Magnetic Resonance Imaging (MRI) based three-dimensional analysis of knee cartilage has evolved to become fully automatic. However, when implementing these measurements across multiple clinical centers, scanner variability becomes a critical consideration. Our purposes were to quantify and compare same-model variability (between repeated scans on the same MRI system) and cross-model variability (across different MRI systems) in knee cartilage thickness measurements using MRI scanners from five manufacturers, as analyzed with a specific 3D volume analysis software. Ten healthy volunteers (eight males and two females, aged 22-60 years) underwent two scans of their right knee on 3T MRI systems from five manufacturers (Canon, Fujifilm, GE, Philips, and Siemens). The imaging protocol included fat-suppressed spoiled gradient echo and proton density weighted sequences. Cartilage regions were automatically segmented into 7 subregions using a specific deep learning-based 3D volume analysis software. This resulted in 350 measurements for same-model variability and 2,800 measurements for cross-model variability. For same-model variability, 82% of measurements showed variability ≤0.10 mm, and 98% showed variability ≤0.20 mm. For cross-model variability, 51% showed variability ≤0.10 mm, and 84% showed variability ≤0.20 mm. The mean same-model variability (0.06 ± 0.05 mm) was significantly lower than cross-model variability (0.11 ± 0.09 mm) (p < 0.001). This study demonstrates that knee cartilage thickness measurements exhibit significantly higher variability across different MRI systems compared to repeated measurements on the same system, when analyzed using this specific software. This finding has important implications for multi-center studies and longitudinal assessments using different MRI systems and highlights the software-dependent nature of such variability assessments.

Interval breast cancers (IBCs) are detected between regular mammographic screenings after an initially negative result. Studies have shown that the prognosis of IBCs is similar to that of unscreened symptomatic cancers and is hence a surrogate used to assess the effectiveness of screening programs. This systematic review consolidates the current literature available on strategies to reduce the rates of IBC. Following PRISMA guidelines, three databases were searched from inception till October 29, 2023 to identify papers, which reported IBC rates. Key search terms included "interval breast cancer", "mammogram", "tomosynthesis" and "screening". A total of 32 articles were included. Fourteen studies discussed the use of digital breast tomosynthesis (DBT) as an alternative screening modality to mammograms. Six studies discussed the use of artificial intelligence (AI) on mammograms, five studies discussed the use of supplemental modalities including ultrasonography (US) in addition to mammograms, five studies discussed varying screening intervals and two studies discussed tamoxifen use. The trajectory of IBCs can be altered by early detection when they are more amenable to treatment, through advanced screening techniques, adjusting inter-screening intervals and modifiable risk factors. The goal is to create a screening protocol that is economically effective and accessible to various populations.

To develop an interpretable machine learning (ML) model using dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) radiomic data, dosimetric parameters, and clinical data for predicting radiation-induced hepatic toxicity (RIHT) in patients with hepatocellular carcinoma (HCC) following intensity-modulated radiation therapy (IMRT). A retrospective analysis of 150 HCC patients was performed, with a 7:3 ratio used to divide the data into training and validation cohorts. Radiomic features from the original MRI sequences and Delta-radiomic features were extracted. Seven ML models based on radiomics were developed: logistic regression (LR), random forest (RF), support vector machine (SVM), eXtreme Gradient Boosting (XGBoost), adaptive boosting (AdaBoost), decision tree (DT), and artificial neural network (ANN). The predictive performance of the models was evaluated using receiver operating characteristic (ROC) curve analysis and calibration curves. Shapley additive explanations (SHAP) were employed to interpret the contribution of each variable and its risk threshold. Original radiomic features and Delta-radiomic features were extracted from DCE-MRI images and filtered to generate Radiomics-scores and Delta-Radiomics-scores. These were then combined with independent risk factors (Body Mass Index (BMI), V5, and pre-Child-Pugh score(pre-CP)) identified through univariate and multivariate logistic regression and Spearman correlation analysis to construct the ML models. In the training cohort, the AUC values were 0.8651 for LR, 0.7004 for RF, 0.6349 for SVM, 0.6706 for XGBoost, 0.7341 for AdaBoost, 0.6806 for Decision Tree, and 0.6786 for ANN. The corresponding accuracies were 84.4%, 65.6%, 75.0%, 65.6%, 71.9%, 68.8%, and 71.9%, respectively. The validation cohort further confirmed the superiority of the LR model, which was selected as the optimal model. SHAP analysis revealed that Delta-radiomics made a substantial positive contribution to the model. The interpretable ML model based on radiomics provides a non-invasive tool for predicting RIHT in patients with HCC, demonstrating satisfactory discriminative performance.

Early and accurate detection of COVID-19 and pneumonia through medical imaging is critical for effective patient management. This study aims to develop a robust framework that integrates synthetic image augmentation with advanced deep learning (DL) models to address dataset imbalance, improve diagnostic accuracy, and enhance trust in artificial intelligence (AI)-driven diagnoses through Explainable AI (XAI) techniques. The proposed framework benchmarks state-of-the-art models (InceptionV3, DenseNet, ResNet) for initial performance evaluation. Synthetic images are generated using Feature Interpolation through Linear Mapping and principal component analysis to enrich dataset diversity and balance class distribution. YOLOv8 and InceptionV3 models, fine-tuned via transfer learning, are trained on the augmented dataset. Grad-CAM is used for model explainability, while large language models (LLMs) support visualization analysis to enhance interpretability. YOLOv8 achieved superior performance with 97% accuracy, precision, recall, and F1-score, outperforming benchmark models. Synthetic data generation effectively reduced class imbalance and improved recall for underrepresented classes. Comparative analysis demonstrated significant advancements over existing methodologies. XAI visualizations (Grad-CAM heatmaps) highlighted anatomically plausible focus areas aligned with clinical markers of COVID-19 and pneumonia, thereby validating the model's decision-making process. The integration of synthetic data generation, advanced DL, and XAI significantly enhances the detection of COVID-19 and pneumonia while fostering trust in AI systems. YOLOv8's high accuracy, coupled with interpretable Grad-CAM visualizations and LLM-driven analysis, promotes transparency crucial for clinical adoption. Future research will focus on developing a clinically viable, human-in-the-loop diagnostic workflow, further optimizing performance through the integration of transformer-based language models to improve interpretability and decision-making.

Lung cancer imaging plays a crucial role in early diagnosis and treatment, where machine learning and deep learning have significantly advanced the accuracy and efficiency of disease classification. This study introduces the Explainable and Lightweight Lung Cancer Net (XLLC-Net), a streamlined convolutional neural network designed for classifying lung cancer from histopathological images. Using the LC25000 dataset, which includes three lung cancer classes and two colon cancer classes, we focused solely on the three lung cancer classes for this study. XLLC-Net effectively discerns complex disease patterns within these classes. The model consists of four convolutional layers and contains merely 3 million parameters, considerably reducing its computational footprint compared to existing deep learning models. This compact architecture facilitates efficient training, completing each epoch in just 60 seconds. Remarkably, XLLC-Net achieves a classification accuracy of 99.62% [Formula: see text] 0.16%, with precision, recall, and F1 score of 99.33% [Formula: see text] 0.30%, 99.67% [Formula: see text] 0.30%, and 99.70% [Formula: see text] 0.30%, respectively. Furthermore, the integration of Explainable AI techniques, such as Saliency Map and GRAD-CAM, enhances the interpretability of the model, offering clear visual insights into its decision-making process. Our results underscore the potential of lightweight DL models in medical imaging, providing high accuracy and rapid training while ensuring model transparency and reliability.

Tuberculous meningitis (TBM) leads to high mortality, especially amongst individuals with HIV. Predicting the incidence of disease-related complications is challenging, for which purpose the value of brain magnetic resonance imaging (MRI) has not been well investigated. We used a convolutional neural network (CNN) to explore the complementary contribution of brain MRI to the conventional prognostic determinants. We pooled data from two randomised control trials of HIV-positive and HIV-negative adults with clinical TBM in Vietnam to predict the occurrence of death or new neurological complications in the first two months after the subject's first MRI session. We developed and compared three models: a logistic regression with clinical, demographic and laboratory data as reference, a CNN that utilised only T1-weighted MRI volumes, and a model that fused all available information. All models were fine-tuned using two repetitions of 5-fold cross-validation. The final evaluation was based on a random 70/30 training/test split, stratified by the outcome and HIV status. Based on the selected model, we explored the interpretability maps derived from the models. 215 patients were included, with an event prevalence of 22.3%. On the test set our non-imaging model had higher AUC (71.2% [Formula: see text] 1.1%) than the imaging-only model (67.3% [Formula: see text] 2.6%). The fused model was superior to both, with an average AUC = 77.3% [Formula: see text] 4.0% in the test set. The non-imaging variables were more informative in the HIV-positive group, while the imaging features were more predictive in the HIV-negative group. All three models performed better in the HIV-negative cohort. The interpretability maps show the model's focus on the lateral fissures, the corpus callosum, the midbrain, and peri-ventricular tissues. Imaging information can provide added value to predict unwanted outcomes of TBM. However, to confirm this finding, a larger dataset is needed.

Currently, there are inconsistencies among different studies on preoperative prediction of Cytokeratin 19 (CK19) expression in HCC using traditional imaging, radiomics, and deep learning. We aimed to systematically analyze and compare the performance of non-invasive methods for predicting CK19-positive HCC, thereby providing insights for the stratified management of HCC patients. A comprehensive literature search was conducted in PubMed, EMBASE, Web of Science, and the Cochrane Library from inception to February 2025. Two investigators independently screened and extracted data based on inclusion and exclusion criteria. Eligible studies were included, and key findings were summarized in tables to provide a clear overview. Ultimately, 22 studies involving 3395 HCC patients were included. 72.7% (16/22) focused on traditional imaging, 36.4% (8/22) on radiomics, 9.1% (2/22) on deep learning, and 54.5% (12/22) on combined models. The magnetic resonance imaging was the most commonly used imaging modality (19/22), and over half of the studies (12/22) were published between 2022 and 2025. Moreover, 27.3% (6/22) were multicenter studies, 36.4% (8/22) included a validation set, and only 13.6% (3/22) were prospective. The area under the curve (AUC) range of using clinical and traditional imaging was 0.560 to 0.917. The AUC ranges of radiomics were 0.648 to 0.951, and the AUC ranges of deep learning were 0.718 to 0.820. Notably, the AUC ranges of combined models of clinical, imaging, radiomics and deep learning were 0.614 to 0.995. Nevertheless, the multicenter external data were limited, with only 13.6% (3/22) incorporating validation. The combined model integrating traditional imaging, radiomics and deep learning achieves excellent potential and performance for predicting CK19 in HCC. Based on current limitations, future research should focus on building an easy-to-use dynamic online tool, combining multicenter-multimodal imaging and advanced deep learning approaches to enhance the accuracy and robustness of model predictions.

Volumetric atlases are an invaluable tool in neuroscience and otolaryngology, greatly aiding experiment planning and surgical interventions, as well as the interpretation of experimental and clinical data. The rat is a major animal model for hearing and balance studies, and a detailed volumetric atlas for the rat central auditory system (Waxholm) is available. However, the Waxholm rat atlas only contains a low-resolution inner ear featuring five structures. In the present work, we segmented and annotated 34 structures in the rat inner ear, yielding a detailed volumetric inner ear atlas which can be integrated with the Waxholm rat brain atlas. We performed iodine-enhanced microCT and iDISCO+-based clearing and fluorescence lightsheet microscopy imaging on a sample of rat temporal bones. Image stacks were segmented in a semiautomated way, and 34 inner ear volumes were reconstructed from five samples. Using geometrical morphometry, high-resolution segmentations obtained from lightsheet and microCT stacks were registered into the coordinate system of the Waxholm rat atlas. Cleared sample autofluorescence was used for the reconstruction of most inner ear structures, including fluid-filled compartments, nerves and sensory epithelia, blood vessels, and connective tissue structures. Image resolution allowed reconstruction of thin ducts (reuniting, saccular and endolymphatic), and the utriculoendolymphatic valve. The vestibulocochlear artery coursing through bone was found to be associated to the reuniting duct, and to be visible both in cleared and microCT samples, thus allowing to infer duct location from microCT scans. Cleared labyrinths showed minimal shape distortions, as shown by alignment with microCT and Waxholm labyrinths. However, membranous labyrinths could display variable collapse of the superior division, especially the roof of canal ampullae, whereas the inferior division (saccule and cochlea) was well preserved, with the exception of Reissner's membrane that could display ruptures in the second cochlear turn. As an example of atlas use, the volumes reconstructed from segmentations were used to separate macrophage populations from the spiral ganglion, auditory neuron dendrites, and Organ of Corti. We have reconstructed 34 structures from the rat temporal bone, which are available as both image stacks and printable 3D objects in a shared repository for download. These can be used for teaching, localizing cells or other features within the ear, modeling auditory and vestibular sensory physiology and training of automated segmentation machine learning tools.

Brain tumors pose a significant medical challenge, necessitating early detection and precise classification for effective treatment. This study aims to address this challenge by introducing an automated brain tumor classification system that utilizes deep learning (DL) and Magnetic Resonance Imaging (MRI) images. The main purpose of this research is to develop a model that can accurately detect and classify different types of brain tumors, including glioma, meningioma, pituitary tumors, and normal brain scans. A convolutional neural network (CNN) architecture with pretrained VGG16 as the base model is employed, and diverse public datasets are utilized to ensure comprehensive representation. Data augmentation techniques are employed to enhance the training dataset, resulting in a total of 17,136 brain MRI images across the four classes. The accuracy of this model was 99.24%, a higher accuracy than other similar works, demonstrating its potential clinical utility. This higher accuracy was achieved mainly due to the utilization of a large and diverse dataset, the improvement of network configuration, the application of a fine-tuning strategy to adjust pretrained weights, and the implementation of data augmentation techniques in enhancing classification performance for brain tumor detection. In addition, a web application was developed by leveraging HTML and Dash components to enhance usability, allowing for easy image upload and tumor prediction. By harnessing artificial intelligence (AI), the developed system addresses the need to reduce human error and enhance diagnostic accuracy. The proposed approach provides an efficient and reliable solution for brain tumor classification, facilitating early diagnosis and enabling timely medical interventions. This work signifies a potential advancement in brain tumor classification, promising improved patient care and outcomes.

Quantitative MRI (qMRI) requires the acquisition of multiple images with parameter changes, resulting in longer measurement times than conventional imaging. Deep learning (DL) for image reconstruction has shown a significant reduction in acquisition time and improved image quality. In qMRI, where the image contrast varies between sequences, preparing large, fully-sampled (FS) datasets is challenging. Recently, methods that do not require FS data such as self-supervised learning (SSL) and zero-shot self-supervised learning (ZSSSL) have been proposed. Another challenge is the large GPU memory requirement for DL-based qMRI image reconstruction, owing to the simultaneous processing of multiple contrast images. In this context, Kellman et al. proposed memory-efficient learning (MEL) to save the GPU memory. This study evaluated SSL and ZSSSL frameworks with MEL to accelerate qMRI. Three experiments were conducted using the following sequences: 2D T2 mapping/MSME (Experiment 1), 3D T1 mapping/VFA-SPGR (Experiment 2), and 3D T2 mapping/DESS (Experiment 3). Each experiment used the undersampled k-space data under acceleration factors of 4, 8, and 12. The reconstructed maps were evaluated using quantitative metrics. In this study, we performed three qMRI reconstruction measurements and compared the performance of the SL- and GT-free learning methods, SSL and ZSSSL. Overall, the performances of SSL and ZSSSL were only slightly inferior to those of SL, even under high AF conditions. The quantitative errors in diagnostically important tissues (WM, GM, and meniscus) were small, demonstrating that SL and ZSSSL performed comparably. Additionally, by incorporating a GPU memory-saving implementation, we demonstrated that the network can operate on a GPU with a small memory (<8GB) with minimal speed reduction. This study demonstrates the effectiveness of memory-efficient GT-free learning methods using MEL to accelerate qMRI.