enables learning from unlabeled data, potentially reducing annotation costs. The primary dataset consisted of 2870 brain magnetic resonance images across 4 classes: glioma, meningioma, pituitary, and nontumor. All models were trained under consistent settings, including data augmentation, early stopping, and 3 independent runs using the different random seeds to account for performance variability. Performance metrics included accuracy, precision, recall, F₁-score, and convergence. To assess robustness and generalization capability, evaluation was performed on unseen test data from both the primary and cross datasets. No retraining or test augmentations were applied to the external data, thereby reflecting realistic deployment conditions. The models demonstrated consistently strong performance in both within-domain and cross-domain evaluations. The results revealed distinct trade-offs; ResNet18 achieved the highest validation accuracy (mean 99.77%, SD 0.00%) and the lowest validation loss, along with a weighted test accuracy of 99% within-domain and 95% cross-domain. SimCLR reached a mean validation accuracy of 97.29% (SD 0.86%) and achieved up to 97% weighted test accuracy within-domain and 91% cross-domain, despite requiring 2-stage training phases involving contrastive pretraining followed by linear evaluation. ViT-B/16 reached a mean validation accuracy of 97.36% (SD 0.11%), with a weighted test accuracy of 98% within-domain and 93% cross-domain. SVM+HOG maintained a competitive validation accuracy of 96.51%, with 97% within-domain test accuracy, though its accuracy dropped to 80% cross-domain. The study reveals meaningful trade-offs between model complexity, annotation requirements, and deployment feasibility-critical factors for selecting models in real-world medical imaging applications.

Stress perfusion cardiovascular magnetic resonance (CMR) is widely used to detect myocardial ischemia, mostly through visual assessment. Recent studies suggest that strain imaging at rest and during stress can also help in prognostic stratification. However, the additional prognostic value of combining both rest and stress strain imaging has not been fully established. This study examined the incremental benefit of combining these strain measures with traditional risk prognosticators and CMR findings to predict major adverse clinical events (MACE) in a cohort of consecutive patients referred for stress CMR. This retrospective, single-center observational study included all consecutive patients with known or suspected coronary artery disease referred for stress CMR between 2016 and 2018. Fully automated machine learning was used to obtain global longitudinal strain at rest (rest-GLS) and global circumferential strain at stress (stress-GCS). The primary outcome was MACE, including cardiovascular death or hospitalization for heart failure. Cox models were used to assess the incremental prognostic value of combining these strain features with traditional prognosticators. Of 2778 patients (age 65±12 years, 68% male), 96% had feasible, fully automated rest-GLS and stress-GCS measurements. After a median follow-up of 5.2 (4.8-5.5) years, 316 (11.1%) patients experienced MACE. After adjustment for traditional prognosticators, both rest-GLS (hazard ratio, 1.09 [95% CI, 1.05-1.13]; <i>P</i><0.001) and stress-GCS (hazard ratio, 1.08 [95% CI, 1.03-1.12]; <i>P</i><0.001) were independently associated with MACE. The best cutoffs for MACE prediction were >-10% for rest-GLS and stress-GCS, with a C-index improvement of 0.02, continuous net reclassification improvement of 15.6%, and integrative discrimination index of 2.2% (all <i>P</i><0.001). The combination of rest-GLS and stress-GCS, with a cutoff of >-10% provided an incremental prognostic value over and above traditional prognosticators, including CMR parameters, for predicting MACE in patients undergoing stress CMR.

<b><i>Background:</i></b> Combining functional magnetic resonance imaging (fMRI) and machine learning (ML) can be used to identify therapeutic targets and evaluate the effect of repetitive transcranial magnetic stimulation (rTMS) in neural networks in tobacco use disorder. We investigated whether large-scale network connectivity can predict the rTMS effect on smoking cessation. <b><i>Methods:</i></b> Smoking cue exposure task-fMRI (T-fMRI) and resting-state fMRI (Rs-fMRI) scans were acquired before and after the 10 sessions of active or sham rTMS (10 Hz, 3000 pulses per session) over the left dorsal lateral prefrontal cortex in 42 treatment-seeking smokers. Five large-scale networks (default model network, central executive network, dorsal attention network, salience network [SN], and reward network) were compared before and after 10 sessions of rTMS, as well as between active and sham rTMS conditions. We performed neural network and regression analysis on the average connectivity of large-scale networks and the effectiveness of rTMS induced by rTMS. <b><i>Results:</i></b> Regression analyses indicated higher salience connectivity in T-fMRI and lower reward connectivity in Rs-fMRI, predicting a better outcome of TMS treatment for smoking cessation (<i>p</i> < 0.01, Bonferroni corrected). Neural Network analyses suggested that SN was the most important predictor of rTMS effectiveness in both T-fMRI (0.33 of feature importance) and Rs-fMRI (0.37 feature importance). <b><i>Conclusions:</i></b> Both T-fMRI and Rs-fMRI connectivity in SN predict a better outcome of TMS treatment for smoking cessation, but in opposite directions. The work shows that ML models can be used to target TMS treatment. Given the small sample size, all ML findings should be replicated in a larger cohort to ensure their validity.

X-ray imaging is a ubiquitous in radiology, yet most existing AI foundation models are limited to chest anatomy and fail to generalize across broader clinical tasks. In this work, we introduce XR-0, the multi-anatomy X-ray foundation model using self-supervised learning on a large, private dataset of 1.15 million images spanning diverse anatomical regions and evaluated across 12 datasets and 20 downstream tasks, including classification, retrieval, segmentation, localization, visual grounding, and report generation. XR-0 achieves state-of-the-art performance on most multi-anatomy tasks and remains competitive on chest-specific benchmarks. Our results demonstrate that anatomical diversity and supervision are critical for building robust, general-purpose medical vision models, paving the way for scalable and adaptable AI systems in radiology.

Major depressive disorder (MDD) is a prevalent mental health condition that negatively impacts both individual well-being and global public health. Automated detection of MDD using structural magnetic resonance imaging (sMRI) and deep learning (DL) methods holds increasing promise for improving diagnostic accuracy and enabling early intervention. Most existing methods employ either voxel-level features or handcrafted regional representations built from predefined brain atlases, limiting their ability to capture complex brain patterns. This paper develops a unified pipeline that utilizes Vision Transformers (ViTs) for extracting 3D region embeddings from sMRI data and Graph Neural Network (GNN) for classification. We explore two strategies for defining regions: (1) an atlas-based approach using predefined structural and functional brain atlases, and (2) an cube-based method by which ViTs are trained directly to identify regions from uniformly extracted 3D patches. Further, cosine similarity graphs are generated to model interregional relationships, and guide GNN-based classification. Extensive experiments were conducted using the REST-meta-MDD dataset to demonstrate the effectiveness of our model. With stratified 10-fold cross-validation, the best model obtained 78.98% accuracy, 76.54% sensitivity, 81.58% specificity, 81.58% precision, and 78.98% F1-score. Further, atlas-based models consistently outperformed the cube-based approach, highlighting the importance of using domain-specific anatomical priors for MDD detection.

Computer-aided diagnosis systems must make critical decisions from medical images that are often noisy, ambiguous, or conflicting, yet today's models are trained on overly simplistic labels that ignore diagnostic uncertainty. One-hot labels erase inter-rater variability and force models to make overconfident predictions, especially when faced with incomplete or artifact-laden inputs. We address this gap by introducing a novel framework that brings uncertainty back into the label space. Our method leverages neural network training dynamics (NNTD) to assess the inherent difficulty of each training sample. By aggregating and calibrating model predictions during training, we generate uncertainty-aware pseudo-labels that reflect the ambiguity encountered during learning. This label augmentation approach is architecture-agnostic and can be applied to any supervised learning pipeline to enhance uncertainty estimation and robustness. We validate our approach on a challenging echocardiography classification benchmark, demonstrating superior performance over specialized baselines in calibration, selective classification, and multi-view fusion.

Multimodal learning has attracted much attention in recent years due to its ability to effectively utilize data features from a variety of different modalities. Diagnosing the vulnerability of atherosclerotic plaques directly from carotid 3D MRI images is relatively challenging for both radiologists and conventional 3D vision networks. In clinical practice, radiologists assess patient conditions using a multimodal approach that incorporates various imaging modalities and domain-specific expertise, paving the way for the creation of multimodal diagnostic networks. In this paper, we have developed an effective strategy to leverage radiologists' domain knowledge to automate the diagnosis of carotid plaque vulnerability through Variation inference and Multimodal knowledge Distillation (VMD). This method excels in harnessing cross-modality prior knowledge from limited image annotations and radiology reports within training data, thereby enhancing the diagnostic network's accuracy for unannotated 3D MRI images. We conducted in-depth experiments on the dataset collected in-house and verified the effectiveness of the VMD strategy we proposed.

Deep learning (DL) based on MRI of medulloblastoma enables risk stratification, potentially aiding in therapeutic decisions. This study aims to develop DL models that identify four medulloblastoma molecular subgroups and prognostic-related genetic signatures. This retrospective study enrolled 325 patients for model development and an independent external validation cohort of 124 patients, totaling 449 MB patients from 2 medical institutes. Consecutive patients with newly diagnosed MB at MRI (T1-weighted, T2-weighted, and contrast-enhanced T1-weighted) at two medical institutes between January 2015 and June 2023 were identified. Two-stage sequential DL models were designed-MB-CNN that first identifies wingless (WNT), sonic hedgehog (SHH), Group 3, and Group 4. Further, prognostic-related genetic signatures using DL models (MB-CNN_TP53/MYC/Chr11) were developed to predict TP53 mutation, MYC amplification, and chromosome 11 loss status. A hybrid model combining MB-CNN and conventional data (clinical information and MRI features) was compared to a logistic regression model constructed only with conventional data. Four-classification tasks were evaluated with confusion matrices (accuracy) and two-classification tasks with ROC curves (area under the curve (AUC)). The datasets comprised 449 patients (mean age ± SD at diagnosis, 13.55 years ± 2.33, 249 males). MB-CNN accurately classified MB subgroups in the external test dataset, achieving a median accuracy of 77.50% (range in 76.29% to 78.71%). MB-CNN_TP53/MYC/Chr11 models effectively predicted signatures (AUC of TP53 in SHH: 0.91, MYC amplification in Group 3: 0.87, chromosome 11 loss in Group 4: 0.89). The accuracy of the hybrid model outperformed the logistic regression model (82.20% vs. 59.14%, P = .009) and showed comparable performance to MB-CNN (82.20% vs. 77.50%, P = 0.105). MRI-based DL models allowed identification of the molecular medulloblastoma subgroups and prognostic-related genetic signatures.

This study investigates the correlation between chest computed tomography (CT) radiomics features and breast density classification, and aiming to develop an automated radiomics model for breast density assessment using chest CT images. The diagnostic performance was evaluated to establish a CT-based alternative for breast density classification in clinical practice. A retrospective analysis was conducted on patients who underwent both mammography and chest CT scans. The breast density classification results based on mammography images were used to guide the development of CT-based breast density classification models. Radiomic features were extracted from breast regions of interest (ROIs) segmented on chest CT images. The diagnostic performance was evaluated to establish a CT-based alternative for breast density classification in clinical practice. Following dimensionality reduction and selection of dominant radiomic features, four four-class classification models were established, including ① Extreme Gradient Boosting (XGBoost), ② One Vs Rest Classifier-Logistic Regression, ③ Gradient Boosting, and ④ Random Forest Classifier. The performance of these models in classifying breast density using CT images was then evaluated. A total of 330 patients, aged 23-79 years, were included for analysis. The breast ROIs were automatically segmented using a U-net neural network model and subsequently refined and calibrated manually. A total of 1427 radiomic features were extracted, and after dimensionality reduction and feature selection, 28 dominant features closely associated with breast density classification were obtained to construct four classification models. Among the tested models-XGBoost, One-vs-Rest Logistic Regression, Gradient Boosting Classifier, and Random Forest Classifier-the XGBoost model achieved the best performance, with a classification accuracy of 86.6%. Analysis of the receiver operating characteristic curves showed Area Under the Curve (AUC) values of 1.00, 0.93, 0.93, and 0.99 for the four breast density categories, along with a micro-averaged AUC of 0.97 and a macro-averaged AUC of 0.96. Chest CT scans, combined with imaging radiomics models, can accurately classify breast density, providing valuable information related to breast cancer risk stratification. The proposed classification model offers a promising tool for automated breast density assessment, which could enhance personalized breast cancer screening and clinical decision-making.

Anomaly detection and classification in medical imaging are critical for early diagnosis but remain challenging due to limited annotated data, class imbalance, and the high cost of expert labeling. Emerging vision foundation models such as DINOv2, pretrained on extensive, unlabeled datasets, offer generalized representations that can potentially alleviate these limitations. In this study, we propose an attention-based global aggregation framework tailored specifically for 3D medical image anomaly classification. Leveraging the self-supervised DINOv2 model as a pretrained feature extractor, our method processes individual 2D axial slices of brain MRIs, assigning adaptive slice-level importance weights through a soft attention mechanism. To further address data scarcity, we employ a composite loss function combining supervised contrastive learning with class-variance regularization, enhancing inter-class separability and intra-class consistency. We validate our framework on the ADNI dataset and an institutional multi-class headache cohort, demonstrating strong anomaly classification performance despite limited data availability and significant class imbalance. Our results highlight the efficacy of utilizing pretrained 2D foundation models combined with attention-based slice aggregation for robust volumetric anomaly detection in medical imaging. Our implementation is publicly available at https://github.com/Rafsani/DinoAtten3D.git.