Accurate detection of cortical infarct is critical for timely treatment and improved patient outcomes. Current brain imaging methods often require invasive procedures that primarily assess blood vessel and structural white matter damage. There is a need for non-invasive approaches, such as functional MRI (fMRI), that better reflect neuronal viability. This study utilized automated machine learning (auto-ML) techniques to identify novel infarct-specific fMRI biomarkers specifically related to chronic cortical infarcts. We analyzed resting-state fMRI data from the multi-center ADNI dataset, which included 20 chronic infarct patients and 30 cognitively normal (CN) controls. This study utilized automated machine learning (auto-ML) techniques to identify novel fMRI biomarkers specifically related to chronic cortical infarcts. Surface-based registration methods were applied to minimize partial-volume effects typically associated with lower resolution fMRI data. We evaluated the performance of 7 previously known fMRI biomarkers alongside 107 new auto-generated fMRI biomarkers across 33 different classification models. Our analysis identified 6 new fMRI biomarkers that substantially improved infarct detection performance compared to previously established metrics. The best-performing combination of biomarkers and classifiers achieved a cross-validation ROC score of 0.791, closely matching the accuracy of diffusion-weighted imaging methods used in acute stroke detection. Our proposed auto-ML fMRI infarct-detection technique demonstrated robustness across diverse imaging sites and scanner types, highlighting the potential of automated feature extraction to significantly enhance non-invasive infarct detection.

Brain age has emerged as a powerful tool to understand neuroanatomical aging and its link to health outcomes like cognition. However, there remains a lack of studies investigating the rate of brain aging and its relationship to cognition. Furthermore, most brain age models are trained and tested on cross-sectional data from primarily Caucasian, adult participants. It is thus unclear how well these models generalize to non-Caucasian participants, especially children. Here, we tested a previously published deep learning model on Singaporean elderly participants (55-88 years old) and children (4-11 years old). We found that the model directly generalized to the elderly participants, but model finetuning was necessary for children. After finetuning, we found that the rate of change in brain age gap was associated with future executive function performance in both elderly participants and children. We further found that lateral ventricles and frontal areas contributed to brain age prediction in elderly participants, while white matter and posterior brain regions were more important in predicting brain age of children. Taken together, our results suggest that there is potential for generalizing brain age models to diverse populations. Moreover, the longitudinal change in brain age gap reflects developing and aging processes in the brain, relating to future cognitive function.

High tumor recurrence after surgery remains a significant challenge in managing hepatocellular carcinoma (HCC). We aimed to construct a multimodal model to forecast the early recurrence of HCC after surgical resection and explore the associated biological mechanisms. Overall, 519 patients with HCC were included from three medical centers. 433 patients from Nanfang Hospital were used as the training cohort, and 86 patients from the other two hospitals comprised validation cohort. Radiomics and deep learning (DL) models were developed using contrast-enhanced computed tomography images. Radiomics feature visualization and gradient-weighted class activation mapping were applied to improve interpretability. A multimodal model (MM-RDLM) was constructed by integrating radiomics and DL models. Associations between MM-RDLM and recurrence-free survival (RFS) and overall survival were analyzed. Gene set enrichment analysis (GSEA) and multiplex immunohistochemistry (mIHC) were used to investigate the biological mechanisms. Models based on hepatic arterial phase images exhibited the best predictive performance, with radiomics and DL models achieving areas under the curve (AUCs) of 0.770 (95 % confidence interval [CI]: 0.725-0.815) and 0.846 (95 % CI: 0.807-0.886), respectively, in the training cohort. MM-RDLM achieved an AUC of 0.955 (95 % CI: 0.937-0.972) in the training cohort and 0.930 (95 % CI: 0.876-0.984) in the validation cohort. MM-RDLM (high vs. low) was notably linked to RFS in the training (hazard ratio [HR] = 7.80 [5.74 - 10.61], P < 0.001) and validation (HR = 10.46 [4.96 - 22.68], P < 0.001) cohorts. GSEA revealed enrichment of the natural killer cell-mediated cytotoxicity pathway in the MM-RDLM low cohort. mIHC showed significantly higher percentages of CD3-, CD56-, and CD8-positive cells in the MM-RDLM low group. The MM-RDLM model demonstrated strong predictive performance for early postoperative recurrence of HCC. These findings contribute to identifying patients at high risk for early recurrence and provide insights into the potential underlying biological mechanisms.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes Coronavirus Disease 2019 (COVID-19) and induces activation of inflammatory pathways, including the inflammasome. The aim was to construct Machine Learning (ML) models to predict critical disease and death in patients with COVID-19. A total of 528 individuals with SARS-CoV-2 infection were included, comprising 308 with critical and 220 with non-critical COVID-19. The ML models included imaging, demographic, inflammatory biomarkers, NLRP3 (rs10754558 and rs10157379) and IL18 (rs360717 and rs187238) inflammasome variants. Individuals with critical COVID-19 were older, higher male/female ratio, body mass index (BMI), rate of type 2 diabetes mellitus (T2DM), hypertension, inflammatory biomarkers, need of orotracheal intubation, intensive care unit admission, incidence of death, and sickness symptom complex (SSC) scores and lower peripheral oxygen saturation (SpO₂) compared to those with non-critical disease. We found that 49.5 % of the variance in the severity of critical COVID-19 was explained by SpO₂ and SSC (negatively associated), chest computed tomography alterations (CCTA), inflammatory biomarkers, severe acute respiratory syndrome (SARS), BMI, T2DM, and age (positively associated). In this model, the NLRP3/IL18 variants showed indirect effects on critical COVID-19 that were mediated by inflammatory biomarkers, SARS, and SSC. Neural network models yielded a prediction of critical disease and death due to COVID-19 with an area under the receiving operating characteristic curve of 0.930 and 0.927, respectively. These ML methods increase the accuracy of predicting severity, critical illness, and mortality caused by COVID-19 and show that the genetic variants contribute to the predictive power of the ML models.

Kernel size selection in Convolutional Neural Networks (CNNs) is a critical but often overlooked design decision that affects receptive field, feature extraction, computational cost, and model accuracy. This paper proposes the Best Kernel Size Estimation Function (BKSEF), a mathematically grounded and empirically validated framework for optimal, layer-wise kernel size determination. BKSEF balances information gain, computational efficiency, and accuracy improvements by integrating principles from information theory, signal processing, and learning theory. Extensive experiments on CIFAR-10, CIFAR-100, ImageNet-lite, ChestX-ray14, and GTSRB datasets demonstrate that BKSEF-guided architectures achieve up to 3.1 percent accuracy improvement and 42.8 percent reduction in FLOPs compared to traditional models using uniform 3x3 kernels. Two real-world case studies further validate the approach: one for medical image classification in a cloud-based setup, and another for traffic sign recognition on edge devices. The former achieved enhanced interpretability and accuracy, while the latter reduced latency and model size significantly, with minimal accuracy trade-off. These results show that kernel size can be an active, optimizable parameter rather than a fixed heuristic. BKSEF provides practical heuristics and theoretical support for researchers and developers seeking efficient and application-aware CNN designs. It is suitable for integration into neural architecture search pipelines and real-time systems, offering a new perspective on CNN optimization.

We conducted a prospective study to evaluate the usefulness of ultralow-dose computed tomography (ULD-CT) with deep-learning reconstruction (DLR) compared with conventional standard-dose CT (SD-CT) for post-endovascular aneurysm repair (EVAR) surveillance. We prospectively performed post-EVAR surveillance using ULD-CT at a single center in 44 patients after they had received SD-CT. The ULD-CT images underwent DLR, whereas the SD-CT images underwent iterative reconstruction. Three radiologists blinded to the patient information and CT conditions independently measured the aneurysmal sac diameter and evaluated the overall image quality. Bland-Altman analysis and a linear mixed-effects model were used to assess and compare the measurement accuracy between SD-CT and ULD-CT. The mean CT dose index volume and dose-length product were significantly lower for ULD-CT (1.0 ± 0.3 mGy and 71.4 ± 26.5 mGy•cm) than that for SD-CT (6.9 ± 0.9 mGy and 500.9 ± 96.0 mGy•cm; p<0.001). The mean short diameters of the aneurysmal sac measured by the 3 observers were 46.7 ± 10.8 mm on SD-CT and 46.3 ± 10.8 mm on ULD-CT. The mean difference in the short diameter of the aneurysmal sac between ULD-CT and SD-CT was -0.37 mm (95% confidence interval, -0.6 to -0.12 mm). The intraobserver limits of agreement (LOA) for measurements by ULD-CT and SD-CT were -3.5 to 2.6, -2.8 to 1.9, and -2.9 to 2.3 for Observers 1, 2, and 3, respectively. The pairwise LOAs for assessing interobserver agreement, such as for the differences between Observers 1 and 2 measurements in SD-CT, were mostly within the predetermined acceptable range. The mean image-quality score was lower for ULD-CT (3.3 ± 0.6) than that for SD-CT (4.5 ± 0.5; p<0.001). Aneurysmal sac diameter measurements by ULD-CT with DLR were sufficiently accurate for post-EVAR surveillance, with substantial radiation reduction versus SD-CT.Clinical ImpactDeep-learning reconstruction (DLR) is implemented as a software-based algorithm rather than requiring dedicated hardware. As such, it is expected to be integrated into standard computed tomography (CT) systems in the near future. The ultralow-dose CT (ULD-CT) with DLR evaluated in this study has the potential to become widely accessible across various institutions. This advancement could substantially reduce radiation exposure in post-endovascular aneurysm repair (EVAR) CT imaging, thereby facilitating its adoption as a standard modality for post-EVAR surveillance.

The clinical application of artificial intelligence (AI) models based on breast ultrasound static images has been hindered in real-world workflows due to operator-dependence of standardized image acquisition and incomplete view of breast lesions on static images. To better exploit the real-time advantages of ultrasound and more conducive to clinical application, we proposed a whole-lesion-aware network based on freehand ultrasound video (WAUVE) scanning in an arbitrary direction for predicting overall breast cancer risk score. The WAUVE was developed using 2912 videos (2912 lesions) of 2771 patients retrospectively collected from May 2020 to August 2022 in two hospitals. We compared the diagnostic performance of WAUVE with static 2D-ResNet50 and dynamic TimeSformer models in the internal validation set. Subsequently, a dataset comprising 190 videos (190 lesions) from 175 patients prospectively collected from December 2022 to April 2023 in two other hospitals, was used as an independent external validation set. A reader study was conducted by four experienced radiologists on the external validation set. We compared the diagnostic performance of WAUVE with the four experienced radiologists and evaluated the auxiliary value of model for radiologists. The WAUVE demonstrated superior performance compared to the 2D-ResNet50 model, while similar to the TimeSformer model. In the external validation set, WAUVE achieved an area under the receiver operating characteristic curve (AUC) of 0.8998 (95% CI = 0.8529-0.9439), and showed a comparable diagnostic performance to that of four experienced radiologists in terms of sensitivity (97.39% vs. 98.48%, p = 0.36), specificity (49.33% vs. 50.00%, p = 0.92), and accuracy (78.42% vs.79.34%, p = 0.60). With the WAUVE model assistance, the average specificity of four experienced radiologists was improved by 6.67%, and higher consistency was achieved (from 0.807 to 0.838). The WAUVE based on non-standardized ultrasound scanning demonstrated excellent performance in breast cancer assessment which yielded outcomes similar to those of experienced radiologists, indicating the clinical application of the WAUVE model promising.

Early identification of unresectable hepatocellular carcinoma (HCC) patients who may benefit from immune checkpoint inhibitors (ICIs) is crucial for optimizing outcomes. Here, we developed a multimodal fusion (MMF) system integrating CT-derived deep learning features and clinical data to predict overall survival (OS) and progression-free survival (PFS). Using retrospective multicenter data (n = 859), the MMF combining an ensemble deep learning (Ensemble-DL) model with clinical variables achieved strong external validation performance (C-index: OS = 0.74, PFS = 0.69), outperforming radiomics (29.8% OS improvement), mRECIST (27.6% OS improvement), clinical benchmarks (C-index: OS = 0.67, p = 0.0011; PFS = 0.65, p = 0.033), and Ensemble-DL (C-index: OS = 0.69, p = 0.0028; PFS = 0.66, p = 0.044). The MMF system effectively stratified patients across clinical subgroups and demonstrated interpretability through activation maps and radiomic correlations. Differential gene expression analysis revealed enrichment of the PI3K/Akt pathway in patients identified by the MMF system. The MMF system provides an interpretable, clinically applicable approach to guide personalized ICI treatment in unresectable HCC.

Brain stroke is a leading cause of disability and mortality worldwide, necessitating the development of accurate and efficient diagnostic models. In this study, we explore the integration of Genetic Algorithm (GA)-based feature selection with three state-of-the-art deep learning architectures InceptionV3, VGG19, and MobileNetV2 to enhance stroke detection from neuroimaging data. GA is employed to optimize feature selection, reducing redundancy and improving model performance. The selected features are subsequently fed into the respective deep-learning models for classification. The dataset used in this study comprises neuroimages categorized into "Normal" and "Stroke" classes. Experimental results demonstrate that incorporating GA improves classification accuracy while reducing computational complexity. A comparative analysis of the three architectures reveals their effectiveness in stroke detection, with MobileNetV2 achieving the highest accuracy of 97.21%. Notably, the integration of Genetic Algorithms with MobileNetV2 for feature selection represents a novel contribution, setting this study apart from prior approaches that rely solely on traditional CNN pipelines. Owing to its lightweight design and low computational demands, MobileNetV2 also offers significant advantages for real-time clinical deployment, making it highly applicable for use in emergency care settings where rapid diagnosis is critical. Additionally, performance metrics such as precision, recall, F1-score, and Receiver Operating Characteristic (ROC) curves are evaluated to provide comprehensive insights into model efficacy. This research underscores the potential of genetic algorithm-driven optimization in enhancing deep learning-based medical image classification, paving the way for more efficient and reliable stroke diagnosis.

Accurate brain tumor classification is crucial in medical imaging to ensure reliable diagnosis and effective treatment planning. This study introduces a novel double ensembling framework that synergistically combines pre-trained deep learning (DL) models for feature extraction with optimized machine learning (ML) classifiers for robust classification. The framework incorporates comprehensive preprocessing and data augmentation of brain magnetic resonance images (MRI), followed by deep feature extraction using transfer learning with pre-trained Vision Transformer (ViT) networks. The novelty lies in the dual-level ensembling strategy: feature-level ensembling, which integrates deep features from the top-performing ViT models, and classifier-level ensembling, which aggregates predictions from hyperparameter-optimized ML classifiers. Experiments on two public Kaggle MRI brain tumor datasets demonstrate that this approach significantly surpasses state-of-the-art methods, underscoring the importance of feature and classifier fusion. The proposed methodology also highlights the critical roles of hyperparameter optimization (HPO) and advanced preprocessing techniques in improving diagnostic accuracy and reliability, advancing the integration of DL and ML for clinically relevant medical image analysis.