Chest X-ray (CXR) imaging remains one of the most widely used diagnostic tools for detecting pulmonary diseases such as tuberculosis (TB) and pneumonia. Recent advances in deep learning, particularly Vision Transformers (ViTs), have shown strong potential for automated medical image analysis. However, most ViT architectures are pretrained on natural images and require three-channel inputs, while CXR scans are inherently grayscale. To address this gap, we propose RepViT-CXR, a channel replication strategy that adapts single-channel CXR images into a ViT-compatible format without introducing additional information loss. We evaluate RepViT-CXR on three benchmark datasets. On the TB-CXR dataset,our method achieved an accuracy of 99.9% and an AUC of 99.9%, surpassing prior state-of-the-art methods such as Topo-CXR (99.3% accuracy, 99.8% AUC). For the Pediatric Pneumonia dataset, RepViT-CXR obtained 99.0% accuracy, with 99.2% recall, 99.3% precision, and an AUC of 99.0%, outperforming strong baselines including DCNN and VGG16. On the Shenzhen TB dataset, our approach achieved 91.1% accuracy and an AUC of 91.2%, marking a performance improvement over previously reported CNN-based methods. These results demonstrate that a simple yet effective channel replication strategy allows ViTs to fully leverage their representational power on grayscale medical imaging tasks. RepViT-CXR establishes a new state of the art for TB and pneumonia detection from chest X-rays, showing strong potential for deployment in real-world clinical screening systems.

In lung CT imaging, motion artifacts caused by cardiac motion and respiration are common. Recently, CLEAR Motion, a deep learning-based reconstruction method that applies motion correction technology, has been developed. This study aims to quantitatively evaluate the clinical usefulness of CLEAR Motion. A total of 129 lung CT was analyzed, and heart rate, height, weight, and BMI of all patients were obtained from medical records. Images with and without CLEAR Motion were reconstructed, and quantitative evaluation was performed using variance of Laplacian (VL) and PSNR. The difference in VL (DVL) between the two reconstruction methods was used to evaluate which part of the lung field (upper, middle, or lower) CLEAR Motion is effective. To evaluate the effect of motion correction based on patient characteristics, the correlation between body mass index (BMI), heart rate and DVL was determined. Visual assessment of motion artifacts was performed using paired comparisons by 9 radiological technologists. With the exception of one case, VL was higher in CLEAR Motion. Almost all the cases (110 cases) showed large DVL in the lower part. BMI showed a positive correlation with DVL (r = 0.55, p < 0.05), while no differences in DVL were observed based on heart rate. The average PSNR was 35.8 ± 0.92 dB. Visual assessments indicated that CLEAR Motion was preferred in most cases, with an average preference score of 0.96 (p < 0.05). Using Clear Motion allows for obtaining images with fewer motion artifacts in lung CT.

Precise preoperative discrimination of invasive lung adenocarcinoma (IA) from preinvasive lesions (adenocarcinoma in situ [AIS]/minimally invasive adenocarcinoma [MIA]) and prediction of high-risk histopathological features are critical for optimizing resection strategies in early-stage lung adenocarcinoma (LUAD). In this multicenter study, 813 LUAD patients (tumors ≤3 cm) formed the training cohort. A total of 1,709 radiomic features were extracted from the PET/CT images. Feature selection was performed using the max-relevance and min-redundancy (mRMR) algorithm and least absolute shrinkage and selection operator (LASSO). Hybrid machine learning models integrating [18F]FDG PET/CT radiomics and clinical-radiological features were developed using H2O.ai AutoML. Models were validated in a prospective internal cohort (N = 256, 2021-2022) and external multicenter cohort (N = 418). Performance was assessed via AUC, calibration, decision curve analysis (DCA) and survival assessment. The hybrid model achieved AUCs of 0.93 (95% CI: 0.90-0.96) for distinguishing IA from AIS/MIA (internal test) and 0.92 (0.90-0.95) in external testing. For predicting high-risk histopathological features (grade-III, lymphatic/pleural/vascular/nerve invasion, STAS), AUCs were 0.82 (0.77-0.88) and 0.85 (0.81-0.89) in internal/external sets. DCA confirmed superior net benefit over CT model. The model stratified progression-free (P = 0.002) and overall survival (P = 0.017) in the TCIA cohort. PET/CT radiomics-based models enable accurate non-invasive prediction of invasiveness and high-risk pathology in early-stage LUAD, guiding optimal surgical resection.

Lung adenocarcinoma remains a leading cause of cancer-related mortality, and the diagnostic performance of computed tomography (CT) is limited when dependent solely on human interpretation. This study aimed to develop and evaluate an interpretable deep learning framework using an attention-enhanced Squeeze-and-Excitation Residual Network (SE-ResNet) to improve automated classification of lung adenocarcinoma from thoracic CT images. Furthermore, Gradient-weighted Class Activation Mapping (Grad-CAM) was applied to enhance model interpretability and assist in the visual localization of tumor regions. A total of 3800 chest CT axial slices were collected from 380 subjects (190 patients with lung adenocarcinoma and 190 controls, with 10 slices extracted from each case). This dataset was used to train and evaluate the baseline ResNet50 model as well as the proposed SE-ResNet50 model. Performance was compared using accuracy, Area Under the Curve (AUC), precision, recall, and F1-score. Grad-CAM visualizations were generated to assess the alignment between the model's attention and radiologically confirmed tumor locations. The SE-ResNet model achieved a classification accuracy of 94% and an AUC of 0.941, significantly outperforming the baseline ResNet50, which had an 85% accuracy and an AUC of 0.854. Grad-CAM heatmaps produced from the SE-ResNet demonstrated superior localization of tumor-relevant regions, confirming the enhanced focus provided by the attention mechanism. The proposed SE-ResNet framework delivers high accuracy and interpretability in classifying lung adenocarcinoma from CT images. It shows considerable potential as a decision-support tool to assist radiologists in diagnosis and may serve as a valuable clinical tool with further validation.

The variability and biases in the real-world performance benchmarking of deep learning models for medical imaging compromise their trustworthiness for real-world deployment. The common approach of holding out a single fixed test set fails to quantify the variance in the estimation of test performance metrics. This study introduces NACHOS (Nested and Automated Cross-validation and Hyperparameter Optimization using Supercomputing) to reduce and quantify the variance of test performance metrics of deep learning models. NACHOS integrates Nested Cross-Validation (NCV) and Automated Hyperparameter Optimization (AHPO) within a parallelized high-performance computing (HPC) framework. NACHOS was demonstrated on a chest X-ray repository and an Optical Coherence Tomography (OCT) dataset under multiple data partitioning schemes. Beyond performance estimation, DACHOS (Deployment with Automated Cross-validation and Hyperparameter Optimization using Supercomputing) is introduced to leverage AHPO and cross-validation to build the final model on the full dataset, improving expected deployment performance. The findings underscore the importance of NCV in quantifying and reducing estimation variance, AHPO in optimizing hyperparameters consistently across test folds, and HPC in ensuring computational feasibility. By integrating these methodologies, NACHOS and DACHOS provide a scalable, reproducible, and trustworthy framework for DL model evaluation and deployment in medical imaging. To maximize public availability, the full open-source codebase is provided at https://github.com/thepanlab/NACHOS.

BACKGROUND This study used CT imaging analyzed with deep learning techniques to assess the diagnostic accuracy of lung metastasis detection in patients with breast cancer. The aim of the research was to create and verify a system for detecting malignant and metastatic lung lesions that uses YOLOv10 and transfer learning. MATERIAL AND METHODS From January 2023 to 2024, CT scans of 16 patients with breast cancer who had confirmed lung metastases were gathered retrospectively from Erzincan Mengücek Gazi Training and Research Hospital. The YOLOv10 deep learning system was used to assess a labeled dataset of 1264 enhanced CT images. RESULTS A total of 1264 labeled images from 16 patients were included. With an accuracy of 96.4%, sensitivity of 94.1%, specificity of 97.1%, and precision of 90.3%, the ResNet-50 model performed best. The robustness of the model was shown by the remarkable area under the curve (AUC), which came in at 0.96. After dataset tuning, the GoogLeNet model's accuracy was 97.3%. These results highlight our approach's improved diagnostic capabilities over current approaches. CONCLUSIONS This study shows how YOLOv10 and transfer learning can be used to improve the diagnostic precision of pulmonary metastases in patients with breast cancer. The model's effectiveness is demonstrated by the excellent performance metrics attained, opening the door for its application in clinical situations. The suggested approach supports prompt and efficient treatment decisions by lowering radiologists; workload and improving the early diagnosis of metastatic lesions.

Deep neural networks excel in radiological image classification but frequently suffer from poor interpretability, limiting clinical acceptance. We present MedicalPatchNet, an inherently self-explainable architecture for chest X-ray classification that transparently attributes decisions to distinct image regions. MedicalPatchNet splits images into non-overlapping patches, independently classifies each patch, and aggregates predictions, enabling intuitive visualization of each patch's diagnostic contribution without post-hoc techniques. Trained on the CheXpert dataset (223,414 images), MedicalPatchNet matches the classification performance (AUROC 0.907 vs. 0.908) of EfficientNet-B0, while substantially improving interpretability: MedicalPatchNet demonstrates substantially improved interpretability with higher pathology localization accuracy (mean hit-rate 0.485 vs. 0.376 with Grad-CAM) on the CheXlocalize dataset. By providing explicit, reliable explanations accessible even to non-AI experts, MedicalPatchNet mitigates risks associated with shortcut learning, thus improving clinical trust. Our model is publicly available with reproducible training and inference scripts and contributes to safer, explainable AI-assisted diagnostics across medical imaging domains. We make the code publicly available: https://github.com/TruhnLab/MedicalPatchNet

In non-small cell lung cancer (NSCLC), non-invasive alternatives to biopsy-dependent driver mutation analysis are needed. We reviewed the effectiveness of radiomics alone or with clinical data and assessed the performance of artificial intelligence (AI) models in predicting oncogene mutation status. A PRISMA-compliant literature review for studies predicting oncogene mutation status in NSCLC patients using radiomics was conducted by a multidisciplinary team. Meta-analyses evaluating the performance of AI-based models developed with CT-derived radiomics features alone or combined with clinical data were performed. A meta-regression to analyze the influence of different predictors was also conducted. Of 890 studies identified, 124 evaluating models for the prediction of epidermal growth factor-1 (EGFR), anaplastic lymphoma kinase (ALK), and Kirsten rat sarcoma virus (KRAS) mutations were included in the systematic review, of which 51 were meta-analyzed. The AI algorithms' sensitivity/false positive rate (FPR) in predicting mutation status using radiomics-based models was 0.754 (95% CI 0.727-0.780)/0.344 (95% CI 0.308-0.381) for EGFR, 0.754 (95% CI 0.638-0.841)/0.225 (95% CI 0.163-0.302) for ALK and 0.475 (95% CI 0.153-0.820)/0.181 (95% CI 0.054-0.461) for KRAS. A meta-analysis of combined models was possible for EGFR mutation, revealing a sensitivity of 0.806 (95% CI 0.777-0.833) and a FPR of 0.315 (95% CI 0.270-0.364). No statistically significant results were obtained in the meta-regression. Radiomics-based models may offer a non-invasive alternative for determining oncogene mutation status in NSCLC. Further research is required to analyze whether clinical data might boost their performance. Question Can imaging-based radiomics and artificial intelligence non-invasively predict oncogene mutation status to improve diagnosis in non-small cell lung cancer (NSCLC)? Findings Radiomics-based models achieved high performance in predicting mutation status in NSCLC; adding clinical data showed limited improvement in predictive performance. Clinical relevance Radiomics and AI tools offer a non-invasive strategy to support molecular profiling in NSCLC. Validation studies addressing clinical and methodological aspects are essential to ensure their reliability and integration into routine clinical practice.

To establish an explainable machine learning (ML) approach using patient-related and noncontrast chest CT-derived features to predict the contrast material arrival time (TARR) in CT pulmonary angiography (CTPA). This retrospective study included consecutive patients referred for CTPA between September 2023 to October 2024. Sixteen clinical and 17 chest CT-derived parameters were used as inputs for the ML approach, which employed recursive feature elimination for feature selection and XGBoost with SHapley Additive exPlanations (SHAP) for explainable modeling. The prediction target was abnormal TARR of the pulmonary artery (ie, TARR <7 seconds or >10 s), determined by the time to peak enhancement in the test bolus, with 2 models distinguishing these cases. External validation was conducted. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC). A total of 666 patients (mean age, 70 [IQR, 59.3 to 78.0]; 46.8% female participants) were split into training (n = 353), testing (n = 151), and external validation (n = 162) sets. 86 cases (12.9%) had TARR <7 seconds, and 138 cases (20.7%) had TARR >10 seconds. The ML models exhibited good performance in their respective testing and external validation sets (AUC: 0.911 and 0.878 for TARR <7 s; 0.834 and 0.897 for TARR >10 s). SHAP analysis identified the measurements of the vena cava and pulmonary artery as key features for distinguishing abnormal TARR. The explainable ML algorithm accurately identified normal and abnormal TARR of the pulmonary artery, facilitating personalized CTPA scans.

Point-of-care ultrasonography has become a valuable tool for assessing diaphragmatic function in critically ill patients receiving invasive mechanical ventilation. However, conventional diaphragm ultrasound assessment remains highly operator-dependent and subjective. Previous research introduced automatic measurement of diaphragmatic excursion and velocity using 2D speckle-tracking technology. This study aimed to develop an artificial intelligence-multimodal learning framework to improve the prediction of weaning failure and guide individualized weaning strategies. This prospective study enrolled critically ill patients older than 18 years who received mechanical ventilation for more than 48 hours and were eligible for a spontaneous breathing trial in 2 intensive care units in Guangzhou, China. Before the spontaneous breathing trial, diaphragm ultrasound videos were collected using a standardized protocol, and automatic measurements of excursion and velocity were obtained. A total of 88 patients were included, with 50 successfully weaned and 38 experiencing weaning failure. Each patient record included 27 clinical and 6 diaphragmatic indicators, selected based on previous literature and phenotyping studies. Clinical variables were preprocessed using OneHotEncoder, normalization, and scaling. Ultrasound videos were interpolated to a uniform resolution of 224×224×96. Artificial intelligence-multimodal learning based on clinical characteristics, laboratory parameters, and diaphragm ultrasonic videos was established. Four experiments were conducted in an ablation setting to evaluate model performance using different combinations of input data: (1) diaphragmatic excursion only, (2) clinical and diaphragmatic indicators, (3) ultrasound videos only, and (4) all modalities combined (multimodal). Metrics for evaluation included classification accuracy, area under the receiver operating characteristic curve (AUC), average precision in the precision-recall curve, and calibration curve. Variable importance was assessed using SHAP (Shapley Additive Explanation) to interpret feature contributions and understand model predictions. The multimodal co-learning model outperformed all single-modal approaches. The accuracy improved when predicted through diaphragm ultrasound video data using Video Vision Transformer (accuracy=0.8095, AUC=0.852), clinical or ultrasound indicators (accuracy=0.7381, AUC=0.746), and the multimodal co-learning (accuracy=0.8331, AUC=0.894). The proposed co-learning model achieved the highest score (average precision=0.91) among the 4 experiments. Furthermore, calibration curve analysis demonstrated that the proposed colearning model was well calibrated, as the curve was closest to the perfectly calibrated line. Combining ultrasound and clinical data for colearning improved the accuracy of the weaning outcome prediction. Multimodal learning based on automatic measurement of point-of-care ultrasonography and automated collection of objective clinical indicators greatly enhanced the practical operability and user-friendliness of the system. The proposed model offered promising potential for widespread clinical application in intensive care settings.