Faculty Bibliography

BACKGROUND:Accurate and equitable prediction of trauma-related in-hospital mortality is critical for guiding clinical decisions and optimising trauma care resources. Traditional severity scoring systems like the Injury Severity Score (ISS) do not account for demographic factors, potentially limiting their fairness and generalisability across diverse populations. METHODS:We developed and externally validated an artificial intelligence (AI) model based on ISS and integrated demographic features (age and sex) to predict in-hospital mortality after trauma. Data from the Korean Trauma Data Bank were used for model development and internal validation, comprising 121,418 patients with trauma aged ≥15 years treated at 19 trauma centres in South Korea (2017-2022). External validation was performed on an independent cohort of 7458 patients from five trauma centres (four in South Korea and one in Australia, 2022-2024). The primary outcome was trauma-related in-hospital mortality. Predictive performance was assessed using area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, accuracy, and balanced accuracy. Fairness was evaluated by comparing AUROC differences across age (<65 vs ≥65 years) and sex (female vs male) subgroups. FINDINGS/RESULTS:The ISS-based AI model incorporating age and sex achieved high predictive performance (internal validation AUROC, 0.934; external validation AUROC range, 0.901-0.920), outperforming conventional ISS-based methods. The model also demonstrated improved fairness, showing reduced AUROC differences across subgroups (age: 0.068 vs 0.091; sex: 0.021 vs 0.046 for AI model vs ISS, respectively). INTERPRETATION/CONCLUSIONS:Scaling an ISS-based AI model through demographic integration yielded accurate, fair, and generalisable predictions of trauma-related in-hospital mortality. This approach may enhance trauma care decision-making and enable more equitable resource allocation across diverse clinical settings. FUNDING/BACKGROUND:This research was supported by the MSIT (Ministry of Science and ICT), Korea, under the ITRC (Information Technology Research Center) support program (IITP-2025-RS-2024-00438239) and the Institute of Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (MSIT) (RS-2024-00509257, Global AI Frontier Lab). In addition, this research was supported by the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (RS-2025-02220492).

Early identification of high-risk trauma patients in the prehospital setting is crucial for optimizing resource allocation and improving survival. We developed and externally validated a real-time AI model predicting emergency room mortality using 21 prehospital variables. Model development and internal validation utilized the Korean Trauma Data Bank (KTDB; 204,189 patients), and external validation included four South Korean trauma centers (8,358 patients) and one Australian Level 1 center (3,578 patients). Our Prehospital-AI model, an ensemble of XGBoost, LightGBM, and random forest, achieved an AUROC of 0.923 (sensitivity: 0.780, specificity: 0.880) on the test set, outperforming the shock index (AUROC: 0.712). External validation yielded AUROCs of 0.925-0.956 across South Korean centers and 0.895 in the Australian center. Here we show that the Prehospital-AI model enables accurate, real-time risk assessment in the prehospital setting, outperforming traditional triage tools and improving trauma system efficiency. Nonetheless, additional multinational studies are warranted to further evaluate its generalizability across diverse trauma care systems.

Prostate MRI has transformed lesion detection and risk stratification in prostate cancer, but its impact is constrained by the high cost of the exam, variability in interpretation, and limited scalability. False negatives, false positives, and moderate inter-reader agreement undermine reliability, while long acquisition times restrict throughput. Artificial intelligence (AI) offers potential solutions to address many of the limitations of prostate MRI in the clinical management pathway. Machine learning-based triage can refine patient selection to optimize resources. Deep learning reconstruction enables accelerated acquisition while preserving diagnostic quality, with multiple FDA-cleared products now in clinical use. Ongoing development of automated quality assessment and artifact correction aims to improve reliability by reducing nondiagnostic exams. In image interpretation, AI models for lesion detection and clinically significant prostate cancer prediction achieve performance comparable to radiologists, and the PI-CAI international reader study has provided the strongest evidence to date of non-inferiority at scale. More recent work extends MRI-derived features into prognostic modeling of recurrence, metastasis, and functional outcomes. This review synthesizes progress across five domains-triage, accelerated acquisition and reconstruction, image quality assurance, diagnosis, and prognosis-highlighting the level of evidence, validation status, and barriers to adoption. While acquisition and reconstruction are furthest along, with FDA-cleared tools and prospective evaluations, triage, quality control, and prognosis remain earlier in development. Ensuring equitable performance across populations, incorporating uncertainty estimation, and conducting prospective workflow trials will be essential to move from promising prototypes to routine practice. Ultimately, AI could accelerate the adoption of prostate MRI toward a scalable platform for earlier detection and population-level prostate cancer management. EVIDENCE LEVEL: N/A TECHNICAL EFFICACY: 3.

PURPOSE/OBJECTIVE: METHODS:Like many other MRI simulators, ours discretizes magnetic fields in space. However, we extend the MR signal simulation at each grid point from the 0th-order approximation, which assumes piecewise constant fields, to a 1st-order approximation, which assumes piecewise linear fields. We solve the signal equation by analytically integrating over each grid cube, assuming linear field variations, and then summing over all cubes. We provide analytical integrals for several pulse sequences. RESULTS:The 1st-order approximation captures strongly varying fields and associated intravoxel dephasing more accurately, avoiding severe "ringing" artifacts present in the usual 0th-order simulations. This enables simulations on a much coarser grid, facilitating computational feasibility. CONCLUSION/CONCLUSIONS:The first-order simulator enables the evaluation of unconventional scanner designs with strongly varying magnetic fields.

OBJECTIVES/OBJECTIVE:Despite its high negative predictive value (NPV) for clinically significant prostate cancer (csPCa), MRI suffers from a substantial number of false positives, especially for intermediate-risk cases. In this work, we determine whether a deep learning model trained with PI-RADS-guided representation learning can disambiguate the PI-RADS 3 classification, detect csPCa from bi-parametric prostate MR images, and avoid unnecessary benign biopsies. MATERIALS AND METHODS/METHODS:This study included 28,263 MR examinations and radiology reports from 21,938 men imaged for known or suspected prostate cancer between 2015 and 2023 at our institution (21 imaging locations with 34 readers), with 6352 subsequent biopsies. We trained a deep learning model, a representation learner (RL), to learn how radiologists interpret conventionally acquired T2-weighted and diffusion-weighted MR images, using exams in which the radiologists are confident in their risk assessments (PI-RADS 1 and 2 for the absence of csPCa vs. PI-RADS 4 and 5 for the presence of csPCa, n=21,465). We then trained biopsy-decision models to detect csPCa (Gleason score ≥7) using these learned image representations, and compared them to the performance of radiologists, and of models trained on other clinical variables (age, prostate volume, PSA, and PSA density) for treatment-naïve test cohorts consisting of only PI-RADS 3 (n=253, csPCa=103) and all PI-RADS (n=531, csPCa=300) cases. RESULTS:On the 2 test cohorts (PI-RADS-3-only, all-PI-RADS), RL-based biopsy-decision models consistently yielded higher AUCs in detecting csPCa (AUC=0.73 [0.66, 0.79], 0.88 [0.85, 0.91]) compared with radiologists (equivocal, AUC=0.79 [0.75, 0.83]) and the clinical model (AUCs=0.69 [0.62, 0.75], 0.78 [0.74, 0.82]). In the PIRADS-3-only cohort, all of whom would be biopsied using our institution's standard of care, the RL decision model avoided 41% (62/150) of benign biopsies compared with the clinical model (26%, P<0.001), and improved biopsy yield by 10% compared with the PI-RADS ≥3 decision strategy (0.50 vs. 0.40). Furthermore, on the all-PI-RADS cohort, RL decision model avoided 27% of additional benign biopsies (138/231) compared to radiologists (33%, P<0.001) with comparable sensitivity (93% vs. 92%), higher NPV (0.87 vs. 0.77), and biopsy yield (0.75 vs. 0.64). The combination of clinical and RL decision models further avoided benign biopsies (46% in PI-RADS-3-only and 62% in all-PI-RADS) while improving NPV (0.82, 0.88) and biopsy yields (0.52, 0.76) across the 2 test cohorts. CONCLUSIONS:Our PI-RADS-guided deep learning RL model learns summary representations from bi-parametric prostate MR images that can provide additional information to disambiguate intermediate-risk PI-RADS 3 assessments. The resulting RL-based biopsy decision models also outperformed radiologists in avoiding benign biopsies while maintaining comparable sensitivity to csPCa for the all-PI-RADS cohort. Such AI models can easily be integrated into clinical practice to supplement radiologists' reads in general and improve biopsy yield for any equivocal decisions.

Serial Magnetic Resonance Imaging (MRI) exams are often performed in clinical practice, offering shared anatomical and motion information across imaging sessions. However, existing reconstruction methods process each session independently without leveraging this valuable longitudinal information. In this work, we propose a novel concept of longitudinal dynamic MRI, which incorporates patient-specific prior images to exploit temporal correlations across sessions. This framework enables progressive acceleration of data acquisition and reduction of scan time as more imaging sessions become available. The concept is demonstrated using the 4D Golden-angle RAdial Sparse Parallel (GRASP) MRI, a state-of-the-art dynamic imaging technique. Longitudinal reconstruction is performed by concatenating multi-session time-resolved 4D GRASP datasets into an extended dynamic series, followed by a low-rank subspace-based reconstruction algorithm. A series of experiments were conducted to evaluate the feasibility and performance of the proposed method. Results show that longitudinal 4D GRASP reconstruction consistently outperforms standard single-session reconstruction in image quality, while preserving inter-session variations. The approach demonstrated robustness to changes in anatomy, imaging intervals, and body contour, highlighting its potential for improving imaging efficiency and consistency in longitudinal MRI applications. More generally, this work suggests a new context-aware imaging paradigm in which the more we see a patient, the faster we can image.

Ultra low-field (ULF) MRI is an accessible neuroimaging modality that can bridge healthcare disparities and advance population-level brain health research. However, the inherently low signal-to-noise ratio of ULF-MRI often necessitates reductions in spatial resolution and, combined with the field-dependency of MRI contrast, challenges the accurate extraction of clinically relevant brain morphology. We evaluate the current state of ULF-MRI brain volumetry utilizing techniques for enhancing spatial resolution and leveraging recent advancements in brain segmentation. This is based on the agreement between ULF and corresponding high-field (HF) MRI brain volumes, and test-retest repeatability for multiple ULF scans. In this study, we find that accurate brain volumes can be measured from ULF-MRIs when combining orthogonal imaging directions for T₂-weighted images to form a higher resolution image volume. We also demonstrate that not all orthogonal imaging directions contribute equally to volumetric accuracy and provide a recommended scan protocol given the constraints of the current technology.

This perspective proposes adapting video-text generative AI to 3D medical imaging (CT/MRI) and medical videos (endoscopy/laparoscopy) by treating 3D images as videos. The approach leverages modern video models to analyze multiple sequences simultaneously and provide real-time AI assistance during procedures. The paper examines medical imaging's unique characteristics (synergistic information, metadata, and world model), outlines applications in automated reporting, case retrieval, and education, and addresses challenges of limited datasets, benchmarks, and specialized training.