Supracondylar humerus fractures in children are among the most common elbow fractures in pediatrics. However, their diagnosis can be particularly challenging due to the anatomical characteristics and imaging features of the pediatric skeleton. In recent years, convolutional neural networks (CNNs) have achieved notable success in medical image analysis, though their performance typically relies on large-scale, high-quality labeled datasets. Unfortunately, labeled samples for pediatric supracondylar fractures are scarce and difficult to obtain. To address this issue, this paper introduces a deep learning-based multi-scale patch residual network (MPR) for the automatic detection and localization of subtle pediatric supracondylar fractures. The MPR framework combines a CNN for automatic feature extraction with a multi-scale generative adversarial network to model skeletal integrity using healthy samples. By leveraging healthy images to learn the normal skeletal distribution, the approach reduces the dependency on labeled fracture data and effectively addresses the challenges posed by limited pediatric datasets. Datasets from two different hospitals were used, with data augmentation techniques applied during both training and validation. On an independent test set, the proposed model achieves an accuracy of 90.5%, with 89% sensitivity, 92% specificity, and an F1 score of 0.906-outperforming the diagnostic accuracy of emergency medicine physicians and approaching that of pediatric radiologists. Furthermore, the model demonstrates a fast inference speed of 1.1 s per sheet, underscoring its substantial potential for clinical application.

To perform a systematic review on artificial intelligence (AI) studies focused on identifying and differentiating pelvic gynecological tumors on ultrasound scans. Studies developing or validating AI models for diagnosing gynecological pelvic tumors on ultrasound scans were eligible for inclusion. We systematically searched PubMed, Embase, Web of Science, and Cochrane Central from their database inception until April 30th, 2024. To assess the quality of the included studies, we adapted the QUADAS-2 risk of bias tool to address the unique challenges of AI in medical imaging. Using multi-level random effects models, we performed a meta-analysis to generate summary estimates of the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. To provide a reference point of current diagnostic support tools for ultrasound examiners, we descriptively compared the pooled performance to that of the well-recognized ADNEX model on external validation. Subgroup analyses were performed to explore sources of heterogeneity. From 9151 records retrieved, 44 studies were eligible: 40 on ovarian, three on endometrial, and one on myometrial pathology. Overall, 95% were at high risk of bias - primarily due to inappropriate study inclusion criteria, the absence of a patient-level split of training and testing image sets, and no calibration assessment. For ovarian tumors, the summary AUC for AI models distinguishing benign from malignant tumors was 0.89 (95% CI: 0.85-0.92). In lower-risk studies (at least three low-risk domains), the summary AUC dropped to 0.87 (0.83-0.90), with deep learning models outperforming radiomics-based machine learning approaches in this subset. Only five studies included an external validation, and six evaluated calibration performance. In a recent systematic review of external validation studies, the ADNEX model had a pooled AUC of 0.93 (0.91-0.94) in studies at low risk of bias. Studies on endometrial and myometrial pathologies were reported individually. Although AI models show promising discriminative performances for diagnosing gynecological tumors on ultrasound, most studies have methodological shortcomings that result in a high risk of bias. In addition, the ADNEX model appears to outperform most AI approaches for ovarian tumors. Future research should emphasize robust study designs - ideally large, multicenter, and prospective cohorts that mirror real-world populations - along with external validation, proper calibration, and standardized reporting. This study was pre-registered with Open Science Framework (OSF): https://doi.org/10.17605/osf.io/bhkst.

Early detection and accurate diagnosis are essential to improving patient outcomes. The use of convolutional neural networks (CNNs) for tumor detection has shown promise, but existing models often suffer from overparameterization, which limits their performance gains. In this study, fuzzy sigmoid convolution (FSC) is introduced along with two additional modules: top-of-the-funnel and middle-of-the-funnel. The proposed methodology significantly reduces the number of trainable parameters without compromising classification accuracy. A novel convolutional operator is central to this approach, effectively dilating the receptive field while preserving input data integrity. This enables efficient feature map reduction and enhances the model's tumor detection capability. In the FSC-based model, fuzzy sigmoid activation functions are incorporated within convolutional layers to improve feature extraction and classification. The inclusion of fuzzy logic into the architecture improves its adaptability and robustness. Extensive experiments on three benchmark datasets demonstrate the superior performance and efficiency of the proposed model. The FSC-based architecture achieved classification accuracies of 99.17%, 99.75%, and 99.89% on three different datasets. The model employs 100 times fewer parameters than large-scale transfer learning architectures, highlighting its computational efficiency and suitability for detecting brain tumors early. This research offers lightweight, high-performance deep-learning models for medical imaging applications.

Thoracolumbar stump ribs are one of the essential indicators of thoracolumbar transitional vertebrae or enumeration anomalies. While some studies manually assess these anomalies and describe the ribs qualitatively, this study aims to automate thoracolumbar stump rib detection and analyze their morphology quantitatively. To this end, we train a high-resolution deep-learning model for rib segmentation and show significant improvements compared to existing models (Dice score 0.997 vs. 0.779, p-value < 0.01). In addition, we use an iterative algorithm and piece-wise linear interpolation to assess the length of the ribs, showing a success rate of 98.2%. When analyzing morphological features, we show that stump ribs articulate more posteriorly at the vertebrae (-19.2 +- 3.8 vs -13.8 +- 2.5, p-value < 0.01), are thinner (260.6 +- 103.4 vs. 563.6 +- 127.1, p-value < 0.01), and are oriented more downwards and sideways within the first centimeters in contrast to full-length ribs. We show that with partially visible ribs, these features can achieve an F1-score of 0.84 in differentiating stump ribs from regular ones. We publish the model weights and masks for public use.

Background and ObjectivesBottom-of-sulcus dysplasia (BOSD) is a diagnostically challenging subtype of focal cortical dysplasia, 60% being missed on patients first MRI. Automated MRI-based detection methods have been developed for focal cortical dysplasia, but not BOSD specifically. Use of FDG-PET alongside MRI is not established in automated methods. We report the development and performance of an automated BOSD detector using combined MRI+PET data. MethodsThe training set comprised 54 mostly operated patients with BOSD. The test sets comprised 17 subsequently diagnosed patients with BOSD from the same center, and 12 published patients from a different center. 81% patients across training and test sets had reportedly normal first MRIs and most BOSDs were <1.5cm3. In the training set, 12 features from T1-MRI, FLAIR-MRI and FDG-PET were evaluated using a novel "pseudo-control" normalization approach to determine which features best distinguished dysplastic from normal-appearing cortex. Using the Multi-centre Epilepsy Lesion Detection groups machine-learning detection method with the addition of FDG-PET, neural network classifiers were then trained and tested on MRI+PET features, MRI-only and PET-only. The proportion of patients whose BOSD was overlapped by the top output cluster, and the top five output clusters, were assessed. ResultsCortical and subcortical hypometabolism on FDG-PET were superior in discriminating dysplastic from normal-appearing cortex compared to MRI features. When the BOSD detector was trained on MRI+PET features, 87% BOSDs were overlapped by one of the top five clusters (69% top cluster) in the training set, 76% in the prospective test set (71% top cluster) and 75% in the published test set (42% top cluster). Cluster overlap was similar when the detector was trained and tested on PET-only features but lower when trained and tested on MRI-only features. ConclusionDetection of BOSD is possible using established MRI-based automated detection methods, supplemented with FDG-PET features and trained on a BOSD-specific cohort. In clinical practice, an MRI+PET BOSD detector could improve assessment and outcomes in seemingly MRI-negative patients being considered for epilepsy surgery.

Accurate estimation of treatment response can help clinicians identify patients who would potentially benefit from systemic therapy. This study aimed to develop and externally validate a model for predicting treatment response to systemic therapy in advanced gallbladder cancer (GBC). We recruited 399 eligible GBC patients across four institutions. Multivariable logistic regression analysis was performed to identify independent clinical factors related to therapeutic efficacy. This deep learning (DL) radiomics signature was developed for predicting treatment response using multiphase enhanced CT images. Then, the DL radiomic-clinical (DLRSC) model was built by combining the DL signature and significant clinical factors, and its predictive performance was evaluated using area under the curve (AUC). Gradient-weighted class activation mapping analysis was performed to help clinicians better understand the predictive results. Furthermore, patients were stratified into low- and high-score groups by the DLRSC model. The progression-free survival (PFS) and overall survival (OS) between the two different groups were compared. Multivariable analysis revealed that tumor size was a significant predictor of efficacy. The DLRSC model showed great predictive performance, with AUCs of 0.86 (95% CI, 0.82-0.89) and 0.84 (95% CI, 0.80-0.87) in the internal and external test datasets, respectively. This model showed great discrimination, calibration, and clinical utility. Moreover, Kaplan-Meier survival analysis revealed that low-score group patients who were insensitive to systemic therapy predicted by the DLRSC model had worse PFS and OS. The DLRSC model allows for predicting treatment response in advanced GBC patients receiving systemic therapy. The survival benefit provided by the DLRSC model was also assessed. Question No effective tools exist for identifying patients who would potentially benefit from systemic therapy in clinical practice. Findings Our combined model allows for predicting treatment response to systemic therapy in advanced gallbladder cancer. Clinical relevance With the help of this model, clinicians could inform patients of the risk of potential ineffective treatment. Such a strategy can reduce unnecessary adverse events and effectively help reallocate societal healthcare resources.

The rapid evolution of large language models (LLMs) offers promising opportunities for radiology report annotation, aiding in determining the presence of specific findings. This study evaluates the effectiveness of a human-optimized prompt in labeling radiology reports across multiple institutions using LLMs. Six distinct institutions collected 500 radiology reports: 100 in each of 5 categories. A standardized Python script was distributed to participating sites, allowing the use of one common locally executed LLM with a standard human-optimized prompt. The script executed the LLM's analysis for each report and compared predictions to reference labels provided by local investigators. Models' performance using accuracy was calculated, and results were aggregated centrally. The human-optimized prompt demonstrated high consistency across sites and pathologies. Preliminary analysis indicates significant agreement between the LLM's outputs and investigator-provided reference across multiple institutions. At one site, eight LLMs were systematically compared, with Llama 3.1 70b achieving the highest performance in accurately identifying the specified findings. Comparable performance with Llama 3.1 70b was observed at two additional centers, demonstrating the model's robust adaptability to variations in report structures and institutional practices. Our findings illustrate the potential of optimized prompt engineering in leveraging LLMs for cross-institutional radiology report labeling. This approach is straightforward while maintaining high accuracy and adaptability. Future work will explore model robustness to diverse report structures and further refine prompts to improve generalizability.

In recent years, deformable medical image registration techniques have made significant progress. However, existing models still lack efficiency in parallel extraction of coarse and fine-grained features. To address this, we construct a new pyramid registration network based on feature and deformation field (FF-PNet). For coarse-grained feature extraction, we design a Residual Feature Fusion Module (RFFM), for fine-grained image deformation, we propose a Residual Deformation Field Fusion Module (RDFFM). Through the parallel operation of these two modules, the model can effectively handle complex image deformations. It is worth emphasizing that the encoding stage of FF-PNet only employs traditional convolutional neural networks without any attention mechanisms or multilayer perceptrons, yet it still achieves remarkable improvements in registration accuracy, fully demonstrating the superior feature decoding capabilities of RFFM and RDFFM. We conducted extensive experiments on the LPBA and OASIS datasets. The results show our network consistently outperforms popular methods in metrics like the Dice Similarity Coefficient.

BackgroundIn glioblastoma (GBM), promoter methylation of the O6-methylguanine-DNA methyltransferase (MGMT) is associated with beneficial chemotherapy but has not been accurately evaluated based on radiological and pathological sections. To develop and validate an MRI and pathology image-based deep learning radiopathomics model for predicting MGMT promoter methylation in patients with GBM. MethodsA retrospective collection of pathologically confirmed isocitrate dehydrogenase (IDH) wild-type GBM patients (n=207) from three centers was performed, all of whom underwent MRI scanning within 2 weeks prior to surgery. The pre-trained ResNet50 was used as the feature extractor. Features of 1024 dimensions were extracted from MRI and pathological images, respectively, and the features were screened for modeling. Then feature fusion was performed by calculating the normalized multimode MRI fusion features and pathological features, and prediction models of MGMT based on deep learning radiomics, pathomics, and radiopathomics (DLRM, DLPM, DLRPM) were constructed and applied to internal and external validation cohorts. ResultsIn the training, internal and external validation cohorts, the DLRPM further improved the predictive performance, with a significantly better predictive performance than the DLRM and DLPM, with AUCs of 0.920 (95% CI 0.870-0.968), 0.854 (95% CI 0.702-1), and 0.840 (95% CI 0.625-1). ConclusionWe developed and validated cross-scale radiology and pathology models for predicting MGMT methylation status, with DLRPM predicting the best performance, and this cross-scale approach paves the way for further research and clinical applications in the future.

Accurate estimation of cerebral amyloid-{beta} (A{beta}) burden is critical for early detection and risk stratification in Alzheimers disease (AD). While A{beta} positron emission tomography (PET) remains the gold standard, its high cost, invasive nature and limited accessibility hinder broad clinical application. Blood-based biomarkers offer a non-invasive and cost-effective alternative, but their standalone predictive accuracy remains limited due to biological heterogeneity and limited reflection of central nervous system pathology. Here, we present a high-precision, multimodal prediction machine learning model that integrates plasma biomarkers, brain structural magnetic resonance imaging (sMRI) features, diffusion tensor imaging (DTI)-derived structural connectomes, and genetic risk profiles. The model was trained on 150 participants from the Alzheimers Disease Neuroimaging Initiative (ADNI) and externally validated on 111 participants from the SILCODE cohort. Multimodal integration substantially improved A{beta} prediction, with R{superscript 2} increasing from 0.515 using plasma biomarkers alone to 0.637 when adding imaging and genetic features. These results highlight the potential of this multimodal machine learning approach as a scalable, non-invasive, and economically viable alternative to PET for estimating A{beta} burden.