Walter R. Witschey
VerifiedUniversity of Pennsylvania · Rehabilitation Medicine
Active 1989–2026
About
Walter R. Witschey, PhD, is an Associate Professor of Radiology at the University of Pennsylvania and serves as the Associate Vice Chair of Research IT. He is also the Director of the Advanced Cardiovascular Imaging Lab at the Perelman School of Medicine. His core research focus is on cardiac magnetic resonance imaging (CMR), imaging technologies, and artificial intelligence. His lab employs a multidisciplinary and translational approach, integrating techniques from electrical engineering, computer science, biology, and bioengineering. His expertise includes MR physics, pulse sequence development such as T1ρ and quantitative susceptibility mapping (QSM), MR spectroscopy including 1H downfield MRS, non-proton imaging and spectroscopy (23Na and 31P), machine learning, computer vision, and closed-loop systems for medical imaging. His translational research interests encompass myocardial ischemia, coronary artery disease, reperfusion injury, non-ischemic heart disease, congenital heart disease, and metabolic imaging of the heart.
Research topics
- Biology
- Genetics
- Chemistry
- Medicine
- Biochemistry
- Internal medicine
- Physics
Selected publications
A generalizable deep learning system for cardiac MRI
Nature Biomedical Engineering · 2026-03-25
articleOpen accessCardiac MRI allows for a comprehensive assessment of myocardial structure, function and tissue characteristics. Here we describe a foundational vision system for cardiac MRI, capable of representing the breadth of human cardiovascular disease and health. Our deep-learning model is trained via self-supervised contrastive learning, in which visual concepts in cine-sequence cardiac MRI scans are learned from the raw text of the accompanying radiology reports. We train and evaluate our model on data from four large academic clinical institutions in the United States. We additionally showcase the performance of our models on the UK BioBank and two additional publicly available external datasets. We explore emergent capabilities of our system and demonstrate remarkable performance across a range of tasks, including the problem of left-ventricular ejection fraction regression and the diagnosis of 39 different conditions such as cardiac amyloidosis and hypertrophic cardiomyopathy. We show that our deep-learning system is capable of not only contextualizing the staggering complexity of human cardiovascular disease but can be directed towards clinical problems of interest, yielding impressive, clinical-grade diagnostic accuracy with a fraction of the training data typically required for such tasks.
The International Journal of Cardiovascular Imaging · 2025-07-05 · 1 citations
articleOpen accessQuantitative native cardiovascular magnetic resonance imaging (CMR) techniques such as T1, T2, and T1ρ mapping can provide myocardial tissue characteristics non-invasively. The objective of this study was to evaluate the alterations of T1, T2, and T1ρ in patients with precapillary pulmonary hypertension (PH) and their prognostic value. We prospectively enrolled 35 PH patients and 15 healthy controls between 2014 and 2017. PH patients were followed for all-cause death. CMR functional and tissue relaxation parameters were acquired and analyzed. Multivariate Cox regression models and time-dependent receiver operator characteristic (ROC) curves were constructed to determine the risk assessment/predictive value of T1, T2, and T1ρ for all-cause mortality. During a median follow-up of 79 months (Interquartile range 40-103 months), 15 out of 34 patients died. Compared with healthy controls, T1, T2, and T1ρ values were significantly higher in patients with precapillary PH. T1, T2, and T1ρ values at right ventricular (RV) insertion points (IPs) showed the high correlation with RV functional parameters. T1_RVIP showed the highest predictive value in time-dependent ROC analysis for all-cause mortality during the entire observation time (3-9 years). T2_RVIP of 56.1 ms showed the highest value predictive of 3-year mortality with an AUC of 0.950 (sensitivity: 1.000, specificity 0.824) and T1_RVIP of 1215.2 ms had excellent discrimination of AUC of 0.870 (sensitivity: 0.946, specificity 0.922). According to Kaplan-Meier analyses, a native T1_RVIP of 1215.2 ms (P = 0.003), a native T2_RVIP of 56.1 ms (P = 0.002), and a native T1ρ_RVIP of 109.5 ms (P = 0.005), were predictive of all-cause mortality. In univariate Cox regression models, T1_RVIP, T2_RVIP, and T1ρ_RVIP, as well as RVEDVi and NT-proBNP, are predictive of all-cause-mortality. In multivariate analyses, T2_RVIP was the sole independent predictor for mortality (hazard ratio, 2.5, [1.1-5.9], P = 0.035). Native tissue relaxation parameters T1, T2, and T1ρ at RVIPs are highly correlated with RV functional parameters and may have prognostic value. T2_RVIP is independently predictive of all-cause mortality in precapillary PH during long-term follow-up in this small prospective study.
Circulation · 2025-11-03
articleIntroduction: Patients with non-ischemic cardiomyopathy (NICM), present with complex intramural ventricular tachycardia (VT) circuitry and relatively high post-ablation recurrences. Hypothesis: We hypothesize that left ventricular epicardial adipose tissue (LV-EAT) is associated with myocardial lipomatous metaplasia (LM) and post-ablation VT recurrence in NICM patients. Methods: In this retrospective study, we quantified LV-EAT and LM in a cohort of consecutive NICM patients with cardiac contrast-enhanced CT prior to VT ablation. Cox proportional hazard regression models were used to determine the association of LV-EAT and LM with time to VT recurrence and mortality as the competing risk. Results: Among 113 patients, 50 experienced VT recurrence during 1.7 (IQR 0.7, 4.0) years median follow-up. Patients with VT recurrence demonstrated significantly greater LV-EAT volume (median 32.9 vs. 28.2 ml, p = 0.026), greater LV-EAT thickness (median 1.0 vs. 0.9 mm, p = 0.043), and a more negative LV-EAT attenuation (median -65.1 vs. -58.4 HU, p < 0.001; Figure 1A). Univariable Cox regression demonstrated associations between all LV-EAT parameters and VT recurrence, which persisted after adjustment for heart failure and amiodarone use (adjusted hazard ratio [aHR]: 1.01 [95% CI: 1.00–1.02] per 1 ml increase of LV-EAT volume; 1.79 [1.14–2.80] per 1 mm increase of thickness; 0.94 [0.91–0.97] per 1 HU increase of attenuation value). When incorporating LM, only LV-EAT attenuation remained independently associated (aHR 0.96 per 1 HU increase, p = 0.039), alongside LM volume (aHR 1.09 per 1 ml increase, p = 0.012). Distribution analysis, using the AHA-17 segment model, demonstrated an association between lateral and inferior wall LV-EAT and VT recurrence (Figure 1B). Moderate correlation was noted between LV-EAT attenuation and LM volume (r = 0.37). Mediation analysis revealed that 10% of the effect of LM on VT recurrence was mediated by LV-EAT attenuation, while 55% of the effect of LV-EAT attenuation on VT recurrence was mediated by LM (Figure 1C). Conclusions: LV-EAT demonstrates independent association with post-ablation VT recurrence in NICM patients while exhibiting significant statistical interaction with LM.
Mapping rare protein-coding variants on multi-organ imaging traits
Nature Communications · 2025-12-23
articleOpen accessHuman organ structure and function are important endophenotypes for clinical outcomes. Genome-wide association studies (GWAS) have identified numerous common variants associated with phenotypes derived from magnetic resonance imaging (MRI) of the brain and body. However, the role of rare protein-coding variations affecting organ size and function is largely unknown. Here we present an exome-wide association study that evaluates 596 multi-organ MRI traits across over 50,000 individuals from the UK Biobank. We identified 107 variant-level associations and 224 gene-based burden associations (67 unique gene-trait pairs) across all MRI modalities, including PTEN with total brain volume, TTN with regional peak circumferential strain in the heart left ventricle, and TNFRSF13B with spleen volume. The singleton burden model and AlphaMissense annotations contributed 8 unique gene-trait pairs including the association between an approved drug target gene of KCNA5 and brain functional activity. The identified rare coding signals elucidate some shared genetic effects across organs, prioritize previously identified GWAS loci, and are enriched for drug targets. Overall, we demonstrate how rare variants enhance our understanding of genetic effects on human organ morphology and function and their connections to complex diseases. Here, the authors report an exome-wide association study for multi-organ imaging traits by leveraging recent bioinformatic tools such as AlphaMissense. The identified signals elucidate the genetic effects from rare variants on human organs and their connections to complex diseases
Proceedings on CD-ROM - International Society for Magnetic Resonance in Medicine. Scientific Meeting and Exhibition/Proceedings of the International Society for Magnetic Resonance in Medicine, Scientific Meeting and Exhibition · 2025-09-16
articleSenior authorMotivation: There is a need for specific imaging of ischemia-reperfusion injury (IRI) as it initiates infarct iron deposition, catalyzing reactive oxygen species (ROS) production and heart failure. Goal(s): Our goal was to assess and image the relationships between susceptibility, R2*, and ROS activity. Approach: Six swine with IRI were imaged to quantify iron using QSM and R2*, and to measure ROS activity. Results: Regions of hemorrhage and infarct had higher susceptibility, R2* and ROS activity compared to remote regions. There was a positive correlation between infarct ROS activity and R2* in the hemorrhage, and between infarct ROS activity and susceptibility in the hemorrhage. Impact: The observed associations between magnetic susceptibility and R2* with ROS activity allows for better understanding of IRI and potential to develop new targeted interventions. This suggests that iron could be a catalyst for ROS production in ischemia-reperfusion injury.
Scientific Reports · 2025-11-10
articleOpen accessThis study examines the association of aortic geometric traits with flow characteristics and disease outcomes in 3204 patients from the Penn Medicine Biobank (PMBB). Using an nnU-Net, the thoracic aorta was segmented from CT scans to measure traits such as diameter and length. A one-dimensional reduced-order Navier-Stokes model (ROM) simulated aortic pulse pressure under various physiological conditions. Phenome-wide association studies (PheWAS) were conducted to link aortic traits to diseases using electronic health records (EHR). Significant associations were identified between aortic pulse pressure and conditions like aortic aneurysms, heart valve disorders, hypertension, and obesity. Notably, pulse pressure-but not aortic diameter-was also linked to diseases such as diabetes mellitus, wheezing, and chronic airway obstruction. The ROM-simulated pulse pressure showed not only previously recognized associations with diseases such as aortic aneurysm and hypertension, but also associations with conditions affecting organs outside the aorta. ROM hemodynamic simulations can be applied to thoracic images at the scale of thousands of patients. The ROM-simulated pulse pressure showed not only previously recognized associations with diseases including aortic aneurysm and hypertension, but also other diseases outside the aorta.
Annals of Biomedical Engineering · 2025-07-03 · 2 citations
articleOpen accessSenior authorAbstract Purpose Right ventricular (RV) remodeling in repaired tetralogy of Fallot (rToF) is a multifactorial process that may be affected by downstream hemodynamics. We therefore sought to characterize hemodynamics in the pulmonary arteries (PAs) of rToF patients using cardiovascular magnetic resonance (CMR)-derived computational fluid dynamics (CFD) and to study these variables in association with RV measurements at follow-up. Methods We selected patients with two CMRs who had magnetic resonance angiography (MRA) performed at baseline. The PA was segmented from the main PA (MPA) through the first bifurcation of the left PA (LPA) and right PA (RPA). Both steady and pulsatile simulations were performed. For each vessel, we calculated curvature, tortuosity, and both average (avg) and peak steady WSS (WSS steady ), time-averaged WSS (taWSS), WSS in systole (WSS systole ), and WSS in diastole (WSS diastole ), as well as oscillatory shear index (OSI). We studied these variables in association with RV metrics at follow-up including: RV end-diastolic volume index (RVEDVi), RV end-systolic volume index (RVESVi), RV stroke volume index (RVSVi), and RV ejection fraction (RVEF), as well as the outcome of pulmonic valve replacement (PVR). Results 22 patients met the inclusion criteria. Several focal hemodynamic metrics in the main and branch PAs, including WSS steady , taWSS, WSS systole , WSS diastole, and OSI were associated with RV measurements at follow-up, including RVEDVi, RVESVi, and RVSVi. LPA WSS steady,avg , RPA WSS steady,peak , whole vessel OSI avg , and MPA OSI avg were associated with likelihood of PVR. Conclusion CFD-derived hemodynamic variables in the PAs of rToF patients are associated with both PVR and RV remodeling.
Circulation · 2025-11-03
articleSenior authorIntroduction: In repaired tetralogy of Fallot (rToF), asymmetric remodeling of the pulmonary arteries (PA) leads to branch-specific hemodynamic changes. Geometric factors such as curvature influence wall shear stress (WSS) patterns, with distinct effects between the left (LPA) and right pulmonary arteries (RPA). Oscillatory shear index (OSI), which quantifies directional changes in WSS over the cardiac cycle, is a key marker of disturbed flow. This study investigates how curvature and other geometric factors influence hemodynamics in these two branches. Hypothesis: We hypothesize that geometric features influence PA hemodynamics in a branch-specific manner, with curvature having a stronger association with shear-related metrics in certain regions compared to others. Methods: Patient-specific PA models (n = 22) were reconstructed from cardiac magnetic resonance imaging, and computational fluid dynamics simulations were performed under steady and pulsatile flow conditions with patient-derived boundary conditions. Geometric parameters, including curvature and tortuosity, and hemodynamic metrics, including time-averaged WSS and OSI, were quantified. Spearman correlations assessed branch-specific relationships. Results: In the LPA, curvature showed a strong positive correlation with time-averaged WSS (ρ = 0.56, p = 0.006) and a negative correlation with OSI (ρ = -0.52, p = 0.013), indicating that higher curvature segments exhibit more unidirectional, high-shear flow (Figures 1 and 2) . In contrast, RPA curvature did not correlate significantly with any of the measured hemodynamic variables (all p > 0.28). The LPA curvature was significantly greater than the RPA curvature (p = 0.015). Tortuosity did not show significant correlations with hemodynamics in either branch (p > 0.17), suggesting that curvature is the dominant geometric modulator of wall shear stress (Table 1) . Conclusions: The LPA’s curvature-dependent hemodynamics characterized by significant time-averaged WSS and OSI patterns contrast with the RPA’s lack of such correlations. Anatomically, the RPA’s straighter anatomy minimizes flow disruption whereas the LPA curvature increases flow disruption. This study’s results align with prior studies showing sharper angulation in the LPA post-repair, promoting flow acceleration. Clinically, these findings highlight the importance of branch-specific geometric and hemodynamic assessments in rToF follow-up.
Circulation · 2025-11-03
articleIntroduction/Background: Left atrial (LA) pericardial and epicardial fat are associated with atrial fibrillation (AF). However, the association of LA intramyocardial fat, i.e. lipomatous metaplasia (LALM), with AF persistence and ablation outcomes is understudied. Research Question/Hypothesis: Do the total and regional distributions of LALM differ between paroxysmal and persistent AF, and does higher LALM burden predict early AF recurrence after catheter ablation? Methods/Approach: We retrospectively analyzed 100 patients who underwent contrast-enhanced cardiac computed tomography less than one year before AF ablation. The LA was segmented using ADAS software into standardized regions (Panel A) and LALM was quantified using intensity thresholds [−180 to 0 Hounsfield unit (HU)]. Total and regional LALM burdens were expressed as percentages of total LA myocardial and respective regional volumes. For transmural analysis, the myocardium was equally divided into 20%, 40%, 60%, and 80% wall depth-layers from sub-endocardium to sub-epicardium. Patients were stratified by AF persistence and by freedom from or recurrence of AF (≥30 s) beyond a three-month blanking period. Results/Data: Total LALM comprised a median of 16.6% (interquartile range: 11.0–21.1%) of the total LA myocardial volume. Regional burden was highest at the right (28.1%) and left (19.8%) pulmonary vein (PV) carinae, followed by the inferior wall (19.6%), interatrial septum (15.0%), posterior wall (14.1%), lateral wall (12.7%), and superior wall (10.7%). Patients with persistent AF (n = 43) had a significantly higher LALM percentage than those with paroxysmal AF (n = 57; 17.2% vs 16.1%, p = 0.03) (Panel B). No significant difference was found between patients with and without AF recurrence following ablation (16.4% vs 16.8%, p = 0.82) (Panel C). A transmural gradient was observed, with LALM increasing from the sub-endocardium (3.5%) to the sub-epicardium (24.9%, p < 0.001) (Panel D). Conclusions: LALM percentage is significantly greater in persistent than paroxysmal AF. Regionally, the highest percentages are found at the right and left PV carinae, followed by the inferior wall, and interatrial septum. Further investigation into LALM pathogenesis and its association with treatment outcomes is warranted.
Ascending Aortic Dimensions and Body Size
JACC. Cardiovascular imaging · 2025-08-22 · 2 citations
article
Recent grants
NIH · $77k · 2009
Non-invasive imaging of reactive oxygen species in reperfusion injury myocardial infarction
NIH · $3.0M · 2023–2027
NIH · $9.4M · 2021–2027
NIH · $8.8M · 2015
Cardiac MR imaging of hemorrhagic reperfusion injury after myocardial infarction
NIH · $2.8M · 2018–2024
Frequent coauthors
- 61 shared
Robert C. Gorman
University of Nebraska–Lincoln
- 49 shared
Ravinder Reddy
- 47 shared
Yuchi Han
- 39 shared
Julio A. Chirinos
University of Pennsylvania
- 38 shared
Joseph H. Gorman
- 36 shared
Francisco Contijoch
- 35 shared
James J. Pilla
University of Pennsylvania
- 34 shared
Arijitt Borthakur
California University of Pennsylvania
Labs
Advanced Cardiovascular Imaging LabPI
- Resume-aware match score
- Save to shortlist
- AI-drafted outreach
See your match with Walter R. Witschey
PhdFit ranks faculty by your research interests, methods, and publications — grounded in their actual work, not templates.
- Free to start
- No credit card
- 30-second signup