To propose a multi-parametric ultrasound imaging-based deep learning method for accurately classifying metastatic and non-metastatic axillary lymph nodes in breast cancer patients. The proposed method integrates the conventional ultrasound B-mode imaging with shear wave elastography and color Doppler images of 174 patients to train a transfer learning-based network comprising pretrained MobileNetv2 with a custom shallow head consisting of a convolutional neural network with mixed pooling, weighted sum mixed pooling and squeeze-and-excite attention mechanisms for the first time in the context of ALN classification. The proposed method was evaluated using five-fold cross-validation, achieving a mean classification accuracy of 0.91, specificity of 0.91, sensitivity of 0.93, F1 score of 0.93, area under the precision-recall curve of 0.94, and a cross-validated AUC (cvAUC) of 0.92. A network ablation study confirmed the robustness of the model, with relatively narrow 95% confidence intervals (CIs) for cvAUC. Comparative analysis showed that the proposed network (Acc: 0.91) outperformed state-of-the-art deep learning models (Acc: 0.67-0.88) for ALN classification and exhibited narrower CIs, highlighting its relative stability. Additionally, results demonstrated that multi-parametric imaging significantly enhanced classification performance, reducing the 95% CI width by nearly half compared to uni-parametric data, further supporting the method's robustness and reliability. The integration of multi-parametric ultrasound imaging with deep learning network can remarkably improve the classification of metastatic and non-metastatic ALNs in breast cancer patients.

This study developed a multiomics model combining radiomics, pathomics, and temporal imaging to predict major pathological response in patients with locally advanced non-small cell lung cancer (NSCLC) undergoing neoadjuvant immunochemotherapy. A retrospective, multicenter study was conducted, enrolling 271 patients with stage IB-III NSCLC who received neoadjuvant immunochemotherapy. High-resolution CT images were enhanced using a generative adversarial network-based super-resolution technique. Radiomics features were extracted from multi-sequence CT scans at multiple time points, while pathomics features were derived from whole-slide imaging of surgical specimens. A transformer-based attention mechanism was used to integrate radiomics, pathomics, and temporal imaging data. The model was trained and validated on data from one center and tested on external cohorts. Performance was evaluated using area under the curve (AUC), net reclassification improvement, integrated discrimination improvement, and decision curve analysis. The Trans-Model demonstrated superior predictive performance, achieving an AUC of 0.858 (95% CI 0.783 to 0.933) in the external test cohort. It outperformed Rad-Model (AUC: 0.839) and Patho-Model (AUC: 0.753). The Trans-Model effectively stratified patients by survival outcomes, with major pathological response (MPR)-positive patients exhibiting significantly improved 3-year overall survival (87.3% vs 76.1%, p=0.034) and 5-year progression-free survival (45.8% vs 34.7%, p=0.033) compared with MPR-negative patients. Decision curve analysis confirmed the model's clinical utility across a wide range of threshold probabilities. The multiomics model, integrating multi-temporal, multi-sequence data with attention-based feature fusion, improves MPR prediction in patients with NSCLC receiving neoadjuvant immunochemotherapy, enabling personalized treatment by identifying responders and optimizing outcomes.

Thrombosis, the formation of blood clots within blood vessels, poses serious health risks including pulmonary embolism and post-thrombotic syndrome. Ultrasound (US) imaging is a widely used, non-invasive diagnostic tool owing to its real-time capability and safety profile; however, its effectiveness is often limited by operator dependency and variability in interpretation. This scoping review investigates how deep learning (DL) techniques have been applied to enhance thrombosis detection and risk assessment using US imaging across venous, arterial, and cardiac contexts. A comprehensive literature search was conducted in PubMed and Scopus following PRISMA-ScR methodology, targeting studies that used DL models for thrombus detection, classification, segmentation, or risk prediction in conjunction with vascular US modalities such as B-mode, Doppler, intravascular ultrasound (IVUS), and transesophageal echocardiography (TEE). Out of 233 records initially identified, 22 studies met the eligibility criteria. The most frequently used models included convolutional neural networks (CNNs), U-Net, Residual Neural Networks (ResNet), and Artificial Neural Networks (ANNs). DL models mainly aided deep vein thrombosis (DVT) diagnosis by evaluating vein compressibility and supporting point-of-care ultrasound (POCUS) imaging. Arterial thrombosis applications focused on plaque segmentation and vessel reconstruction using IVUS, while cardiac studies employed TEE to differentiate thrombi from tumours. Studies often reported high sensitivity, specificity, accuracy, and area under the curve (AUC), frequently outperforming traditional rule-based or manual interpretation methods, although considerable variability in datasets and validation approaches was observed. Overall, DL-enhanced US imaging shows great promise for improving diagnostic precision and clinical decision-making in thrombosis care. Future research should prioritize model interpretability, real-world integration, and the development of standardized, publicly accessible datasets.

Lung ultrasound (LUS) is an essential tool for diagnosing lung diseases. However, its effectiveness is often limited by its reproducibility, making interpretation challenging for clinicians. LUS diagnosis typically relies on subjective assessments of pleural line and vertical artifacts. To address this limitation, we introduce a novel quantitative approach aimed at reducing the need to rely on human operators (HOs) for LUS data assessment (i.e., improving the reproducibility). In the first phase of our study, we propose a hybrid method that integrates motion estimation and K-means clustering for automated segmentation of LUS images. The technique utilizes K-means clustering to identify pleural line based on intensity variations, while motion estimation detects vertical artifacts by analyzing motion vectors between consecutive frames. Rather than employing a conventional learning-based classification model, we develop an interpretable scoring framework that assigns scores to individual video frames according to standard scoring criteria. A threshold-based approach is then applied to aggregate frame-level scores, determining the final score for each video. We evaluated our method on a clinical dataset comprising 420 neonatal LUS videos from 70 patients, with annotations provided by three HOs. When using the majority vote among HOs as the reference standard, our method achieved a video-level accuracy of 0.72. For cases with full agreement among HOs, accuracy improved to 0.77. These results demonstrate that our approach offers comparable or superior performance to state-of-the-art deep learning (DL)-based methods in terms of scoring consistency, while reducing the need for a huge training dataset.

Peritoneal metastases (PM) significantly impact treatment options and prognosis of patients with cancer. Early detection and accurate evaluation are essential for guiding clinical decisions. This systematic review and meta-analysis aimed to provide a comprehensive overview and evaluate the performance of Artificial Intelligence (AI) and radiomics models for diagnosis and prognosis of PM on imaging. A systematic search of PubMed, Embase, and the Cochrane Library was conducted for studies published up to July 2024 that evaluated AI or radiomics models analyzing imaging data for diagnosing or predicting prognosis in PM. Data were extracted, and if more than 3 studies evaluated the same endpoint and reported true/false positive and negative values, a meta-analysis was conducted to obtain pooled area under the curve (AUC), sensitivity, and specificity. Bias was assessed using the PROBAST + AI tool. This review included 24 studies, of which 18 evaluated PM presence, 2 assessed PM severity (low versus high Peritoneal Cancer Index (PCI)), and 4 focused on prognosis or treatment efficacy. Meta-analysis of 13 studies evaluating PM presence revealed a pooled AUC of 0.84, sensitivity of 0.75, and specificity of 0.80. Subgroup analysis indicated comparable performance for 2D and 3D imaging data, and lower performance for models detecting occult PM compared to all PM presentations. Incorporating clinical factors into AI and radiomics models improved performance. AI and radiomics models demonstrated promising performance outcomes for PM evaluation on imaging, showing potential to aid in diagnosis and prognosis prediction. However, large validation studies are needed to evaluate their effects in clinical practice.

We characterized, in patients with severe acute traumatic spinal cord injuries, the relationships between intraoperative spinal cord blood flow (SCBF) and postoperative injury-site metabolism and physiology, preoperative magnetic resonance imaging (MRI) features, and neurological outcome. Twenty-six adults with severe, acute traumatic spinal cord injuries (American Spinal Injury Association Impairment Scale, grades A-C) had surgery within 72 h of injury. All had preoperative spine MRI and intraoperative laser speckle contrast imaging of SCBF. For four days after operation, we monitored from the injury site, intraspinal pressure (ISP), and spinal cord perfusion pressure (SCPP) as well as tissue metabolism with surface microdialysis. We observed three intraoperative SCBF patterns: necrosis-penumbra SCBF (SCBF-necr) in 34.6% of patients, patchy-perfusion SCBF (SCBF-patchy) in 38.5% of patients, and hyperperfusion SCBF (SCBF-hyper) in 26.9% of patients. On preoperative MRI, SCBF-necr was associated with higher Brain and Spinal Injury Center MRI score versus SCBF-patchy or SCBF-hyper (median 4 vs. 2 or 2.5). SCBF-necr was associated with higher postoperative ISP, lower postoperative SCPP, and more deranged postoperative injury-site metabolism (lower glucose; higher lactate, glutamate, and glycerol) than SCBF-patchy or SCBF-hyper, with little difference between SCBF-patchy and SCBF-hyper. Machine learning analysis of physiological-metabolic data considered as seven-dimensional vectors (ISP, SCPP, glucose, pyruvate, lactate, glutamate, and glycerol) accurately distinguished between the three SCBF patterns with an area under the curve of 0.85-0.95. The seven-dimensional physiological-metabolic vectors were segregated as SCBF-necr, SCBF-patchy, and SCBF-hyper in Kohonen self-organizing maps. SCBF-patchy was associated with greater improvement in motor score than SCBF-necr or SCBF-hyper (35.3 vs. 5.2 or 2.2), independent of admission American Spinal Injury Association Impairment Scale grade. Our findings challenge the prevailing concept in the field, derived from animal experiments, that spinal cord injury causes necrosis at the injury site with surrounding penumbra. In humans, spinal cord injury causes three abnormal SCBF patterns detected intraoperatively, with distinct postoperative physiological-metabolic signatures, preoperative MRI characteristics, and neurological outcomes.

Epicardial adipose tissue is gaining increasing interest as a cardiometabolic imaging biomarker, but its exact role in coronary artery disease is not fully understood. This study aimed to investigate the relationship between epicardial adipose tissue, coronary plaque characteristics, and risk of myocardial infarction in patients with suspected coronary artery disease, and in those with concomitant cardiometabolic disease. In a post-hoc analysis of the SCOT-HEART trial, epicardial adipose tissue volume and attenuation were quantified automatically from computed tomography (CT) angiography using deep-learning. Quantitative and high-risk coronary plaque characteristics were also assessed. The primary endpoint was fatal or non-fatal myocardial infarction. The study population consisted of 1770 patients (58 ± 9 years, 56% males) of whom 313 (18%) with cardiometabolic disease. Epicardial adipose tissue volume was higher in patients withcardiometabolic disease (123 ± 44 versus 88 ± 36 mL, p < 0.001), and increased with the coronary calcium score (0: 82 ± 35 mL, 1-400: 97 ± 38 mL, > 400: 113 ± 44 mL; p < 0.001), and low-attenuation plaque burden (burden ≤ 4%: 85 ± 36mL, burden > 4%: 103 ± 41mL; p < 0.001), while there were no interactions between these features and epicardial adipose tissue attenuation (p > 0.05 for all). During a median follow-up of 8.6 years, 82 (4.6%) patients experienced myocardial infarction. In the total study cohort, epicardial adipose tissue volume predicted myocardial infarction both in univariable analysis, and after adjustment for established markers of cardiovascular risk. In patients with cardiometabolic disease, epicardial adipose tissue volume independently predicted myocardial infarction after adjustment for clinical risk factors and plaque features but this relationship was not found in those without cardiometabolic disease. CT-derived Epicardial adipose tissue volume correlates with quantitative and high-risk plaque features, and independently predicts risk of myocardial infarction in patients with cardiometabolic disease.

Breast cancer remains a leading cause of mortality in women worldwide, with notable disparities in incidence and prognosis across regions. This systematic review explores the application of Deep Learning-based computer-aided diagnostic (CAD) systems for breast cancer detection, with a special focus on Asia to highlight underrepresented perspectives and challenges. We conducted a systematic Literature review in accordance with PRISMA guidelines. A comprehensive search of Scopus and Web of Science databases was performed to identify relevant studies published between January 2018 and November 2023, with an additional hand search for recent studies from 2024 to 2025. After screening 1051 records, 287 articles were included based on predefined inclusion and exclusion criteria. Quality assessment focused on the relevance of deep learning-based approaches to mammographic breast cancer detection, emphasizing global research trends and focused analysis of studies involving Asian populations. The review identified major research trends in deep learning-based mammographic analysis, with most studies focusing on lesion classification while comparatively fewer addressed detection, segmentation, and breast density assessment. Studies using Asian datasets revealed unique challenges, including higher breast density, limited annotations, and under-representation in public datasets. Analysis of methodologies highlighted varied use of image preprocessing and augmentation techniques. Focus maps were used to visualize contributions across tasks and populations, revealing gaps in multi-class BI-RADS classification and a global research bias toward Caucasian datasets (> 80%). This review reveals that most deep learning models for breast cancer detection are trained predominantly on Caucasian datasets, creating significant limitations when applied to other populations due to demographic differences in breast density and imaging characteristics. To improve breast cancer screening globally, researchers must develop deep learning systems using diverse datasets that represent different populations, validate these models across various ethnic groups, and ensure clinical testing includes women from multiple demographic backgrounds. PROSPERO CRD 42,023,478,896.

Microvascular invasion (MVI) is of great significance for the individualized treatment of hepatocellular carcinoma (HCC) and preoperative noninvasive prediction of MVI is still an urgent clinical problem. To explore the effects of different regions of interest (ROI) and image input dimensions on the performance of deep learning (DL) models, and to select the best result to develop and validate a DL model for preoperative prediction of MVI. A total of 206 patients with pathologically confirmed HCC from three hospitals were retrospectively enrolled and divided into training, internal validation and external test set. Based on hepatobiliary phase images (HBP) of gadoxetic acid-enhanced MRI, 2D DL, 3D DL and 2.5D deep multi-instance learning (MIL) models were established. The receiver operating characteristic curve (ROC) was used to evaluate the predictive efficacy of the above models. Based on the optimal performance model, the T1WI-FS and T2WI-FS images were preprocessed correspondingly, and a multimodal prediction model including three sequences was constructed. The ROC, and decision curve were used to visualize the predictive ability of the model. Compared with 2D DL and 3D DL models, the 2.5D DL model based on all axial images of ROI had the highest performance, with the AUC values of 0.802 (95% CI, 0.669-0.936) and 0.759 (95% CI, 0.643-0.875) in the validation and test sets. The AUCs of the multimodal MRI model were 0.954 (95% CI, 0.920-0.989) in the training set, 0.857 (95% CI, 0.736-0.978) in the validation set, and 0.788 (95% CI, 0.681-0.895) in the test set. The DL model that selects all axial slices of intratumor and peritumor as input shows robust capability in predicting MVI, which is expected to help clinical decision-making of individualized treatment for HCC.

This study evaluates radiomics correlation with mortality and suitability as prognostic indicator for troponin for pulmonary embolism to enhance prognostic accuracy and guide personalized treatment strategies with the help of machine learning. We conducted an initial study focusing on texture information of the arterial thrombus. Computed tomography (CT) of the lung from 86 patients with pulmonary embolism was used. As target variables, we used patients 30-day mortality and troponin results. Each arterial thrombus was manually segmented. After the extraction of their radiomics features and the reduction via correlation analysis and 12 feature selection methods, these and the target variables were given to 12 different classification methods to record the accuracies (Acc.), F1-scores (F1) and ROC curve areas under the curve (AUC) for comparison and evaluation. The resulting accuracy achieved was 0.967, the F1-score 0.973 for class 0 and 0.967 for class 1 and the AUC around 0.9686. The feature selection methods which resulted in the highest results were ReliefF (RF), Logistic Regression (LOR) and CART Classification (CARTC). For the classification methods, Support Vector Machines (SVM), eXtreme Gradiant Boosting (XGB) and Ensemble Bagged Trees (EBT) lead to the highest results. Firstorder, Shape and gray-level co-occurrence matrix (GLCM) were the most selected radiomics feature classes. Within this study, we conducted radiomics feature extraction within a medical image data analysis pipeline with subsequent correlation analysis and training of classifiers for patients with pulmonary lung embolism. We could show that the radiomics features correlated with patient's morphology as well as troponin range with an accuracy of 0.967 and 0.9302, respectively, yield high potential for prognosis and treatment strategy of pulmonary embolism patients in the future.