Breast cancer continues to be a global public health challenge. An early and precise diagnosis is crucial for improving prognosis and efficacy. While deep learning (DL) methods have shown promising advances in breast cancer classification from mammogram images, most existing DL models remain static, single-view image-based, and overlook the longitudinal progression of lesions and patient-specific clinical context. Moreover, the majority of models also limited their clinical usability by designing tests for subtype classification in isolation (i.e., not predicting disease stages simultaneously). This paper introduces BreastXploreAI, a simple yet powerful multimodal, multitask deep learning framework for breast cancer diagnosis to fill these gaps. TransBreastNet, a hybrid architecture that combines convolutional neural networks (CNNs) for spatial encoding of lesions, a Transformer-based modular approach for temporal encoding of lesions, and dense metadata encoders for fusion of patient-specific clinical information, forms the backbone of our system. The breast cancer subtype and disease stage are predicted simultaneously from a dual-head classifier. They are then used to construct temporal lesion sequences, either by employing genuine longitudinal data or by adding sequence augmentation to sample sequences, thereby strengthening the model's ability to learn Progression Patterns. We conduct extensive experiments on a public mammogram dataset and demonstrate that our model outperforms several state-of-the-art baselines in both subtype classification, achieving a macro accuracy of 95.2%, and stage Prediction, with a macro accuracy of 93.8%. We also provide ablation studies, which confirm how every module contributes to the framework. Unlike prior static single-view models, our framework jointly models spatial, temporal, and clinical features using a CNN-Transformer hybrid design. It simultaneously predicts breast cancer subtypes and lesion progression stages, while generating synthetic temporal lesion sequences where longitudinal data is scarce. Built-in explainability modules enhance interpretability and clinical trust. BreastXploreAI offers a robust, scalable, and clinically relevant approach to diagnosing breast cancer from full-field digital mammogram (FFDM) images. ZH is computationally capable of analyzing spatial, temporal, and clinical features simultaneously, which enables a more informed diagnosis and lays the foundation for improved clinical decision support systems in oncology.

In benign tumors with potential for malignant transformation, sampling error during pre-operative biopsy can significantly change patient counseling and surgical planning. Sinonasal inverted papilloma (IP) is the most common benign soft tissue tumor of the sinuses, yet it can undergo malignant transformation to squamous cell carcinoma (IP-SCC), for which the planned surgery could be drastically different. Artificial intelligence (AI) could potentially help with this diagnostic challenge. CT images from 19 institutions were used to train the Google Cloud Vertex AI platform to distinguish between IP and IP-SCC. The model was evaluated on a holdout test dataset of images from patients whose data were not used for training or validation. Performance metrics of area under the curve (AUC), sensitivity, specificity, accuracy, and F1 were used to assess the model. Here we show CT image data from 958 patients and 41099 individual images that were labeled to train and validate the deep learning image classification model. The model demonstrated a 95.8 % sensitivity in correctly identifying IP-SCC cases from IP, while specificity was robust at 99.7 %. Overall, the model achieved an accuracy of 99.1%. A deep automated machine learning model, created from a publicly available artificial intelligence tool, using pre-operative CT imaging alone, identified malignant transformation of inverted papilloma with excellent accuracy.

Segmentation of medical images is critical for the evaluation, diagnosis, and treatment of various medical conditions. While deep learning-based approaches are the dominant methodology, they rely heavily on abundant labeled data and face significant challenges when data is limited. Semi-supervised learning methods mitigate this issue but there are still some challenges associated with them. Additionally, these approaches can be improved specifically for medical images considering their unique properties (e.g., smooth boundaries). In this work, we adapt and enhance the well-established pseudo-labeling approach specifically for medical image segmentation. Our exploration consists of modifying the network's loss function, pruning the pseudo-labels, and refining pseudo-labels by integrating traditional image processing methods with semi-supervised learning. This integration enables traditional segmentation techniques to complement deep semi-supervised methods, particularly in capturing fine edges where deep models often struggle. It also incorporates the smoothness of the edges in the segmentation and achieves a balance between deep learning and traditional methods through tunable parameters. Moreover, to address the problem of noisy or unreliable pseudo-labels, we utilize uncertainty-based pixel-level and image-level pruning of the pseudo-labels using a specific loss function, thereby improving the accuracy and robustness of the segmentation. We evaluated our approach on three different datasets from two imaging modalities (CT and MRI) and demonstrated its superior performance, highlighting its accuracy and robustness in the presence of limited labeled data. With only 15% of the labeled data, on the Sunnybrook Cardiac dataset, our approaches increased endocardium segmentation accuracy from 82.1% to 87.5%, and epicardium segmentation from 82.5% to 86.7%. On the COVID-19 CT lung and infection segmentation dataset, our approach improved left lung segmentation accuracy from 72.5% to 79.3%, and right lung segmentation from 75.8% to 81.6% when using only 15% of labeled data. On the Automated Cardiac Diagnostic Challenge dataset, with just 10% of labeled data, our approach increased endocardium segmentation from 91% to 93.7%, myocardium from 69.8% to 74.5%, and right ventricle from 76.7% to 82.1%. Our codes will be published in https://github.com/behnam-rahmati .

Accurate diagnosis of brain tumors is critical in understanding the prognosis in terms of the type, growth rate, location, removal strategy, and overall well-being of the patients. Among different modalities used for the detection and classification of brain tumors, a computed tomography (CT) scan is often performed as an early-stage procedure for minor symptoms like headaches. Automated procedures based on artificial intelligence (AI) and machine learning (ML) methods are used to detect and classify brain tumors in Computed Tomography (CT) scan images. However, the key challenges in achieving the desired outcome are associated with the model's complexity and generalization. To address these issues, we propose a hybrid model that extracts features from CT images using classical machine learning. Additionally, although MRI is a common modality for brain tumor diagnosis, its high cost and longer acquisition time make CT scans a more practical choice for early-stage screening and widespread clinical use. The proposed framework has different stages, including image acquisition, pre-processing, feature extraction, feature selection, and classification. The hybrid architecture combines features from ResNet50, AlexNet, LBP, HOG, and median intensity, classified using a multilayer perceptron. The selection of the relevant features in our proposed hybrid model was extracted using the SelectKBest algorithm. Thus, it optimizes the proposed model performance. In addition, the proposed model incorporates data augmentation to handle the imbalanced datasets. We employed a scoring function to extract the features. The Classification is ensured using a multilayer perceptron neural network (MLP). Unlike most existing hybrid approaches, which primarily target MRI-based brain tumor classification, our method is specifically designed for CT scan images, addressing their unique noise patterns and lower soft-tissue contrast. To the best of our knowledge, this is the first work to integrate LBP, HOG, median intensity, and deep features from both ResNet50 and AlexNet in a structured fusion pipeline for CT brain tumor classification. The proposed hybrid model is tested on data from numerous sources and achieved an accuracy of 94.82%, precision of 94.52%, specificity of 98.35%, and sensitivity of 94.76% compared to state-of-the-art models. While MRI-based models often report higher accuracies, the proposed model achieves 94.82% on CT scans, within 3-4% of leading MRI-based approaches, demonstrating strong generalization despite the modality difference. The proposed hybrid model, combining hand-crafted and deep learning features, effectively improves brain tumor detection and classification accuracy in CT scans. It has the potential for clinical application, aiding in early and accurate diagnosis. Unlike MRI, which is often time-intensive and costly, CT scans are more accessible and faster to acquire, making them suitable for early-stage screening and emergency diagnostics. This reinforces the practical and clinical value of the proposed model in real-world healthcare settings.

Alzheimer's disease is a progressive neurodegenerative disorder that is challenging to diagnose at an early stage. Affecting over 55 million people worldwide, its prevalence is expected to rise sharply by 2030. The use of artificial intelligence (AI) techniques has become increasingly important to improve the speed and accuracy of diagnosis. In this study, we propose the NCA-Enhanced Voting Algorithm for Alzheimer's Classification (NCA-EVA) to support computer-aided diagnosis. A total of 66 models were trained for four-class data and six models for two-class data. The proposed method successfully classified all four stages of Alzheimer's disease, achieving 98.97% accuracy in four-class classification and 99.87% accuracy in binary classification. Moreover, with a processing time of just 1.26 s, NCA-EVA is approximately 1200 times faster than a comparable study using NCA-based feature selection. These findings demonstrate that Alzheimer's diagnosis can be performed both quickly and with high accuracy, and the proposed approach has potential applications in other healthcare data analysis tasks.

Early and accurate diagnosis of pulmonary hypertension (PH), including differentiating pre-capillary from post-capillary PH, is crucial for guiding effective clinical management. This study developed and validated a deep learning-based diagnostic model to classify patients into non-PH, pre-capillary PH, or post-capillary PH categories. A retrospective dataset from 204 patients (112 pre-capillary PH, 32 post-capillary PH, and 60 non-PH controls) was collected at the First Affiliated Hospital of Nanjing Medical University, with diagnoses confirmed by right heart catheterization (RHC). Patients were randomly divided into training (186 patients, 90%) and testing sets (18 patients, 10%) stratified by diagnostic category. We trained and evaluated the model using 35 repeated random splits. The proposed deep learning model combined graph convolutional networks (GCN), convolutional neural networks (CNN), and Transformers to analyze multimodal data, including cine short-axis (SAX) sequences, four-chamber (4CH) sequences, and clinical parameters. Across test splits, the model achieved an overall area under the receiver operating characteristic curve (AUC) of 0.814 ± 0.06 and accuracy (ACC) of 0.734 ± 0.06 (mean ± SD). Class-specific AUCs were 0.745 ± 0.11 for non-PH, 0.863 ± 0.06 for pre-capillary PH, and 0.834 ± 0.10 for post-capillary PH, indicating good discriminative ability. This study demonstrated three-class PH classification using multimodal inputs. By fusing imaging and clinical data, the model may support accurate and timely clinical decision-making in PH.

Glioblastoma (GBM) remains one of the most formidable brain malignancies, characterized by a heterogeneous genetic profile that significantly influences patient prognosis. Per the 2021 WHO central nervous system classification, GBM is defined as an isocitrate dehydrogenase (IDH) wild-type diffuse astrocytic tumor. We analyzed two multi-institutional cohorts, UPENN-GBM (644 patients) and UCSF-PDGM (420 patients); after excluding the 116 and 42 IDH-mutant records, 528 and 378 wild-type cases remained for modelling. MGMT promoter methylation, present in 43% of GBM cases, correlates with enhanced survival outcomes, demonstrating a median survival of 504 days versus 329 days in unmethylated cases. In this study, we present a novel integration of imaging phenotypes, clinical characteristics, and molecular markers through the application of advanced machine learning methodologies, including Random Forest, XGBoost, LightGBM, and an optimized dense neural network (Dense NN). This integrative approach aims to refine survival prediction in GBM patients. MRI data were meticulously processed using the MRIPreprocessor tool and the radiomics Python library, facilitating the extraction of high-dimensional radiomic features. Our findings reveal that the proposed custom Dense NN model outperformed traditional tree-based algorithms, with the Dense NN achieving a concordance index (CI) of 0.86 on the UPENN-GBM dataset and 0.83 on the UCSF-PDGM dataset. The optimized Dense NN architecture features three hidden layers with 256, 128, and 64 units respectively, employing ReLU activation, L1/L2 regularization to mitigate overfitting, batch normalization to stabilize training, and dropout for improved generalization. This specific configuration was determined through hyperparameter tuning using techniques like RandomizedSearchCV. This integrative, non-invasive methodology provides a more nuanced assessment of tumor biology, thereby advancing the development of personalized therapeutic strategies. Our results underscore the transformative potential of artificial intelligence in delineating disease trajectories and optimizing treatment paradigms. Moreover, this research establishes a robust framework for future investigations in glioblastoma survival prediction, illustrating the efficacy of combining clinical, genetic, and imaging data to enhance prognostic accuracy within precision medicine paradigms for GBM patients.

Accurate and efficient classification of lung diseases from medical images remains a significant challenge in computer-aided diagnosis systems. This research presents a novel approach integrating transfer learning techniques with fuzzy decision support systems for multi-class lung disease classification. We compare the performance of three pre-trained CNN architectures-VGG16, VGG19, and ResNet50-enhanced with a fuzzy logic decision layer. The proposed methodology employs transfer learning to leverage knowledge from large-scale datasets while adapting to the specific characteristics of lung disease images. A k-symbol Lerch transcendent function is implemented for image enhancement during preprocessing, significantly improving feature extraction capabilities by 23.4% in contrast enhancement and 18.7% in feature visibility. The fuzzy decision support system addresses inherent uncertainties in medical image classification through membership functions and rule-based inference mechanisms specifically designed for lung pathology features. Experimental evaluation was conducted on a comprehensive dataset of 8,409 chest X-ray images across six disease classes: COVID-19, Pneumonia, Tuberculosis, Lung Opacity, Cardiomegaly, and Normal cases. Results demonstrate that the ResNet50-based model with fuzzy integration achieves superior classification accuracy of 98.7%, sensitivity of 98.4%, and specificity of 98.8%, outperforming standard implementations of VGG16 (97.8% accuracy) and VGG19 (98.2% accuracy). The proposed approach shows particular strength in handling borderline cases where traditional CNN confidence falls below 75%, achieving 8.4% improvement in uncertain case classification. Statistical significance testing confirms meaningful performance gains (p < 0.05) across all architectures, with ResNet50 showing the most substantial enhancement (p = 0.0018). The fuzzy inference system activates an average of 8.4 rules per classification decision, providing transparent reasoning pathways that enhance clinical interpretability while maintaining real-time processing capability (0.23 s per image). This research contributes to advancing automated lung disease diagnosis systems with improved accuracy, uncertainty handling, and clinical interpretability for computer-aided diagnostic applications.

Motion artifacts remain a key limitation in brain MRI, particularly during 3D acquisitions in cognitively impaired patients. Most deep learning (DL) reconstruction techniques improve signal-to-noise ratio but lack explicit mechanisms to correct for motion. This study aims to validate a DL reconstruction method that integrates retrospective motion correction into the reconstruction pipeline for 3D T1-weighted brain MRI. This prospective, intra-individual comparison study included a controlled-motion cohort of healthy volunteers and a clinical cohort of patients undergoing evaluation for memory loss. Each cohort was scanned at distinct imaging sites between October 2022 and August 2023 in staggered periods. All participants underwent 4-fold under-sampled 3D magnetization-prepared rapid gradient-echo imaging with integrated Scout Accelerated Motion Estimation and Reduction (SAMER) acquisition. Image volumes were reconstructed using standard-of-care methods and the proposed DL approach. Quantitative morphometric accuracy was assessed by comparing brain segmentation results of instructed-motion scans to motion-free reference scans in the healthy volunteers. Image quality was rated by two board-certified neuroradiologists using a five-point Likert scale. Statistical analysis included Wilcoxon tests and intraclass correlation coefficients. A total of 41 participants (15 women [37%]; mean age, 58 years) and 154 image volumes were evaluated. The DL-based method with integrated motion correction significantly reduced segmentation error under moderate and severe motion (12.4% to 3.5% and 44.2% to 12.5%, respectively; P < .001). Visual ratings showed improved scores across all criteria compared with standard reconstructions (overall image quality, 4.26 ± 0.72 vs. 3.59 ± 0.82; P < .001). In 47% of cases, motion artifact severity was improved following DL-based processing. Inter-reader agreement ranged from moderate to substantial. Motion-informed DL reconstruction improved both morphometric accuracy and perceived image quality in 3D T1-weighted brain MRI. This technique may enhance diagnostic utility and reduce scan failure rates in motion-prone patients with cognitive impairment. AD = Alzheimer's disease; DL = deep learning; ICC = intra-class correlation coefficient; SAMER = scout accelerated motion estimation and reduction.

Motion artifacts in brain MRI, mainly from rigid head motion, degrade image quality and hinder downstream applications. Conventional methods to mitigate these artifacts, including repeated acquisitions or motion tracking, impose workflow burdens. This study introduces Res-MoCoDiff, an efficient denoising diffusion probabilistic model specifically designed for MRI motion artifact correction. Res-MoCoDiff exploits a novel residual error shifting mechanism during the forward diffusion process to incorporate information from motion-corrupted images. This mechanism allows the model to simulate the evolution of noise with a probability distribution closely matching that of the corrupted data, enabling a reverse diffusion process that requires only four steps. The model employs a U-net backbone, with attention layers replaced by Swin Transformer blocks, to enhance robustness across resolutions. Furthermore, the training process integrates a combined l1+l2 loss function, which promotes image sharpness and reduces pixel-level errors. Res-MoCoDiff was evaluated on both an in-silico dataset generated using a realistic motion simulation framework and an in-vivo MR-ART dataset. Comparative analyses were conducted against established methods, including CycleGAN, Pix2pix, and a diffusion model with a vision transformer backbone (MT-DDPM), using quantitative metrics such as peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and normalized mean squared error (NMSE). The proposed method demonstrated superior performance in removing motion artifacts across minor, moderate, and heavy distortion levels. Res-MoCoDiff consistently achieved the highest SSIM and the lowest NMSE values, with a PSNR of up to 41.91±2.94 dB for minor distortions. Notably, the average sampling time was reduced to 0.37 seconds per batch of two image slices, compared with 101.74 seconds for conventional approaches. Res-MoCoDiff offers a robust and efficient solution for correcting MRI motion artifacts, preserving fine structural details while significantly reducing computational overhead. Its speed and restoration fidelity underscore its potential for integration into clinical workflows, enhancing diagnostic accuracy and patient care.