Coronary Artery Disease-Reporting and Data System (CAD-RADS), a standardized reporting system of stenosis severity from coronary computed tomography angiography (CCTA), is performed manually by expert radiologists, being time-consuming and prone to interobserver variability. While deep learning methods automating CAD-RADS scoring have been proposed, radiomics-based machine-learning approaches are lacking, despite their improved interpretability. This study aims to introduce a novel radiomics-based machine-learning approach for automating CAD-RADS scoring from CCTA images with multiplanar reconstruction. This retrospective monocentric study included 251 patients (male 70 %; mean age 60.5 ± 12.7) who underwent CCTA in 2016-2018 for clinical evaluation of CAD. Images were automatically segmented, and radiomic features were extracted. Clinical characteristics were collected. The image dataset was partitioned into training and test sets (90 %-10 %). The training phase encompassed feature scaling and selection, data balancing and model training within a 5-fold cross-validation. A cascade pipeline was implemented for both 6-class CAD-RADS scoring and 4-class therapy-oriented classification (0-1, 2, 3-4, 5), through consecutive sub-tasks. For each classification task the cascade pipeline was applied to develop clinical, radiomic, and combined models. The radiomic, combined and clinical models yielded AUC = 0.88 [0.86-0.88], AUC = 0.90 [0.88-0.90], and AUC = 0.66 [0.66-0.67] for the CAD-RADS scoring, and AUC = 0.93 [0.91-0.93], AUC = 0.97 [0.96-0.97], and AUC = 79 [0.78-0.79] for the therapy-oriented classification. The radiomic and combined models significantly outperformed (DeLong p-value < 0.05) the clinical one in class 1 and 2 (CAD-RADS cascade) and class 2 (therapy-oriented cascade). This study represents the first CAD-RADS classification radiomic model, guaranteeing higher explainability and providing a promising support system in coronary artery stenosis assessment.

Coronary artery calcium scoring (CACS), i.e. quantification of Agatston (AS) or volume score (VS), can be time consuming. The aim of this study was to compare automated, artificial intelligence (AI)-based CACS to manual scoring, in cardiac and chest CT for lung cancer screening. We selected 684 participants (59 ± 4.8 years; 48.8 % men) who underwent cardiac and non-ECG-triggered chest CT, including 484 participants with AS > 0 on cardiac CT. AI-based results were compared to manual AS and VS, by assessing sensitivity and accuracy, intraclass correlation coefficient (ICC), Bland-Altman analysis and Cohen's kappa for classification in AS strata (0;1-99;100-299;≥300). AI showed high CAC detection rate: 98.1% in cardiac CT (accuracy 97.1%) and 92.4% in chest CT (accuracy 92.1%). AI showed excellent agreement with manual AS (ICC:0.997 and 0.992) and manual VS (ICC:0.997 and 0.991), in cardiac CT and chest CT, respectively. In Bland-Altman analysis, there was a mean difference of 2.3 (limits of agreement (LoA):-42.7, 47.4) for AS on cardiac CT; 1.9 (LoA:-36.4, 40.2) for VS on cardiac CT; -0.3 (LoA:-74.8, 74.2) for AS on chest CT; and -0.6 (LoA:-65.7, 64.5) for VS on chest CT. Cohen's kappa was 0.952 (95%CI:0.934-0.970) for cardiac CT and 0.901 (95%CI:0.875-0.926) for chest CT, with concordance in 95.9 and 91.4% of cases, respectively. AI-based CACS shows high detection rate and strong correlation compared to manual CACS, with excellent risk classification agreement. AI may reduce evaluation time and enable opportunistic screening for CAC on low-dose chest CT.

ImportanceBrain MRI is the main imaging modality for pediatric brain tumors (PBTs); however, incomplete MRI exams are common in pediatric neuro-oncology settings and pose a barrier to the development and application of deep learning (DL) models, such as tumor segmentation and prognostic risk estimation. ObjectiveTo evaluate DL-based strategies (image-dropout training and generative image synthesis) and heuristic imputation approaches for handling missing MRI sequences in PBT imaging from clinical acquisition protocols, and to determine their impact on segmentation accuracy and prognostic risk estimation. DesignThis cohort study included 715 patients from the Childrens Brain Tumor Network (CBTN) and BraTS-PEDs, and 43 patients with longitudinal MRI (157 timepoints) from PNOC003/007 clinical trials. We developed a dropout-trained nnU-Net tumor segmentation model that randomly omitted FLAIR and/or T1w (no contrast) sequences during training to simulate missing inputs. We compared this against three imputation approaches: a generative model for image synthesis, copy-substitution heuristics, and zeroed missing inputs. Model-generated tumor volumes from each segmentation method were compared and evaluated against ground truth (expert manual segmentations) and incorporated into time-varying Cox regression models for survival analysis. SettingMulti-institutional PBT datasets and longitudinal clinical trial cohorts. ParticipantsAll patients had multi-parametric MRI and expert manual segmentations. The PNOC cohort had a median of three imaging timepoints and associated clinical data. Main Outcomes and MeasuresSegmentation accuracy (Dice scores), image quality metrics for synthesized scans (SSIM, PSNR, MSE), and survival discrimination (C-index, hazard ratios). ResultsThe dropout model achieved robust segmentation under missing MRI, with [&le;]0.04 Dice drop and a stable C-index of 0.65 compared to complete-input performance. DL-based MRI synthesis achieved high image quality (SSIM > 0.90) and removed artifacts, benefiting visual interpretability. Performance was consistent across cohorts and missing data scenarios. Conclusion and RelevanceModality-dropout training yields robust segmentation and risk-stratification on incomplete pediatric MRI without the computational and clinical complexity of synthesis approaches. Image synthesis, though less effective for these tasks, provides complementary benefits for artifact removal and qualitative assessment of missing or corrupted MRI scans. Together, these approaches can facilitate broader deployment of AI tools in real-world pediatric neuro-oncology settings.

Multiple Sclerosis (MS) is a chronic autoimmune disorder of the central nervous system that can lead to significant neurological disability. Retinal imaging--particularly Scanning Laser Ophthalmoscopy (SLO) and Optical Coherence Tomography (OCT)--provides valuable biomarkers for early MS diagnosis through non-invasive visualization of neurodegenerative changes. This study proposes a foundation-based bi-modal classification framework that integrates SLO images and OCT-derived retinal thickness maps for MS diagnosis. To facilitate this, we introduce two modality-specific foundation models--SLOFound and TMFound--fine-tuned from the RETFound-Fundus backbone using an independent dataset of 203 healthy eyes, acquired at Noor Ophthalmology Hospital with the Heidelberg Spectralis HRA+OCT system. This dataset, which contains only normal cases, was used exclusively for encoder adaptation and is entirely disjoint from the classification dataset. For the classification stage, we use a separate dataset comprising IR-SLO images from 32 MS patients and 70 healthy controls, collected at the Kashani Comprehensive MS Center in Isfahan, Iran. We first assess OCT-derived maps layer-wise and identify the Ganglion Cell-Inner Plexiform Layer (GCIPL) as the most informative for MS detection. All subsequent analyses utilize GCIPL thickness maps in conjunction with SLO images. Experimental evaluations on the MS classification dataset demonstrate that our foundation-based bi-modal model outperforms unimodal variants and a prior ResNet-based state-of-the-art model, achieving a classification accuracy of 97.37%, with perfect sensitivity (100%). These results highlight the effectiveness of leveraging pre-trained foundation models, even when fine-tuned on limited data, to build robust, efficient, and generalizable diagnostic tools for MS in medical imaging contexts where labeled datasets are often scarce.

Major depressive disorder (MDD) is a highly prevalent mental health condition with significant public health implications. Early detection is crucial for timely intervention, but current diagnostic methods often rely on subjective clinical assessments, leading to delayed or inaccurate diagnoses. Advances in neuroimaging and machine learning (ML) offer the potential for objective and accurate early detection. This study aimed to develop and validate ML models using multisite functional magnetic resonance imaging data for the early detection of MDD, compare their performance, and evaluate their clinical applicability. We used functional magnetic resonance imaging data from 1200 participants (600 with early-stage MDD and 600 healthy controls) across 3 public datasets. In total, 4 ML models-support vector machine, random forest, gradient boosting machine, and deep neural network-were trained and evaluated using a 5-fold cross-validation framework. Models were assessed for accuracy, sensitivity, specificity, F1-score, and area under the receiver operating characteristic curve. Shapley additive explanations values and activation maximization techniques were applied to interpret model predictions. The deep neural network model demonstrated superior performance with an accuracy of 89% (95% CI 86%-92%) and an area under the receiver operating characteristic curve of 0.95 (95% CI 0.93-0.97), outperforming traditional diagnostic methods by 15% (P<.001). Key predictive features included altered functional connectivity between the dorsolateral prefrontal cortex, anterior cingulate cortex, and limbic regions. The model achieved 78% sensitivity (95% CI 71%-85%) in identifying individuals who developed MDD within a 2-year follow-up period, demonstrating good generalizability across datasets. Our findings highlight the potential of artificial intelligence-driven approaches for the early detection of MDD, with implications for improving early intervention strategies. While promising, these tools should complement rather than replace clinical expertise, with careful consideration of ethical implications such as patient privacy and model biases.

Glioma is the most common primary malignant brain tumor and preoperative genetic profiling is essential for the management of glioma patients. Our study focused on tumor regions segmentation and predicting the World Health Organization (WHO) grade, isocitrate dehydrogenase (IDH) mutation, and 1p/19q codeletion status using deep learning models on preoperative MRI. To achieve accurate tumor segmentation, we developed an optimal mass transport (OMT) approach to transform irregular MRI brain images into tensors. In addition, we proposed an algebraic preclassification (APC) model utilizing multimode OMT tensor singular value decomposition (SVD) to estimate preclassification probabilities. The fully automated deep learning model named OMT-APC was used for multitask classification. Our study incorporated preoperative brain MRI data from 3,565 glioma patients across 16 datasets spanning Asia, Europe, and America. Among these, 2,551 patients from 5 datasets were used for training and internal validation. In comparison, 1,014 patients from 11 datasets, including 242 patients from The Cancer Genome Atlas (TCGA), were used as independent external test. The OMT segmentation model achieved mean lesion-wise Dice scores of 0.880. The OMT-APC model was evaluated on the TCGA dataset, achieving accuracies of 0.855, 0.917, and 0.809, with AUC scores of 0.845, 0.908, and 0.769 for WHO grade, IDH mutation, and 1p/19q codeletion, respectively, which outperformed the four radiologists in all tasks. These results highlighted the effectiveness of our OMT and tensor SVD-based methods in brain tumor genetic profiling, suggesting promising applications for algebraic and geometric methods in medical image analysis.

Interfraction variations during radiation therapy pose a challenge for patients with cervical cancer, highlighting the benefits of online adaptive radiation therapy (oART). However, adaptation decisions rely on subjective image reviews by physicians, leading to high interobserver variability and inefficiency. This study explores the feasibility of using artificial intelligence for decision-making in oART. A total of 24 patients with cervical cancer who underwent 671 fractions of daily fan-beam computed tomography (FBCT) guided oART were included in this study, with each fraction consisting of a daily FBCT image series and a pair of scheduled and adaptive plans. Dose deviations of scheduled plans exceeding predefined criteria were labeled as "trigger," otherwise as "nontrigger." A data set comprising 588 fractions from 21 patients was used for model development. For the machine learning model (ML), 101 morphologic, gray-level, and dosimetric features were extracted, with feature selection by the least absolute shrinkage and selection operator (LASSO) and classification by support vector machine (SVM). For deep learning, a Siamese network approach was used: the deep learning model of contour (DL_C) used only imaging data and contours, whereas a deep learning model of contour and dose (DL_D) also incorporated dosimetric data. A 5-fold cross-validation strategy was employed for model training and testing, and model performance was evaluated using the area under the curve (AUC), accuracy, precision, and recall. An independent data set comprising 83 fractions from 3 patients was used for model evaluation, with predictions compared against trigger labels assigned by 3 experienced radiation oncologists. Based on dosimetric labels, the 671 fractions were classified into 492 trigger and 179 nontrigger cases. The ML model selected 39 key features, primarily reflecting morphologic and gray-level changes in the clinical target volume (CTV) of the uterus (CTV_U), the CTV of the cervix, vagina, and parametrial tissues (CTV_C), and the small intestine. It achieved an AUC of 0.884, with accuracy, precision, and recall of 0.825, 0.824, and 0.827, respectively. The DL_C model demonstrated superior performance with an AUC of 0.917, accuracy of 0.869, precision of 0.860, and recall of 0.881. The DL_D model, which incorporated additional dosimetric data, exhibited a slight decline in performance compared with DL_C. Heatmap analyses indicated that for trigger fractions, the deep learning models focused on regions where the reference CT's CTV_U did not fully encompass the daily FBCT's CTV_U. Evaluation on an independent data set confirmed the robustness of all models. The weighted model's prediction accuracy significantly outperformed the physician consensus (0.855 vs 0.795), with comparable precision (0.917 vs 0.925) but substantially higher recall (0.887 vs 0.790). This study proposes machine learning and deep learning models to identify treatment fractions that may benefit from adaptive replanning in radical radiation therapy for cervical cancer, providing a promising decision-support tool to assist clinicians in determining when to trigger the oART workflow during treatment.

Cone beam computed tomography (CBCT)-based online adaptive radiation therapy is carried out using a synthetic CT (sCT) created through deformable registration between the patient-specific fan-beam CT, fan-beam computed tomography (FBCT), and daily CBCT. Ethos 2.0 allows for plan calculation directly on HyperSight CBCT and uses artificial intelligence-informed tools for daily contouring without the use of a priori information. This breaks an important link between daily adaptive sessions and initial reference plan preparation. This study explores adaptive radiation therapy for spine metastases without prior patient-specific imaging or treatment planning. We hypothesize that adaptive plans can be created when patient-specific positioning and anatomy is incorporated only once the patient has arrived at the treatment unit. An Ethos 2.0 emulator was used to create initial reference plans on 10 patient-specific FBCTs. Reference plans were also created using FBCTs of (1) a library patient with clinically acceptable contours and (2) a water-equivalent phantom with placeholder contours. Adaptive sessions were simulated for each patient using the 3 different starting points. Resulting adaptive plans were compared with determine the significance of patient-specific information prior to the start of treatment. The library patient and phantom reference plans did not generate adaptive plans that differed significantly from the standard workflow for all clinical constraints for target coverage and organ at risk sparing (P > .2). Gamma comparison between the 3 adaptive plans for each patient (3%/3 mm) demonstrated overall similarity of dose distributions (pass rate > 95%), for all but 2 cases. Failures occurred mainly in low-dose regions, highlighting difference in fluence used to achieve the same clinical goals. This study confirmed feasibility of a procedure for treatment of spine metastases that does not rely on previously acquired patient-specific imaging, contours or plan. Reference-free direct-to-treatment workflows are possible and can condense a multistep process to a single location with dedicated resources.

To establish an artificial intelligence (AI)-empowered multistep integrated (MSI) radiation therapy (RT) workflow for patients with nasopharyngeal carcinoma (NPC) and evaluate its feasibility and clinical performance. Patients with NPC scheduled for MSI RT workflow were prospectively enrolled. This workflow integrates RT procedures from computed tomography (CT) scan to beam delivery, all performed with the patient on the treatment couch. Workflow performance, tumor response, patient-reported acute toxicities, and quality of life were evaluated. From March 2022 to October 2023, 120 newly diagnosed, nonmetastatic patients with NPC were enrolled. Of these, 117 completed the workflow with a median duration of 23.2 minutes (range, 16.3-45.8). Median translation errors were 0.2 mm (from CT scan to planning approval) and 0.1 mm (during beam delivery). AI-generated contours required minimal revision for the high-risk clinical target volume and organs at risk, minor revision for the involved cervical lymph nodes and low-risk clinical target volume (median Dice similarity coefficients (DSC), 0.98 and 0.94), and more revision for the gross tumor at the primary site and the involved retropharyngeal lymph nodes (median DSC, 0.84). Of 117 AI-generated plans, 108 (92.3%) passed after the first optimization, with ≥97.8% of target volumes receiving ≥100% of the prescribed dose. Dosimetric constraints were met for most organs at risk, except the thyroid and submandibular glands. One hundred and fifteen patients achieved a complete response at week 12 post-RT, while 14 patients reported any acute toxicity as "very severe" from the start of RT to week 12 post-RT. AI-empowered MSI RT workflow for patients with NPC is clinically feasible in a single institutional setting compared with standard, human-based RT workflow.

We propose a tumor tracking framework for 2D cine magnetic resonance imaging (MRI) based on a pair of deep learning (DL) models relying on patient-specific (PS) training. The chosen DL models are: (1) an image registration transformer and (2) an auto-segmentation convolutional neural network (CNN). We collected over 1,400,000 cine MRI frames from 219 patients treated on a 0.35 T MRI-linac plus 7500 frames from additional 35 patients that were manually labeled and subdivided into fine-tuning, validation, and testing sets. The transformer was first trained on the unlabeled data (without segmentations). We then continued training (with segmentations) either on the fine-tuning set or for PS models based on 8 randomly selected frames from the first 5 seconds of each patient's cine MRI. The PS auto-segmentation CNN was trained from scratch with the same 8 frames for each patient, without pre-training. Furthermore, we implemented B-spline image registration as a conventional model, as well as different baselines. Output segmentations of all models were compared on the testing set using the Dice similarity coefficient, the 50% and 95% Hausdorff distance (HD_50%/HD_95%), and the root-mean-square-error of the target centroid in superior-inferior direction. The PS transformer and CNN significantly outperformed all other models, achieving a median (interquartile range) dice similarity coefficient of 0.92 (0.03)/0.90 (0.04), HD_50% of 1.0 (0.1)/1.0 (0.4) mm, HD_95% of 3.1 (1.9)/3.8 (2.0) mm, and root-mean-square-error of the target centroid in superior-inferior direction of 0.7 (0.4)/0.9 (1.0) mm on the testing set. Their inference time was about 36/8 ms per frame and PS fine-tuning required 3 min for labeling and 8/4 min for training. The transformer was better than the CNN in 9/12 patients, the CNN better in 1/12 patients, and the 2 PS models achieved the same performance on the remaining 2/12 testing patients. For targets in the thorax, abdomen, and pelvis, we found 2 PS DL models to provide accurate real-time target localization during MRI-guided radiotherapy.