Citations

ABSTRACT Background The brain–heart axis is central to vascular health, yet no imaging biomarkers capture integrated dysfunction across neural and coronary microvascular networks. Although coronary microvascular dysfunction links to cognitive decline, neural correlates connecting cerebral efficiency with coronary physiology remain unclear. Objectives To determine whether the Unified Structural and Functional Connectivity (USFC)—a multimodal magnetic resonance imaging (MRI) “traffic map” of brain network efficiency—predicts coronary endothelial function and cognition in women with ischemia and no obstructive coronary artery disease (INOCA). Methods Thirty‐three women with suspected INOCA from the Women's Ischemia Syndrome Evaluation (WISE) study (NCT03876223) underwent invasive coronary function testing, cardiac MRI, cognitive evaluation, and multimodal brain MRI. USFC, structural connectivity (SC), and functional connectivity (FC) were computed for predefined 10 backbone pathways. Support vector regression and logistic classification assessed predictive performance. Results USFC explained 16%–20% more variance in coronary endothelial function, myocardial perfusion reserve, and cognition than SC or FC alone ( p < 0.05). Connectivity between the left caudate–superior medial orbital gyrus and right calcarine–inferior occipital gyrus emerged as robust predictors of crystallized cognition ( r = –0.78, p FDR < 0.05) and coronary endothelial function ( r = 0.70, p FDR < 0.05), respectively. USFC also best discriminated low versus high coronary blood flow (area under the ROC curve [AUC]: USFC 0.622 vs. SC 0.349 and FC 0.425; p < 0.05). Conclusions USFC identifies neuro–cardiac pathways linking cerebral efficiency with coronary endothelial function. These results introduce a sensitive biomarker of systemic vulnerability, highlighting occipital and frontostriatal pathways as shared substrates of dysfunction. USFC offers a mechanistic framework for detecting vascular risk across metabolically demanding tissues. Trial Registration ClinicalTrials.gov identifier: NCT03876223

Abstract Objective Ceribell Inc.'s point‐of‐care electroencephalographic (EEG) system and artificial intelligence‐based Automated Seizure Burden Estimator (ASBE; ClarityPro) have US Food and Drug Administration clearance for diagnosing electrographic status epilepticus (ESE). The AccuRASE study using ASBE version 6 (V6) showed high negative predictive value (NPV) but limited sensitivity and positive predictive value (PPV) at certain thresholds. Version 8 (V8) is the updated algorithm trained with additional EEG samples. Methods We tested V8 on the previously used rapid‐response EEG test dataset (not used to train V8). Sensitivity, specificity, PPV, and NPV were compared against blinded expert annotations. ESE, ESE and possible ESE (ESE/pESE), electrographic seizures (ESz), and Esz with highly epileptiform patterns (Esz/HEPs) at burden thresholds of >1%, >10%, >20%, >50%, and >90% were analyzed. Additionally, we evaluated the built‐in ESE and ESE/pESE alerts (≥90% over 5 min, ≥20% over 1 h, or ≥10 continuous minutes of seizure). Results V8 showed 100% sensitivity at lower thresholds (>10%–20%) for ESE (V6 .86), without a specificity loss (V8 .9 vs. V6 .85), and retained NPV (V8 1.0 vs. V6 .99). Sensitivity at 50% threshold was .86 for ESE (V6 .71), and specificity was .94 (V6 .91); PPV for ESE was 30% for ESE and 80% for ESE/pESE. The specificity for ESE remained high for 90% burden (V8 .96, V6 .97) but the sensitivity dropped (.29 vs. V6 .43). Sensitivity was lower for ESz/HEPs (.54 → .44 at >10% threshold), although the algorithm is not designed to detect HEPs. Built‐in alert analysis showed significantly improved sensitivity for ESE (.57 → .86); specificity and NPV remained high (>.94). Significance ASBE V8 showed meaningful improvements in sensitivity, continued excellent NPV, and improved built‐in alert accuracy for ESE and ESE/pESE. The best seizure burden threshold for ruling out ESE now is 20% (vs. 10% with V6) and that for ruling in ESE/pESE with reasonable certainty is 50% (vs. 90% with V6).

Abstract Background An accurate discrete spot scanning (DSS) beam delivery time (BDT) model is essential, particularly as 4D dynamic dose accumulation calculations rely on machine‐specific time models for reliable interplay evaluation. Purpose To derive a BDT model for a synchrotron‐based Hitachi Probeat PBS proton system from oscilloscope measurements of key BDT structures, including multi‐energy extraction (MEE) characteristics. Methods Key BDT structures were characterized using oscilloscope measurements, supplemented by delivery log files where appropriate. Spot‐to‐spot scanning trajectories were visualized using a CROSSmini 2D strip ionization chamber (Liverage Biomedical Inc, Taiwan) via the dose‐driven continuous scanning (DDCS) mode. The MEE design and MEE charge recapture efficiencies were also characterized. In addition, the impact of the number of delivered spots on both the extractable fraction of charge and MEE recapture efficiency was investigated. Results An improved synchrotron‐based BDT model was derived that incorporates dead times between radiofrequency knockout (RFK) and high‐speed steering magnet (HSST) signals. Spot‐to‐spot movement follows a “hockey‐stick”‐like behavior where the effective spot‐to‐spot move time is governed by the slower axis. An energy‐dependent variation in the number of MEE layers deliverable per spill was also observed, with fewer MEE layers per spill (1–3) was observed in low and high energy ranges compared to the mid‐energy range (4–5), deviating from the nominal vendor value of five MEE layers per spill per block. A reduction in the extractable fraction of charge was observed with an increasing number of spots per spill, while increasing the number of delivered spots in the preceding layer significantly reduced the MEE recapture efficiency. Conclusion MEE characteristics should be incorporated into synchrotron‐based time models for accurate beam delivery modeling. The methodology presented here also provides a framework for centers seeking to derive machine‐specific delivery time models for their proton systems.

ABSTRACT Background & Aims Metabolic dysfunction‐associated steatotic liver disease (MASLD) is highly prevalent and increasingly recognized as a multisystem cardiometabolic disorder. The global burden, incidence and prognostic implications of coexisting chronic kidney disease (CKD) in MASLD remain uncertain. Methods In this meta‐analysis, we searched MEDLINE and EMBASE from inception through September 25, 2025, for observational studies reporting CKD prevalence and/or incidence among adults (≥ 18 years) with MASLD. The primary outcomes were pooled CKD prevalence, estimated using random‐effects models on the logit scale with the Hartung–Knapp adjustment, and pooled CKD incidence (per 1000 person‐years), estimated using random‐effects models with exact Poisson confidence intervals. Secondary outcomes included all‐cause mortality and cardiovascular, renal and cancer outcomes. Results From 2811 records, 36 observational studies met the inclusion criteria. Twenty‐three studies contributed prevalence estimates (575 615 participants; 56 248 CKD cases) and thirteen studies contributed incidence estimates (27 996 new events). The pooled global CKD prevalence among individuals with MASLD was 15.22% (95% CI 9.69–23.12). The pooled CKD incidence was 22.17 per 1000 person‐years (95% CI 11.44–32.89) in the MASLD population. Hypertension (OR 1.49, 95% CI 1.29–1.72), dyslipidaemia (OR 1.22, 95% CI 1.20–1.24) and diabetes (OR 1.94, 95% CI 1.69–2.22) were associated with higher odds of CKD in the MASLD population. All‐cause mortality rates were 18.28 per 1000 person‐years (95% CI 3.45–33.11) in coexistent MASLD and CKD, more than double those in the MASLD‐only population (7.26 per 1000 person‐years (95% CI 2.97–11.56)). Conclusions Coexistent CKD affects one in seven individuals with MASLD worldwide, conferring a more than two‐fold increased risk of all‐cause mortality compared with individuals with MASLD alone.

Abstract Background & Objectives Cholangiocarcinoma (CCA) is a rare and aggressive malignancy of the bile ducts characterized by complex imaging patterns, high inter-reader variability, and limited subspecialty expertise. Despite the critical importance of early detection for improving patient outcomes, most artificial intelligence (AI) tools developed for radiology have focused on other organ systems, leaving CCA largely underserved. Furthermore, clinicians with relevant domain expertise frequently lack the programming background required to participate meaningfully in AI development. This work aimed to create an accessible, educational AI framework “AI Starter” designed to empower radiologists and clinicians to initiate medical image analysis projects for CCA without requiring extensive coding knowledge. Methods A structured, end-to-end AI pipeline was introduced covering the key stages of image acquisition, preprocessing, segmentation, feature extraction, and predictive modeling. Open-source frameworks, specifically MONAI and PyTorch, were employed to ensure reproducibility and accessibility. Transfer learning strategies were incorporated to address the challenge of limited CCA-specific training datasets. The pipeline was designed with modular components to reduce technical barriers and accommodate users across varying levels of computational experience. Results Radiologists and clinicians were able to successfully engage with and deploy AI tools for CCA imaging analysis using this framework. Transfer learning demonstrably improved model performance under data-limited conditions, and the open-source implementation supported reproducible research workflows. Key challenges identified included scanner variability, segmentation instability, feature instability, and model overfitting, all of which were transparently mapped within the pipeline. Conclusion The AI Starter framework offers a practical and educational entry point for the CCA community including basic scientists and clinicians to meaningfully engage with and understand AI-driven imaging analysis.

Abstract Myelodysplastic syndromes (MDS) are clinically and biologically diverse disorders, emphasizing the need for personalized treatment approaches. The International Working Group for Prognostication of MDS (IWG_PM) recently introduced a molecular classification, referred to as the MDS taxonomy, that categorizes patients into 16 subgroups based on 21 gene mutations, 6 cytogenetic abnormalities, and loss of heterozygosity (LOH) at TP53 and TET2 loci. This study sought to validate and enhance the clinical relevance of the MDS taxonomy by analyzing a large retrospective cohort ( n = 5136) and transcriptomic data from a prospective cohort ( n = 477). The taxonomy successfully identified subgroups with distinct clinical characteristics and disease progression patterns. However, incorporating gene interactions from taxonomy subgroups did not improve the prognostic performance of the Molecular International Prognostic Scoring System (IPSS‐M). We further assessed whether the taxonomy could guide management in patients receiving disease‐modifying therapies. Except for the “ TP53 ‐complex” subgroup, taxonomy classifications were not predictive of hypomethylating agent response or transplant outcomes. Nonetheless, they correlated with overall survival, suggesting that while both IPSS‐M and the taxonomy capture disease biology, other non‐genetic factors may influence treatment response. RNA sequencing confirmed the biological distinctiveness of the taxonomy groups. Transcriptomic profiling of CD34+ bone marrow cells revealed unique, homogeneous gene expression patterns, particularly within the AML‐like, bi TET2 , SF3B1 , and TP53 ‐complex subgroups. Further integration of multi‐omics data may refine MDS classification, improving clinical decision‐making and guiding the development of targeted therapies.

Abstract Large Language Models (LLMs) demonstrate strong capabilities in healthcare applications. However, recruiting human evaluators to assess usability and effectiveness becomes challenging in the large-scale LLM-generated outcomes. Existing metrics are often outdated for evaluating LLM-based healthcare technologies. This paper proposes a Research through Evaluation (RtE) approach that can refine metrics and improve assessments of LLM-generated outputs. We deploy RtE in a retrospective comparative study of prostate cancer patient inquiries, comparing responses from human clinical care teams with those generated by a GPT-4-based in-basket bot. Through three rounds of co-evaluations with clinical professionals and an LLM-as-graders study with four LLMs, the RtE method shows that starting with small, iterative evaluations helps human graders adjust metrics to suit their daily practices better. Additionally, in the LLM-as-graders study, LLMs exhibit similar behavior, preferring the in-basket bot's responses over the clinical care team. Our methodology and findings demonstrate that the RtE method can enhance the transparency and robustness of evaluation metrics for LLM-generated content, with the potential to generalize to evaluations in other clinical practices.

Abstract Spatial transcriptomics has been widely applied to study the spatial distribution of cell types, cell states, and specific gene expression in tissue samples. However, we show that there is a prevalent transcript leakage problem in spatial transcriptomics data, where transcripts expressed by a cell diffuse to its neighborhood and are recurrently detected in the nearby cells. By analyzing published data sets, we show that this problem is general across data produced from different tissues and different species using different imaging-based and sequencing-based spatial transcriptomics platforms. It affects both upstream tasks such as expression quantification as well as downstream tasks such as cell-type annotation and detection of spatially-dependent gene expression. To tackle the transcript leakage problem, we propose a reference-free Bayesian model-based method, DeLeakage, which cleans up the data much more effectively than existing denoising methods. DeLeakage also improves cell-type annotation and avoids false detection of spatially dependent expression.

Objective: To evaluate the institutional learning curve of robotic liver resection (RLR), its association with surgical complexity and outcomes in an international multicenter cohort. Summary of Background Data: RLR is increasingly adopted worldwide, but the dynamics of the learning process and its impact on surgical outcomes remain poorly defined. Methods: This international, multicenter, retrospective cohort study included consecutive patients who underwent RLR between 06/2013 and 12/2024 across 22 expert centers. Based on the first 150 cases at each center, three phases were used to define the learning curve: operative time for competency, major complications for proficiency, and textbook outcome for mastery. Risk-adjusted generalized additive models were used to assess the learning process. Benchmark performance was compared to international reference values. Results: Among 2,331 patients, the most common indications were hepatocellular carcinoma (36%), benign lesions (23%) and colorectal liver metastases (21%). Operative time decreased after achieving competency (case #62), from 240 to 220 minutes ( P <0.001). Major morbidity reached a plateau during the proficiency phase (cases #63–138; 9% vs. 11%, P =0.144), while progression to mastery (cases #139–150) was associated with a higher rate of textbook outcome (76% vs. 87%, P =0.021). In contrast, conversion (5.1% vs. 4.9% vs. 2%), bile leak (4.0% vs. 5.3% vs. 3.8%), and 90-day mortality (0.7% vs. 0.9% vs. 0.9%) remained low and stable across all learning phases. Importantly, benchmark criteria for operative time, hospital stay, and textbook outcome were met across all complexity groups, including during the early learning phase. Conclusions: Robotic liver surgery shows progressive outcome improvement with early achievement of benchmark-level results and stable performance across increasing complexity. These findings indicate that, in experienced centers, RLR can be safely implemented and integrated into clinical practice, supporting its key role in modern hepatobiliary surgery.

Abstract Hematoxylin and eosin (H&E) slides are routinely available in cancer care, but molecular profiling often requires additional tissue processing and turnaround time. We introduce HistoMethyl, a DNA methylation-aware pathology foundation model that aligns whole-slide histopathology with matched genome-scale methylation beta-value profiles during pre-training while requiring only an H&E slide at inference. We evaluated HistoMethyl across four task groups: gene mutation prediction, morphology-associated classification, overall survival prediction, and direct DNA methylation beta-value recovery. Evaluation spanned TCGA cross-validation, a disease-held-out lower-grade glioma cohort, and external cohort validation in CPTAC glioblastoma and SurGen rectal adenocarcinoma, totaling 14 cancer cohorts and 81 gene mutation tasks. The best-performing configuration improved mean mutation AUROC by 5.03 percentage points. By converting DNA methylation supervision into an H&E-only representation, His-toMethyl could support an early molecular triage layer that helps prioritize cases for confirmatory sequencing, methylation profiling, immunohistochemistry, and molecular tumor board review. These image-only predictions are intended to accelerate downstream molecular testing and tissue allocation while leaving final diagnosis and treatment selection anchored in validated molecular assays.