Implant-based breast reconstruction is the most common surgical option following mastectomy for breast cancer. Despite its prevalence, up to one-third of patients develop complications within two years. Existing machine-learning models for predicting the complications rely solely on structured clinical data, overlooking prognostic information in narrative pathology reports. Recent advances in large language models (LLMs) enable extraction of numeric and semantic information from clinical text, offering opportunities to improve predictive performance and interpretability. We developed a fully on-premises, open-source model that extracts numeric morphometrics and contextual embeddings from free-text pathology reports, and fuses them with 63 structured variables via a CLIP-style dual encoder. In a single-center cohort of 963 patients (Jan 2007-Jan 2022), the multimodal model improved composite-complication risk discrimination (AUROC from 0.691 with logistic regression, increased to 0.740 with clinical features, and further improved to 0.764 with the addition of pathology report text features: p=0.027) and enhanced sensitivity and positive predictive value at clinical thresholds. The automated extraction module we developed for numeric morphometrics (e.g. mastectomy-specimen weight) from free-text pathology reports achieved an accuracy of 96.3%. SHAP analyses confirmed established risk factors expander-to-implant interval, body-mass index, and total mastectomy weight as dominant drivers. In subgroup analyses, model performance remained robust, with particularly strong discrimination observed among specific populations, such as shorter expander-to-implant interval (EII), the AUROC reached 0.796 and accuracy was 0.7999. These results show that on-premises, open-source LLMs reliably extract and fuse textual and structured clinical features to achieve clinically meaningful gains in predicting complications after implant-based breast reconstruction. While traditional models are constrained by the limited scope of structured variables, pathology text when analyzed with modern language models adds new, clinically relevant signals. Even modest statistical gains yield more accurate identification of high-risk patients, potentially informing surgical planning, patient counseling, and postoperative follow-up. These findings demonstrate that privacy-preserving language models, when integrated with contrastive multimodal alignment, can unlock prognostic information embedded in narrative pathology reports and enable interpretable, patient-level decision support. This interpretable, privacy-preserving multimodal framework offers a generalizable approach for enhancing risk prediction and clinical decision-making across surgical oncology. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement This study did not receive any funding. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The Institutional Review Board of Cedars-Sinai Medical Center gave ethical approval for this work. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes The de-identified structured clinical dataset is available from the corresponding author upon reasonable request. To protect patient privacy, the free-text pathology reports are not publicly available but may be shared with qualified researchers following the execution of a Data Use Agreement (DUA) and institutional approval.

Radiomics and deep learning both offer powerful tools for quantitative medical imaging, but most existing fusion approaches only leverage global radiomic features and overlook the complementary value of spatially resolved radiomic parametric maps. We propose a unified framework that first selects discriminative radiomic features and then injects them into a radiomics-enhanced nnUNet at both the global and voxel levels for pancreatic ductal adenocarcinoma (PDAC) detection. On the PANORAMA dataset, our method achieved AUC = 0.96 and AP = 0.84 in cross-validation. On an external in-house cohort, it achieved AUC = 0.95 and AP = 0.78, outperforming the baseline nnUNet; it also ranked second in the PANORAMA Grand Challenge. This demonstrates that handcrafted radiomics, when injected at both global and voxel levels, provide complementary signals to deep learning models for PDAC detection. Our code can be found at https://github.com/briandzt/dl-pdac-radiomics-global-n-paramaps

Splatting (GS)-based shared geometry framework adopts a two-stage training strategy, in which an explicit, subject-specific Gaussian scaffold encoding anatomical geometry is first learned from the isotropic structural scan and then reused to fit appearance for target modalities acquired with sparse slices. Experiments on the UK Biobank, GBM, and ABCD datasets for through-plane super-resolution across multiple modalities (T2-weighted, FLAIR, DWI, ASL), degradation factors (× 3, × 5, × 7), and pathological abnormalities (glioblastoma) demonstrate state-of-the-art reconstruction fidelity. The shared Gaussian geometry enables arbitrary-view generation for target modalities with strong structural consistency and further shows potential for self-supervised in-plane super-resolution. This work establishes explicit geometry-guided representations as a novel, flexible, and interpretable pathway toward retrospective multi-contrast MRI harmonization and reliable clinical reference construction. Source code is available at: https://github.com/yfgao76/AtlasGS

Relentless mechanical work of the heart is powered by continuous oxygen consumption. How the heart uses oxygen is a defining feature of its health. Invasive studies have established that impaired oxygen consumption by the myocardium predicts contractile dysfunction and adverse outcomes. Despite its importance, noninvasive quantification of myocardial oxygen use remains limited. Magnetic resonance imaging (MRI) signal is known to be sensitive to blood oxygenation and has the potential to quantify myocardial oxygen consumption noninvasively, without exogenous contrast agents and free of ionizing radiation. However, its clinical translation has been impeded by the need for complex biophysical calibration, vulnerability to imaging artifacts and consistent vital motions, and the requirement of lengthy acquisition times. Here, we introduce a rapid, self-calibrated cardiac MRI framework that overcomes these barriers through high-resolution, motion-resolved coronary sinus oximetry, which can quantify myocardial oxygen extraction of the whole heart within 3 minutes. We optimized the imaging parameters via numerical simulations and validated them against invasive coronary sinus catheterization in a porcine model. We combined the method with clinical MRI sequences and demonstrated the feasibility of quantifying myocardial oxygen consumption and myocardial oxygen efficiency in patients with and without heart failure secondary to myocardial infarction in a single institution. This needle-free approach establishes a practical framework for noninvasive characterization of myocardial oxygen metabolism. It holds the potential to facilitate early disease detection, inform personalized therapeutic strategies, and guide the development of cardiometabolic therapies aimed at addressing the ongoing heart failure epidemic.

The 2024 revisions of the McDonald diagnostic criteria for multiple sclerosis (MS) have incorporated susceptibility-based magnetic resonance imaging (MRI) biomarkers to improve diagnostic sensitivity and specificity. However, the addition of imaging sequences used to visualize these biomarkers (Time of Acquisition, TA: ~6 minutes) increases the overall patient scan time. This study addresses this by combining parallel imaging (TA: ~2 minutes) with our proposed deep learning-based image denoising method, complex-valued denoising convolutional neural network (ℂDnCNN), to generate high quality data. Network layers used in the denoising convolutional neural network (DnCNN) were extended to the complex domain for learning complex-valued MR image features in the image domain. Four real-valued and complex-valued versions of this network (2D DnCNN, 2D ℂDnCNN, 3D DnCNN, and 3D ℂDnCNN) were developed for testing on a simulated noise testing set and a real-world noise testing set. In the simulated noise testing set, 3D ℂ DnCNN outperformed the other approaches for denoising magnitude and complex MRI data across all levels of simulated noise. In the real-world noise testing set, 2D DnCNN yielded the highest increases for NRMSE, PSNR and CNR measures, while 3D DnCNN yielded the highest improvement for SSIM when denoising T2*-weighted magnitude images at the highest acceleration factor. For the complex-valued data, 3D ℂDnCNN outperformed 2D ℂDnCNN to produce high quality magnitude and phase data at all acceleration factors in all image quality metrics except for CNR measures. Overall, our study demonstrates the capability of deep learning-based image denoising methods to efficiently denoise ultra-fast submillimeter isotropic data. Our proposed ℂDnCNN is able to denoise complex-valued MRI data which further enables the generation of high-quality quantitative susceptibility mapping (QSM). Besides that, our experimental results using different denoising models indicate that CNNs should be designed to learn 3-dimensional image features in the complex domain to achieve optimal performance on MRI data.

We present Distortion-Guided Restoration (DGR), a physics-informed hybrid CNN-diffusion framework for acquisition-free correction of severe susceptibility-induced distortions in prostate single-shot EPI diffusion-weighted imaging (DWI). DGR is trained to invert a realistic forward distortion model using large-scale paired distorted and undistorted data synthesized from distortion-free prostate DWI and co-registered T2-weighted images from 410 multi-institutional studies, together with 11 measured B0 field maps from metal-implant cases incorporated into a forward simulator to generate low-b DWI (b = 50 s per mm squared), high-b DWI (b = 1400 s per mm squared), and ADC distortions. The network couples a CNN-based geometric correction module with conditional diffusion refinement under T2-weighted anatomical guidance. On a held-out synthetic validation set (n = 34) using ground-truth simulated distortion fields, DGR achieved higher PSNR and lower NMSE than FSL TOPUP and FUGUE. In 34 real clinical studies with severe distortion, including hip prostheses and marked rectal distension, DGR improved geometric fidelity and increased radiologist-rated image quality and diagnostic confidence. Overall, learning the inverse of a physically simulated forward process provides a practical alternative to acquisition-dependent distortion-correction pipelines for prostate DWI.

The quantitative multicontrast atherosclerosis characterization (qMATCH) technique enables high-resolution 3D multicontrast evaluation of carotid atherosclerosis, but the optimal acquisition orientation for reproducible image quality remains unclear. This prospective study compared image quality, motion robustness, and reproducibility between transverse and coronal qMATCH acquisitions. Thirty participants with moderate-to-severe carotid stenosis each underwent 2 scans, totaling 60 imaging sessions. Coronal imaging demonstrated significantly fewer motion artifacts compared with transverse imaging (34.7% versus 41.5%; P < .05) and yielded higher image quality scores (2.78 versus 2.72; P < .05) and arterial wall apparent SNR (5.10 versus 4.36; P < .05). In motion-corrupted images, coronal imaging also showed superior image quality (2.75 versus 2.65; P < .05) and arterial wall apparent SNR (4.89 versus 4.03; P < .05). Reproducibility for T1/T2 quantification was better for coronal acquisitions, particularly in motion-affected cases. These findings suggest that coronal qMATCH is more robust for motion-prone patients and better suited for longitudinal carotid imaging studies.

Background: Body composition is recognized as a major determinant of health outcomes, but its multidimensional nature makes clinical adoption challenging. We sought to develop and validate a body composition index (BCI) for all-cause mortality risk assessment, integrating variables of six body composition tissues. Methods: We analyzed 28509 consecutive patients undergoing myocardial perfusion imaging with routine low-dose chest CT attenuation correction (CTAC) scans acquired during myocardial perfusion imaging (MPI) at 12 centers across four countries. An artificial intelligence-based BCI was developed in a cohort of 15037 patients CTACs by integrating the CT-derived metrics of bone, skeletal muscle, and four adipose tissue compartments, coronary artery calcium score, and basic demographic variables (age, sex, BMI). The performance of BCI for mortality prediction was validated in an internal cohort of 6444 patients and an external cohort of 7028 patients by prognosis, calibration, net benefit, and explainability. Model-based simulation of tissue metrics modification was performed to evaluate estimated mortality risk reduction. Findings: During a median of 3.5 (IQR [1.9, 5.1]) years, 4697 (16%) patients died. In the external testing cohort, the BCI demonstrated excellent discrimination for mortality (area under receiver operating characteristic curve 0.78 (95% CI [0.76, 0.79]) and Harrell concordance index 0.75 [0.73, 0.76]), calibration, and net benefit overall and across pre-specified subgroups stratified by patient characteristics and imaging protocols. Visceral adipose tissue attenuation was the most influential body composition measure, followed by skeletal muscle volume. Simulated improvement in body composition was associated with significant mortality risk reduction. Interpretation: An index combining six body composition measures obtained opportunistically from routine chest CT provides robust mortality risk stratification. By converting complex body composition information into a single interpretable score, the BCI can facilitate clinical implementation of opportunistic CT biomarkers and guide individualized preventive strategies.

Abstract Objective To evaluate the case-level performance of deep-learning segmentation models for detecting extrahepatic bile duct stones on representative intraoperative cholangiography images (IOC) and to characterize the completeness of individual-stone localization. Background Retained bile duct stones can cause biliary obstruction, cholangitis, and pancreatitis. However false-positive interpretation of filling defects may prompt additional downstream procedures. Computer vision has been applied to biliary anatomy recognition and IOC adequacy assessment, but patient-level stone detection and individual-stone localization remain insufficiently studyed. Methods Representative IOC images were annotated for extrahepatic biliary anatomy and stones, with case-level stone status established using a composite clinical reference standard. Two deep-learning models were developed to delineate the common bile duct and common hepatic duct and to detect and localize stones. Case-level diagnostic performance was evaluated against the composite clinical reference standard, and individual-stone localization was evaluated against expert-reviewed annotations. Results On the held-out 125 patients test set, MiT-B2-UNet identified 23 of 25 stone-positive cases and 95 of 100 stone-negative cases, corresponding to a sensitivity of 0.920, specificity of 0.950, and AUC of 0.986. nnU-Net identified 19 of 25 stone-positive cases and 98 of 100 stone-negative cases, corresponding to a sensitivity of 0.760, specificity of 0.980, and AUC of 0.959. At the individual-stone level, MiT-B2-UNet and nnU-Net localized 31 of 59 and 25 of 59 annotated stones, respectively; all annotated stones were localized in 13 of 25 and 12 of 25 stone-positive cases. Conclusions Deep-learning models can identify stone-positive IOC cases and localize individual stones. This technology may help interpreting the IOCs and reduce retained stones and unnecessary downstream interventions.

Imaging-based spatial transcriptomic (ST) assays now profile targeted gene panels for hundreds of thousands of cells while preserving tissue context, yet most downstream analyses still need a compact, interpretable subset. We introduce CiCLoDS (Clustering in Critical & Low-Dimensional Subspace), an unsupervised framework adapted for spatial transcriptomics to jointly optimize clustering and feature selection under a strict, user-defined gene budget. The method returns an explicit feature subset aligned with the discovered partitions, accepts spatial side information via sine–cosine positional encodings, and converges in minutes on commodity CPUs. We benchmark CiCLoDS on three diverse datasets: Vizgen MERFISH mouse liver (1.27 M cells), 10x Xenium human colon (23 K cells), and Human DLPFC. On hepatocyte zonation, CiCLoDS raises the adjusted Rand index by up to +0.36 over PCA and geneBasis, preserves neighborhood structure with a kNN-overlap AUC of 0.89, and detects vessel-associated voids at 92.8% F1. On Human DLPFC (Visium), CiCLoDS demonstrates robust generalization, achieving a mean Adjusted Rand Index (ARI) of 0.40—surpassing BayesCafe (0.32)—and maintaining high accuracy on heterogeneous samples where baselines fail. Furthermore, we reveal a synergistic utility: using CiCLoDS to initialize BayesSpace raises the mean ARI to 0.50, resolving local minima issues to outperform either method alone. These results show that a lightweight, jointly optimized objective can achieve competitive accuracy, robust generalization, and synergistic utility while producing assay-ready feature panels.

Overall survival after transarterial chemoembolization (TACE) for hepatocellular carcinoma (HCC) is heterogeneous, and portal venous remodeling may contribute to outcome variability. We evaluated whether portal venous local vessel volume fraction (LVVF) maps carry prognostic information for 2-year overall survival (OS) after TACE, with a prespecified focus on the intermediate-stage Barcelona Clinic Liver Cancer (BCLC)-B subgroup. In this retrospective study, pre-TACE portal venous CT from 105 patients in the public HCC-TACE–Seg dataset were analyzed. OS was dichotomized at 104 weeks (OS < 104 weeks vs. OS ≥ 104 weeks). LVVF maps were computed by convolving the portal venous vessel mask with a 2-mm-radius 3D spherical kernel to yield voxel-wise local vessel volume fractions. In the full cohort, image-based risk models were trained and evaluated with fixed five-fold patient-level cross-validation using CT-only, LVVF-only, and combined CT+LVVF inputs. In the prespecified BCLC-B subgroup, incremental prognostic value of LVVF was assessed primarily via model ablation (CT-only vs. CT+LVVF) under a matched cross-validation protocol. LVVF-only modeling achieved a cross-validated area under the receiver operating characteristic curve (AUC) of 0.76 ± 0.03 with recall 0.94 ± 0.05. In the BCLC-B subgroup (n = 24; 13 events), incorporating LVVF improved risk stratification compared with CT-only [CT-only AUC/F1, 0.77/0.69 (95

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, pFDR < 0.05) and coronary endothelial function (r = 0.70, pFDR < 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