Early detection of malignant lesions in ultrasound images is crucial for effective cancer diagnosis and treatment. While traditional methods rely on radiologists, deep learning models can improve accuracy, reduce errors, and enhance efficiency. This study explores the application of a deep learning model for classifying benign and malignant lesions, focusing on its performance and interpretability. In this study, we proposed a feature fusion-based deep learning model for classifying benign and malignant lesions in ultrasound images. The model leverages advanced architectures such as MobileNetV2 and DenseNet121, enhanced with feature fusion and attention mechanisms to boost classification accuracy. The clinical dataset comprises 2171 images collected from 1758 patients between December 2020 and May 2024. Additionally, we utilized the publicly available BUSI dataset, consisting of 780 images from female patients aged 25 to 75, collected in 2018. To enhance interpretability, we applied Grad-CAM, Saliency Maps, and shapley additive explanations (SHAP) techniques to explain the model's decision-making. A comparative analysis with radiologists of varying expertise levels is also conducted. The proposed model exhibited the highest performance, achieving an AUC of 0.9320 on our private dataset and an area under the curve (AUC) of 0.9834 on the public dataset, significantly outperforming traditional deep convolutional neural network models. It also exceeded the diagnostic performance of radiologists, showcasing its potential as a reliable tool for medical image classification. The model's success can be attributed to its incorporation of advanced architectures, feature fusion, and attention mechanisms. The model's decision-making process was further clarified using interpretability techniques like Grad-CAM, Saliency Maps, and SHAP, offering insights into its ability to focus on relevant image features for accurate classification. The proposed deep learning model offers superior accuracy in classifying benign and malignant lesions in ultrasound images, outperforming traditional models and radiologists. Its strong performance, coupled with interpretability techniques, demonstrates its potential as a reliable and efficient tool for medical diagnostics. The datasets generated and analyzed during the current study are not publicly available due to the nature of this research and participants of this study, but may be available from the corresponding author on reasonable request.

Muscle wasting in critically ill patients, particularly those with prolonged hospitalization, poses a significant challenge to recovery and long-term outcomes. The aim of this study was to characterize long-term muscle wasting trajectories in ICU patients with acute respiratory distress syndrome (ARDS) due to COVID-19 and acute pancreatitis (AP), to evaluate correlations between muscle wasting and patient outcomes, and to identify clinically feasible thresholds that have the potential to enhance patient care strategies. A collective of 154 ICU patients (100 AP and 54 COVID-19 ARDS) with a minimum ICU stay of 10 days and at least three abdominal CT scans were retrospectively analyzed. AI-driven segmentation of CT scans quantified changes in psoas muscle area (PMA). A mixed model analysis was used to assess the correlation between mortality and muscle wasting, Cox regression was applied to identify potential predictors of survival. Muscle loss rates, survival thresholds and outcome correlations were assessed using Kaplan-Meier and receiver operating characteristic (ROC) analyses. Muscle loss in ICU patients was most pronounced in the first two weeks, peaking at -2.42% and - 2.39% psoas muscle area (PMA) loss per day in weeks 1 and 2, respectively, followed by a progressive decline. The median total PMA loss was 48.3%, with significantly greater losses in non-survivors. Mixed model analysis confirmed correlation of muscle wasting with mortality. Cox regression identified visceral adipose tissue (VAT), sequential organ failure assessment (SOFA) score and muscle wasting as significant risk factors, while increased skeletal muscle area (SMA) was protective. ROC and Kaplan-Meier analyses showed strong correlations between PMA loss thresholds and survival, with daily loss > 4% predicting the worst survival (39.7%). To our knowledge, This is the first study to highlight the substantial progression of muscle wasting in prolonged hospitalized ICU patients. The mortality-related thresholds for muscle wasting rates identified in this study may provide a basis for clinical risk stratification. Future research should validate these findings in larger cohorts and explore strategies to mitigate muscle loss. Not applicable.

The automatic screening of thyroid nodules using computer-aided diagnosis holds great promise in reducing missed and misdiagnosed cases in clinical practice. However, most current research focuses on single-modal images and does not fully leverage the comprehensive information from multimodal medical images, limiting model performance. To enhance screening accuracy, this study uses a deep learning framework that integrates high-dimensional convolutions of B-mode ultrasound (BMUS) and strain elastography (SE) images to predict the malignancy of TI-RADS 4 thyroid nodules with high-risk features. First, we extract nodule regions from the images and expand the boundary areas. Then, adaptive particle swarm optimization (APSO) and contrast limited adaptive histogram equalization (CLAHE) algorithms are applied to enhance ultrasound image contrast. Finally, deep learning techniques are used to extract and fuse high-dimensional features from both ultrasound modalities to classify benign and malignant thyroid nodules. The proposed model achieved an AUC of 0.937 (95 % CI 0.917-0.949) and 0.927 (95 % CI 0.907-0.948) in the test and external validation sets, respectively, demonstrating strong generalization ability. When compared with the diagnostic performance of three groups of radiologists, the model outperformed them significantly. Meanwhile, with the model's assistance, all three radiologist groups showed improved diagnostic performance. Furthermore, heatmaps generated by the model show a high alignment with radiologists' expertise, further confirming its credibility. The results indicate that our model can assist in clinical thyroid nodule diagnosis, reducing the risk of missed and misdiagnosed diagnoses, particularly for high-risk populations, and holds significant clinical value.

In multi-center neuroimaging studies, the technical variability caused by the batch differences could hinder the ability to aggregate data across sites, and negatively impact the reliability of study-level results. Recent efforts in neuroimaging harmonization have aimed to minimize these technical gaps and reduce technical variability across batches. While Generative Adversarial Networks (GAN) has been a prominent method for addressing harmonization tasks, GAN-harmonized images suffer from artifacts or anatomical distortions. Given the advancements of denoising diffusion probabilistic model which produces high-fidelity images, we have assessed the efficacy of the diffusion model for neuroimaging harmonization. While GAN-based methods intrinsically transform imaging styles between two domains per model, we have demonstrated the diffusion model's superior capability in harmonizing images across multiple domains with single model. Our experiments highlight that the learned domain invariant anatomical condition reinforces the model to accurately preserve the anatomical details while differentiating batch differences at each diffusion step. Our proposed method has been tested using T1-weighted MRI images from two public neuroimaging datasets of ADNI1 and ABIDE II, yielding harmonization results with consistent anatomy preservation and superior FID score compared to the GAN-based methods. We have conducted multiple analyses including extensive quantitative and qualitative evaluations against the baseline models, ablation study showcasing the benefits of the learned domain invariant conditions, and improvements in the consistency of perivascular spaces segmentation analysis and volumetric analysis through harmonization.

Structural remodeling of the left atrial appendage (LAA) is characteristic of atrial fibrillation (AF), and LAA morphology impacts radiofrequency catheter ablation (RFCA) outcomes. In this study, we aimed to develop and validate a predictive model for AF ablation outcomes using LAA morphological features, deep learning (DL) radiomics, and clinical variables. In this multicenter retrospective study, 480 consecutive patients who underwent RFCA for AF at three tertiary hospitals between January 2016 and December 2022 were analyzed, with follow-up through December 2023. Preprocedural CT angiography (CTA) images and laboratory data were systematically collected. LAA segmentation was performed using an nnUNet-based model, followed by radiomic feature extraction. Cox proportional hazard regression analysis assessed the relationship between AF recurrence and LAA volume. The dataset was randomly split into training (70%) and validation (30%) cohorts using stratified sampling. An AF recurrence prediction model integrating LAA DL radiomics with clinical variables was developed. The cohort had a median follow-up of 22 months (IQR 15-32), with 103 patients (21.5%) experiencing AF recurrence. The nnUNet segmentation model achieved a Dice coefficient of 0.89. Multivariate analysis showed that LAA volume was associated with a 5.8% increase in hazard risk per unit increase (aHR 1.058, 95% CI 1.021-1.095; p = 0.002). The model combining LAA DL radiomics with clinical variables demonstrated an AUC of 0.92 (95% CI 0.87-0.96) in the test set, maintaining robust predictive performance across subgroups. LAA morphology and volume are strongly linked to AF RFCA outcomes. We developed an LAA segmentation network and a predictive model that combines DL radiomics and clinical variables to estimate the probability of AF recurrence.

Deep learning has been utilized to segment teeth from computed tomography (CT) or cone-beam CT (CBCT). However, the performance of deep learning is unknown due to multiple models and diverse evaluation metrics. This systematic review and meta-analysis aims to evaluate the evolution and performance of deep learning in tooth segmentation. We systematically searched PubMed, Web of Science, Scopus, IEEE Xplore, arXiv.org, and ACM for studies investigating deep learning in human tooth segmentation from CT/CBCT. Included studies were assessed using the Quality Assessment of Diagnostic Accuracy Study (QUADAS-2) tool. Data were extracted for meta-analyses by random-effects models. A total of 30 studies were included in the systematic review, and 28 of them were included for meta-analyses. Various deep learning algorithms were categorized according to the backbone network, encompassing single-stage convolutional models, convolutional models with U-Net architecture, Transformer models, convolutional models with attention mechanisms, and combinations of multiple models. Convolutional models with U-Net architecture were the most commonly used deep learning algorithms. The integration of attention mechanism within convolutional models has become a new topic. 29 evaluation metrics were identified, with Dice Similarity Coefficient (DSC) being the most popular. The pooled results were 0.93 [0.93, 0.93] for DSC, 0.86 [0.85, 0.87] for Intersection over Union (IoU), 0.22 [0.19, 0.24] for Average Symmetric Surface Distance (ASSD), 0.92 [0.90, 0.94] for sensitivity, 0.71 [0.26, 1.17] for 95% Hausdorff distance, and 0.96 [0.93, 0.98] for precision. No significant difference was observed in the segmentation of single-rooted or multi-rooted teeth. No obvious correlation between sample size and segmentation performance was observed. Multiple deep learning algorithms have been successfully applied to tooth segmentation from CT/CBCT and their evolution has been well summarized and categorized according to their backbone structures. In future, studies are needed with standardized protocols and open labelled datasets.

Limited Angle Computed Tomography (LACT) often faces significant challenges due to missing angular information. Unlike previous methods that operate in the image domain, we propose a new method that focuses on sinogram inpainting. We leverage MR-SDEs, a variant of diffusion models that characterize the diffusion process with mean-reverting stochastic differential equations, to fill in missing angular data at the projection level. Furthermore, by combining distillation with constraining the output of the model using the pseudo-inverse of the inpainting matrix, the diffusion process is accelerated and done in a step, enabling efficient and accurate sinogram completion. A subsequent post-processing module back-projects the inpainted sinogram into the image domain and further refines the reconstruction, effectively suppressing artifacts while preserving critical structural details. Quantitative experimental results demonstrate that the proposed method achieves state-of-the-art performance in both perceptual and fidelity quality, offering a promising solution for LACT reconstruction in scientific and clinical applications.

Using massive datasets, foundation models are large-scale, pre-trained models that perform a wide range of tasks. These models have shown consistently improved results with the introduction of new methods. It is crucial to analyze how these trends impact the medical field and determine whether these advancements can drive meaningful change. This study investigates the application of recent state-of-the-art foundation models, DINOv2, MAE, VMamba, CoCa, SAM2, and AIMv2, for medical image classification. We explore their effectiveness on datasets including CBIS-DDSM for mammography, ISIC2019 for skin lesions, APTOS2019 for diabetic retinopathy, and CHEXPERT for chest radiographs. By fine-tuning these models and evaluating their configurations, we aim to understand the potential of these advancements in medical image classification. The results indicate that these advanced models significantly enhance classification outcomes, demonstrating robust performance despite limited labeled data. Based on our results, AIMv2, DINOv2, and SAM2 models outperformed others, demonstrating that progress in natural domain training has positively impacted the medical domain and improved classification outcomes. Our code is publicly available at: https://github.com/sajjad-sh33/Medical-Transfer-Learning.

A critical limitation to deployment and utilization of Artificial Intelligence (AI) algorithms in radiology practice is the actual integration of algorithms directly into the clinical Picture Archiving and Communications Systems (PACS). Here, we sought to integrate an AI-based pipeline for prostate organ and intraprostatic lesion segmentation within a clinical PACS environment to enable point-of-care utilization under a prospective clinical trial scenario. A previously trained, publicly available AI model for segmentation of intra-prostatic findings on multiparametric Magnetic Resonance Imaging (mpMRI) was converted into a containerized environment compatible with MONAI Deploy Express. An inference server and dedicated clinical PACS workflow were established within our institution for evaluation of real-time use of the AI algorithm. PACS-based deployment was prospectively evaluated in two phases: first, a consecutive cohort of patients undergoing diagnostic imaging at our institution and second, a consecutive cohort of patients undergoing biopsy based on mpMRI findings. The AI pipeline was executed from within the PACS environment by the radiologist. AI findings were imported into clinical biopsy planning software for target definition. Metrics analyzing deployment success, timing, and detection performance were recorded and summarized. In phase one, clinical PACS deployment was successfully executed in 57/58 cases and were obtained within one minute of activation (median 33 s [range 21-50 s]). Comparison with expert radiologist annotation demonstrated stable model performance compared to independent validation studies. In phase 2, 40/40 cases were successfully executed via PACS deployment and results were imported for biopsy targeting. Cancer detection rates for prostate cancer were 82.1% for ROI targets detected by both AI and radiologist, 47.8% in targets proposed by AI and accepted by radiologist, and 33.3% in targets identified by the radiologist alone. Integration of novel AI algorithms requiring multi-parametric input into clinical PACS environment is feasible and model outputs can be used for downstream clinical tasks.

'Brain age' is a biological clock typically used to describe brain health with one number, but its relationship with established gradients of cortical organization remains unclear. We address this gap by leveraging a data-driven, region-specific brain age approach in 335 neurologically intact adults, using a convolutional neural network (volBrain) to estimate regional brain ages directly from structural MRI without a predefined set of morphometric properties. Six distinct gradients of brain aging are replicated in two independent cohorts. Spatial patterns of accelerated brain aging in older adults quantitatively align with the archetypal sensorimotor-to-association axis of cortical organization. Other brain aging gradients reflect neurobiological hierarchies such as gene expression and externopyramidization. Participant-level correspondences to brain age gradients are associated with cognitive and sensorimotor performance and explained behavioral variance more effectively than global brain age. These results suggest that regional brain age patterns reflect fundamental principles of cortical organization and behavior.