IntroductionIntegrated digital diagnostics can support complex surgeries in many anatomic sites, and brain tumour surgery represents one of the most complex cases. Neurosurgeons face several challenges during brain tumour surgeries, such as differentiating critical tissue from brain tumour margins. To overcome these challenges, the STRATUM project will develop a 3D decision support tool for brain surgery guidance and diagnostics based on multimodal data processing, including hyperspectral imaging, integrated as a point-of-care computing tool in neurosurgical workflows. This paper reports the protocol for the development and technical validation of the STRATUM tool. Methods and analysisThis international multicentre, prospective, open, observational cohort study, STRATUM-OS (study: 28 months, pre-recruitment: 2 months, recruitment: 20 months, follow-up: 6 months), with no control group, will collect data from 320 patients undergoing standard neurosurgical procedures to: (1) develop and technically validate the STRATUM tool, and (2) collect the outcome measures for comparing the standard procedure versus the standard procedure plus the use of the STRATUM tool during surgery in a subsequent historically controlled non-randomized clinical trial. Ethics and disseminationThe protocol was approved by the participant Ethics Committees. Results will be disseminated in scientific conferences and peer-reviewed journals. Trial registration number[Pending Number] ARTICLE SUMMARYO_ST_ABSStrengths and limitations of this studyC_ST_ABSO_LISTRATUM-OS will be the first multicentre prospective observational study to develop and technically validate a 3D decision support tool for brain surgery guidance and diagnostics in real-time based on artificial intelligence and multimodal data processing, including the emerging hyperspectral imaging modality. C_LIO_LIThis study encompasses a prospective collection of multimodal pre, intra and postoperative medical data, including innovative imaging modalities, from patients with intra-axial brain tumours. C_LIO_LIThis large observational study will act as historical control in a subsequent clinical trial to evaluate a fully-working prototype. C_LIO_LIAlthough the estimated sample size is deemed adequate for the purpose of the study, the complexity of the clinical context and the type of surgery could potentially lead to under-recruitment and under-representation of less prevalent tumour types. C_LI

BackgroundIntracranial hemorrhage (ICH), whether spontaneous or traumatic, is a neurological emergency with high morbidity and mortality. Accurate assessment of severity is essential for neurosurgical decision-making. This study aimed to develop and evaluate a fully automated, deep learning-based tool for the standardized assessment of ICH severity, based on the segmentation of the hemorrhage and intracranial structures, and the computation of an objective severity index. MethodsNon-contrast cranial CT scans from patients with spontaneous or traumatic ICH were retrospectively collected from public datasets and a tertiary care center. Deep learning models were trained to segment hemorrhages and intracranial structures. These segmentations were used to compute a severity index reflecting bleeding burden and mass effect through volumetric relationships. Segmentation performance was evaluated on a hold-out test cohort. In a prospective cohort, the severity index was assessed in relation to expert-rated CT severity, clinical outcomes, and the need for urgent neurosurgical intervention. ResultsA total of 1,110 non-contrast cranial CT scans were analyzed, 900 from the retrospective cohort and 200 from the prospective evaluation cohort. The binary segmentation model achieved a median Dice score of 0.90 for total hemorrhage. The multilabel model yielded Dice scores ranging from 0.55 to 0.94 across hemorrhage subtypes. The severity index significantly correlated with expert-rated CT severity (p < 0.001), the modified Rankin Scale (p = 0.007), and the Glasgow Outcome Scale-Extended (p = 0.039), and independently predicted the need for urgent surgery (p < 0.001). A threshold [~]300 was identified as a decision point for surgical management (AUC = 0.83). ConclusionWe developed a fully automated and openly accessible pipeline for the analysis of non-contrast cranial CT in intracranial hemorrhage. It computes a novel index that objectively quantifies hemorrhage severity and is significantly associated with clinically relevant outcomes, including the need for urgent neurosurgical intervention.

ImportanceUnderstanding the relevance of covert cerebrovascular disease (CCD) for later health will allow clinicians to more effectively monitor and target interventions. ObjectiveTo examine the association between clinically reported CCD, measured using natural language processing (NLP), and subsequent disease risk. Design, Setting and ParticipantsWe conducted a retrospective e-cohort study using linked health record data. From all people with clinical brain imaging in Scotland from 2010 to 2018, we selected people with no prior hospitalisation for neurological disease. The data were analysed from March 2024 to June 2025. ExposureFour phenotypes were identified with NLP of imaging reports: white matter hypoattenuation or hyperintensities (WMH), lacunes, cortical infarcts and cerebral atrophy. Main outcomes and measuresHazard ratios (aHR) for stroke, dementia, and Parkinsons disease (conditions previously associated with CCD), epilepsy (a brain-based control condition) and colorectal cancer (a non-brain control condition), adjusted for age, sex, deprivation, region, scan modality, and pre-scan healthcare, were calculated for each phenotype. ResultsFrom 395,273 people with brain imaging and no history of neurological disease, 145,978 (37%) had [&ge;]1 phenotype. For each phenotype, the aHR of any stroke was: WMH 1.4 (95%CI: 1.3-1.4), lacunes 1.6 (1.5-1.6), cortical infarct 1.7 (1.6-1.8), and cerebral atrophy 1.1 (1.0-1.1). The aHR of any dementia was: WMH, 1.3 (1.3-1.3), lacunes, 1.0 (0.9-1.0), cortical infarct 1.1 (1.0-1.1) and cerebral atrophy 1.7 (1.7-1.7). The aHR of Parkinsons disease was, in people with a report of: WMH 1.1 (1.0-1.2), lacunes 1.1 (0.9-1.2), cortical infarct 0.7 (0.6-0.9) and cerebral atrophy 1.4 (1.3-1.5). The aHRs between CCD phenotypes and epilepsy and colorectal cancer overlapped the null. Conclusions and RelevanceNLP identified CCD and atrophy phenotypes from routine clinical image reports, and these had important associations with future stroke, dementia and Parkinsons disease. Prevention of neurological disease in people with CCD should be a priority for healthcare providers and policymakers. Key PointsO_ST_ABSQuestionC_ST_ABSAre measures of Covert Cerebrovascular Disease (CCD) associated with the risk of subsequent disease (stroke, dementia, Parkinsons disease, epilepsy, and colorectal cancer)? FindingsThis study used a validated NLP algorithm to identify CCD (white matter hypoattenuation/hyperintensities, lacunes, cortical infarcts) and cerebral atrophy from both MRI and computed tomography (CT) imaging reports generated during routine healthcare in >395K people in Scotland. In adjusted models, we demonstrate higher risk of dementia (particularly Alzheimers disease) in people with atrophy, and higher risk of stroke in people with cortical infarcts. However, associations with an age-associated control outcome (colorectal cancer) were neutral, supporting a causal relationship. It also highlights differential associations between cerebral atrophy and dementia and cortical infarcts and stroke risk. MeaningCCD or atrophy on brain imaging reports in routine clinical practice is associated with a higher risk of stroke or dementia. Evidence is needed to support treatment strategies to reduce this risk. NLP can identify these important, otherwise uncoded, disease phenotypes, allowing research at scale into imaging-based biomarkers of dementia and stroke.

Recent advances in medical imaging have established deep learning-based segmentation as the predominant approach, though it typically requires large amounts of manually annotated data. However, obtaining annotations for intracranial hemorrhage (ICH) remains particularly challenging due to the tedious and costly labeling process. Semi-supervised learning (SSL) has emerged as a promising solution to address the scarcity of labeled data, especially in volumetric medical image segmentation. Unlike conventional SSL methods that primarily focus on high-confidence pseudo-labels or consistency regularization, we propose SWDL-Net, a novel SSL framework that exploits the complementary advantages of Laplacian pyramid and deep convolutional upsampling. The Laplacian pyramid excels at edge sharpening, while deep convolutions enhance detail precision through flexible feature mapping. Our framework achieves superior segmentation of lesion details and boundaries through a difference learning mechanism that effectively integrates these complementary approaches. Extensive experiments on a 271-case ICH dataset and public benchmarks demonstrate that SWDL-Net outperforms current state-of-the-art methods in scenarios with only 2% labeled data. Additional evaluations on the publicly available Brain Hemorrhage Segmentation Dataset (BHSD) with 5% labeled data further confirm the superiority of our approach. Code and data have been released at https://github.com/SIAT-CT-LAB/SWDL.

Delayed neurological sequelae are among the most serious complications of carbon monoxide poisoning. However, no reliable tools are available for evaluating its potential risk. We aimed to assess whether machine learning models using imaging features that were automatically extracted from brain MRI can predict the potential delayed neurological sequelae risk in patients with acute carbon monoxide poisoning. This single-center, retrospective, observational study analyzed a prospectively collected registry of acute carbon monoxide poisoning patients who visited our emergency department from April 2011 to December 2015. Overall, 1618 radiomics and 4 lesion-segmentation features from DWI b1000 and ADC images, as well as 62 clinical variables were extracted from each patient. The entire dataset was divided into five subsets, with one serving as the hold-out test set and the remaining four used for training and tuning. Four machine learning models, linear regression, support vector machine, random forest, and extreme gradient boosting, as well as an ensemble model, were trained and evaluated using 20 different data configurations. The primary evaluation metric was the mean and 95% CI of the area under the receiver operating characteristic curve. Shapley additive explanations were calculated and visualized to enhance model interpretability. Of the 373 patients, delayed neurological sequelae occurred in 99 (26.5%) patients (mean age 43.0 ± 15.2; 62.0% male). The means [95% CIs] of the area under the receiver operating characteristic curve, accuracy, sensitivity, and specificity of the best performing machine learning model for predicting the development of delayed neurological sequelae were 0.88 [0.86-0.9], 0.82 [0.8-0.83], 0.81 [0.79-0.83], and 0.82 [0.8-0.84], respectively. Among imaging features, the presence, size, and number of acute brain lesions on DWI b1000 and ADC images more accurately predicted DNS risk than advanced radiomics features based on shape, texture and wavelet transformation. Machine learning models developed using automatically extracted brain MRI features with clinical features can distinguish patients at delayed neurological sequelae risk. The models enable effective prediction of delayed neurological sequelae in patients with acute carbon monoxide poisoning, facilitating timely treatment planning for prevention. ABL = Acute brain lesion; AUROC = area under the receiver operating characteristic curve; CO = carbon monoxide; DNS = delayed neurological sequelae; LR = logistic regression; ML = machine learning; RF = random forest; SVM = support vector machine; XGBoost = extreme gradient boosting.

Prolonged imaging times and motion sensitivity at 7T necessitate advancements in image acceleration techniques. This study evaluates a 7T deep learning (DL)-based image reconstruction by using a deep neural network trained on 7T data, applied to T2-weighted turbo spin-echo imaging. Raw <i>k</i>-space data from 30 consecutive clinical 7T brain MRI patients was reconstructed by using both DL and standard methods. Qualitative assessments included overall image quality, artifacts, sharpness, structural conspicuity, and noise level, while quantitative metrics evaluated contrast-to-noise ratio (CNR) and image noise. DL-based reconstruction consistently outperformed standard methods across all qualitative metrics (<i>P</i> < .001), with a mean CNR increase of 50.8% (95% CI: 43.0%-58.6%) and a mean noise reduction of 35.1% (95% CI: 32.7%-37.6%). These findings demonstrate that DL-based reconstruction at 7T significantly enhances image quality without introducing adverse effects, offering a promising tool for addressing the challenges of ultra-high-field MRI.

Robustness against input data perturbations is essential for deploying deep learning models in clinical practice. Adversarial attacks involve subtle, voxel-level manipulations of scans to increase deep learning models' prediction errors. Testing deep learning model performance on examples of adversarial images provides a measure of robustness, and including adversarial images in the training set can improve the model's robustness. In this study, we examined adversarial training and input modifications to improve the robustness of deep learning models in predicting hematoma expansion (HE) from admission head CTs of patients with acute intracerebral hemorrhage (ICH). We used a multicenter cohort of <i>n</i> = 890 patients for cross-validation/training, and a cohort of <i>n</i> = 684 consecutive patients with ICH from 2 stroke centers for independent validation. Fast gradient sign method (FGSM) and projected gradient descent (PGD) adversarial attacks were applied for training and testing. We developed and tested 4 different models to predict ≥3 mL, ≥6 mL, ≥9 mL, and ≥12 mL HE in an independent validation cohort applying receiver operating characteristics area under the curve (AUC). We examined varying mixtures of adversarial and nonperturbed (clean) scans for training as well as including additional input from the hyperparameter-free Otsu multithreshold segmentation for model. When deep learning models trained solely on clean scans were tested with PGD and FGSM adversarial images, the average HE prediction AUC decreased from 0.8 to 0.67 and 0.71, respectively. Overall, the best performing strategy to improve model robustness was training with 5:3 mix of clean and PGD adversarial scans and addition of Otsu multithreshold segmentation to model input, increasing the average AUC to 0.77 against both PGD and FGSM adversarial attacks. Adversarial training with FGSM improved robustness against similar type attack but offered limited cross-attack robustness against PGD-type images. Adversarial training and inclusion of threshold-based segmentation as an additional input can improve deep learning model robustness in prediction of HE from admission head CTs in acute ICH.

FMRI, a non-invasive neuroimaging technique, can detect emotional brain activation patterns. It allows researchers to observe functional changes in the brain, making it a valuable tool for emotion recognition. For improved emotion recognition systems, it becomes crucial to understand the neural mechanisms behind emotional processing in the brain. There have been multiple studies across the world on the same, however, research on fMRI-based emotion recognition within the Indian population remains scarce, limiting the generalizability of existing models. To address this gap, a culturally relevant neuroimaging dataset has been created https://openneuro.org/datasets/ds005700 for identifying five emotional states i.e., calm, afraid, delighted, depressed and excited-in a diverse group of Indian participants. To ensure cultural relevance, emotional stimuli were derived from Bollywood movie clips. This study outlines the fMRI task design, experimental setup, data collection procedures, preprocessing steps, statistical analysis using the General Linear Model (GLM), and region-of-interest (ROI)-based dynamic functional connectivity (DFC) extraction using parcellation based on the Power et al. (2011) functional atlas. A supervised emotion classification model has been proposed using a Graph Convolutional Neural Network (GCNN), where graph structures were constructed from DFC matrices at varying thresholds. The DFC-GCNN model achieved an impressive 95% classification accuracy across 5-fold cross-validation, highlighting emotion-specific connectivity dynamics in key affective regions, including the amygdala, prefrontal cortex, and anterior insula. These findings emphasize the significance of temporal variability in emotional state classification. By introducing a culturally specific neuroimaging dataset and a GCNN-based emotion recognition framework, this research enhances the applicability of graph-based models for identifying region-wise connectivity patterns in fMRI data. It also offers novel insights into cross-cultural differences in emotional processing at the neural level. Furthermore, the high spatial and temporal resolution of the fMRI dataset provides a valuable resource for future studies in emotional neuroscience and related disciplines.

Fluorodeoxyglucose (FDG) PET to evaluate patients with epilepsy is one of the most common applications for simultaneous PET/MRI, given the need to image both brain structure and metabolism, but is suboptimal due to the radiation dose in this young population. Little work has been done synthesizing diagnostic quality PET images from MRI data or MRI data with ultralow-dose PET using advanced generative AI methods, such as diffusion models, with attention to clinical evaluations tailored for the epilepsy population. Here we compared the performance of diffusion- and non-diffusion-based deep learning models for the MRI-to-PET image translation task for epilepsy imaging using simultaneous PET/MRI in 52 subjects (40 train/2 validate/10 hold-out test). We tested three different models: 2 score-based generative diffusion models (SGM-Karras Diffusion [SGM-KD] and SGM-variance preserving [SGM-VP]) and a Transformer-Unet. We report results on standard image processing metrics as well as clinically relevant metrics, including congruency measures (Congruence Index and Congruency Mean Absolute Error) that assess hemispheric metabolic asymmetry, which is a key part of the clinical analysis of these images. The SGM-KD produced the best qualitative and quantitative results when synthesizing PET purely from T1w and T2 FLAIR images with the least mean absolute error in whole-brain specific uptake value ratio (SUVR) and highest intraclass correlation coefficient. When 1% low-dose PET images are included in the inputs, all models improve significantly and are interchangeable for quantitative performance and visual quality. In summary, SGMs hold great potential for pure MRI-to-PET translation, while all 3 model types can synthesize full-dose FDG-PET accurately using MRI and ultralow-dose PET.

Pragmatics plays a crucial role in effectively conveying messages across various social communication contexts. This aspect is frequently highlighted in the challenges experienced by children diagnosed with autism spectrum disorder (ASD). Notably, there remains a paucity of research investigating how the structural connectome (SC) predicts pragmatic language abilities within this population. Using diffusion tensor imaging (DTI) and deterministic tractography, we constructed the whole-brain white matter structural network (WMSN) in a cohort comprising 92 children with ASD and 52 typically developing (TD) preschoolers, matched for age and gender. We employed network-based statistic (NBS)-Predict, a novel methodology that integrates machine learning (ML) with NBS, to identify dysconnected subnetworks associated with ASD, and then to predict pragmatic language abilities based on the SC derived from the whole-brain WMSN in the ASD group. Initially, NBS-Predict identified a subnetwork characterized by 42 reduced connections across 37 brain regions (p = 0.01), achieving a highest classification accuracy of 79.4% (95% CI: 0.791 ~ 0.796). The dysconnected regions were predominantly localized within the brain's frontotemporal and subcortical areas, with the right superior medial frontal gyrus (SFGmed.R) emerging as the region exhibiting the most extensive disconnection. Moreover, NBS-Predict demonstrated that the optimal correlation coefficient between the predicted pragmatic language scores and the actual measured scores was 0.220 (95% CI: 0.174 ~ 0.265). This analysis revealed a significant association between the pragmatic language abilities of the ASD cohort and the white matter connections linking the SFGmed.R with the bilateral anterior cingulate gyrus (ACG). In summary, our findings suggest that the subnetworks displaying the most significant abnormal connections were concentrated in the frontotemporal and subcortical regions among the ASD group. Furthermore, the observed abnormalities in the white matter connection pathways between the SFGmed.R and ACG may underlie the neurobiological basis for pragmatic language deficits in preschool children with ASD.