About

Jerry Prince is the William B. Kouwenhoven Professor of Electrical and Computer Engineering at Johns Hopkins University. He has more than 30 years of experience in the research and practice of 3-D medical image reconstruction, registration, segmentation, and shape and motion analysis. He holds secondary appointments in the departments of Applied Mathematics and Statistics, and Computer Science, as well as joint appointments in Biomedical Engineering and Radiology at the Johns Hopkins University School of Medicine. He is also a member of the Data Science and AI Institute. Prince’s lab was among the first to develop methods to tag structures during magnetic resonance imaging tests. Algorithms developed by his lab to use MR tagging for studying cardiac motion led to the formation of the companies Diagnosoft and Myocardial Solutions. He is known for work segmenting the human brain cortex from MR images and improving optical coherence tomography (OCT) imaging of the retina. His current research focuses on image processing and computer vision, primarily applied to medical imaging. His ongoing projects include developing imaging and image processing methods to study speech pathologies in patients who have had partial surgical removal of the tongue due to cancer, and developing methods to characterize hydrocephalus, a condition involving fluid accumulation in the brain. Prince has published more than 500 articles on these subjects. He is a Fellow of the IEEE, MICCAI Society, and the American Institute for Medical and Biological Engineering. He is a member of several honor societies and professional organizations, including Sigma Xi, Tau Beta Pi, Eta Kappa Nu, Phi Kappa Phi, ISMRM, and SPIE. Prince has served as an associate editor for IEEE Transactions on Image Processing and IEEE Transactions on Medical Imaging, and is currently on the editorial board of Medical Image Analysis. His awards include the 1993 NSF Presidential Faculty Fellows Award, Maryland’s 1997 Outstanding Young Engineer, and the MICCAI Society Enduring Impact Award in 2012. He earned his BS in Electrical Engineering and Computer Science from the University of Connecticut in 1979 and his Ph.D. in Electrical Engineering from MIT in 1988. He joined the Johns Hopkins faculty in 1989.

Research topics

• Computer Science
• Artificial Intelligence
• Data Mining
• Natural Language Processing
• Medicine
• Mathematics
• Algorithm
• Computer vision
• Statistics

Selected publications

• ECLARE: efficient cross-planar learning for anisotropic resolution enhancement

  Journal of Medical Imaging · 2026-03-04 · 1 citations

  articleOpen access

  PurposeIn clinical imaging, magnetic resonance (MR) image volumes are often acquired as stacks of 2D slices with decreased scan times, improved signal-to-noise ratio, and image contrasts unique to 2D MR pulse sequences. Although this is sufficient for clinical evaluation, automated algorithms designed for 3D analysis perform poorly on multislice 2D MR volumes, especially those with thick slices and gaps between slices. Superresolution (SR) methods aim to address this problem, but previous methods do not address all of the following: slice profile shape estimation, slice gap, domain shift, and noninteger or arbitrary upsampling factors.ApproachWe propose ECLARE (Efficient Cross-planar Learning for Anisotropic Resolution Enhancement), a self-SR method that addresses each of these factors. ECLARE uses a slice profile estimated from the multislice 2D MR volume, trains a network to learn the mapping from low-resolution to high-resolution in-plane patches from the same volume, performs SR with antialiasing, and respects the image FOV during resampling. We compared ECLARE with cubic B-spline interpolation, SMORE, and other contemporary SR methods. We used realistic and representative simulations on human head MR volumes so that quantitative performance against ground truth can be computed. Specifically, healthy T1-w and people with MS T2-w FLAIR datasets were used for evaluations. We used the peak signal-to-noise ratio and structural similarity index measure as signal recovery metrics. We additionally used two independent brain parcellation algorithms, SLANT and SynthSeg, to compute the consistency Dice similarity coefficient and the R2 coefficient of determination, respectively, as comparison metrics.ResultsFor images with up to 5 mm of slice thickness and up to 1.5 mm of gap, ECLARE achieves greater mean PSNR and SSIM compared with other methods. In representative regions of interest, such as the ventricles, caudate, cerebral white matter, and cerebellar white matter, ECLARE performs comparably or better than other approaches. These trends are similar for both investigated datasets.ConclusionsThe use of slice profile estimation, FOV-aware resampling, and self-SR allowed ECLARE to robustly superresolve anisotropic images without the need for external training data. Future work will investigate the utility of ECLARE on other organs, species, modalities, and resolutions. Our code is open-source and available at https://www.github.com/sremedios/eclare.

  PublisherDOI

• Solving a Nonlinear Blind Inverse Problem for Tagged MRI with Physics and Deep Generative Priors

  arXiv (Cornell University) · 2026-03-01

  preprintOpen accessSenior author

  Tagged MRI enables tracking internal tissue motion non-invasively. It encodes motion by modulating anatomy with periodic tags, which deform along with tissue. However, the entanglement between anatomy, tags and motion poses significant challenges for post-processing. The existence of tags and imaging blur hinders downstream tasks such as segmenting anatomy. Tag fading, due to T1-relaxation, disrupts the brightness constancy assumption for motion tracking. For decades, these challenges have been handled in isolation and sub-optimally. In contrast, we introduce a blind and nonlinear inverse framework for tagged MRI that, for the first time, unifies these tasks: anatomical image recovery, high-resolution cine image synthesis, and motion estimation. At its core, the synergy of MR physics and generative priors enables us to blindly estimate the unknown forward imaging models and high-resolution underlying anatomy, while simultaneously tracking 3D diffeomorphic Lagrangian motion over time. Experiments on tagged brain MRI demonstrate that our approach yields high-resolution anatomy images, cine images, and more accurate motion than specialized methods.

  PublisherDOI

• Automated quality assurance of segmentation techniques with statistically‑based failure evaluation (SAFE)

  2026-02-13

  article
  PublisherDOI

• A Severity‐Agnostic Atrophy Pattern in Spinocerebellar Ataxia Type 3: Volumetrics from <scp>ENIGMA</scp> ‐Ataxia

  Movement Disorders · 2026-05-10

  article

  BACKGROUND: Spinocerebellar ataxia type 3 (SCA3) is a rare, inherited neurodegenerative disease characterized by progressive loss of motor coordination. OBJECTIVES: We undertook a multisite magnetic resonance imaging study to profile the spatial spread of atrophy across the brain, determine whether atrophy preferentially maps onto specific functional networks, and investigate the relationship between cerebellar and cerebral atrophy. METHODS: Whole-brain grey and white matter (GM and WM) voxel-based morphometry was performed on 408 individuals with SCA3 (82 pre-ataxic) and 293 controls. The SCA3 cohort was stratified by ataxia severity to study progression. Cerebellar GM atrophy was mapped onto a task-based functional atlas. Cerebrocerebellar volumetric covariance was assessed to determine whether cerebral and cerebellar atrophy were coupled. RESULTS: The atrophy pattern is spatially consistent but progressive in magnitude across the disease course. The greatest atrophy (Cohen's d > 1.5) occurred in the pons, cerebellar WM, and cerebellar peduncles; correlations with ataxia severity and duration were also strongest (-0.4 > r > -0.65) in those regions. Cerebellar GM atrophy was greatest (d ≅ 0.7) in functional regions associated with motor planning/execution, attention, and emotional processing. Sparse cerebral cortical atrophy appears only in the most severe disease subgroup, while striatal atrophy begins in the earliest stages but does not worsen with increasing clinical severity. Reduced cerebrocerebellar volumetric covariance is observed in SCA3 participants versus controls. CONCLUSIONS: Cerebellar and brainstem atrophy underlies greater ataxia severity in SCA3, but the spatial pattern of structural changes remains relatively consistent across the course of the disease. Cerebellar GM atrophy is spatially non-uniform, and occurs maximally in regions consistent with the motor and cognitive clinical presentation of SCA3. Cerebellar atrophy is not mirrored by corresponding cerebral structural changes. © 2026 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

  PublisherDOI

• DICOM-CLIP: Zero-Shot Acquisition Parameter Retrieval from Unprocessed Dicom Files

  2026-04-08

  article

  Contrast variation in magnetic resonance (MR) images arises from acquisition-specific parameters such as echo time, repetition time, and flip angle, which vary widely across individual acquisitions and hardware. These parameters are included in imaging metadata tags, but not often used for image categorization. Instead, broad labels such as ” <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{T}_{1}$</tex>-weighted” and ” <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{T}_{2}$</tex>-FLAIR” are commonly used, but the resultant images exhibit wide intra-label contrast variation. Creating anatomy-invariant contrast representations that are faithful to intra-contrast variation is valuable for image contrast identification and downstream tasks such as image harmonization. However, in large-scale real-world acquisitions, metadata is often erroneous, inconsistent, or missing, making interpretation difficult and complex. To recover these parameters and learn fine-grained representations of MR contrast, we propose DICOM-CLIP, a multimodal contrastive learning framework that embeds raw DICOM slices in the same space as their associated metadata. DICOM-CLIP is trained in a single-stage supervised contrastive (SupCon) setting using ground-truth metadata, optimizing embedding separability using exact-match supervision. Our approach yields up to 6% more accurate retrieval and up to 23 % improvement in Mean Average Error (MAE) over the current state-of-the-art.

  PublisherDOI

• VAID: Valve Artifact Inpainting for Normal Pressure Hydrocephalus from a 3D MRI Diffusion Model

  2026-04-08

  articleSenior author

  Normal Pressure Hydrocephalus (NPH) is a neurological disorder characterized by significant ventricular enlargement, potentially leading to cognitive impairment with gait and incontinence issues. Shunt implants can be used to drain excess cerebrospinal fluid (CSF) from the ventricles and can be an effective treatment in the correct cohorts. However, the presence of the shunt valve introduces artifacts in magnetic resonance images (MRI), which can bias downstream image processing pipelines. This bias hinders the accurate computation of biomarkers in evaluating the efficacy of shunt treatment. To address this, we propose VAID (Valve Artifact Inpainting with Diffusion models), a valve artifact removal method based on a fine-tuned 3D MRI diffusion model. The probabilistic representation learned by the diffusion model serves to provide realistic inpainting in the valve region of the shunt. Experiments with real pre- and post-shunt surgery MRI data demonstrate that our inpainting approach improves the accuracy of biomarker quantification in downstream neuroimaging pipelines such as SLANT and FreeSurfer.

  PublisherDOI

• Synthetic multi-inversion time magnetic resonance images for visualization of subcortical structures

  Journal of Medical Imaging · 2026-01-06

  articleOpen access

  PurposeVisualization of subcortical gray matter is essential in neuroscience and clinical practice, particularly for disease understanding and surgical planning. Although multi-inversion time (multi-TI) T1-weighted (T1-w) magnetic resonance (MR) imaging improves visualization, it is only acquired in specific clinical settings and not available in common public MR datasets.ApproachWe present SyMTIC (synthetic multi-TI contrasts), a deep learning method that generates synthetic multi-TI images using routinely acquired T1-w, T2-weighted (T2-w), and fluid-attenuated inversion recovery (FLAIR) images. Our approach combines image translation via deep neural networks with imaging physics to estimate longitudinal relaxation time (T1) and proton density (ρ) maps. These maps are then used to compute multi-TI images with arbitrary inversion times.ResultsSyMTIC was trained using paired magnetization prepared rapid acquisition with gradient echo (MPRAGE) and fast gray matter acquisition T1 inversion recovery (FGATIR) images along with T2-w and FLAIR images. It accurately synthesized multi-TI images from standard clinical inputs, achieving image quality comparable to that from explicitly acquired multi-TI data. The synthetic images, especially for TI values between 400 to 800 ms, enhanced visualization of subcortical structures and improved segmentation of thalamic nuclei.ConclusionSyMTIC enables robust generation of high-quality multi-TI images from routine MR contrasts. When paired with the HACA3 algorithm, it generalizes well to varied clinical datasets, including those without FLAIR or T2-w images and unknown parameters, offering a practical solution for improving brain MR image visualization and analysis.

  PublisherOA PDFDOI

• Construction of a normative database of human retinal thicknesses from existing UK Biobank OCT participants

  2026-04-01

  article
  PublisherDOI

• Proteomic Age Acceleration in Multiple Sclerosis Precedes Symptom Onset and Associates with Severity

  medRxiv · 2026-04-20

  articleOpen access

  Abstract Biological aging is accelerated in people with multiple sclerosis, but whether such acceleration occurs during the pre-symptomatic phase or varies by organ system is understudied. We analyzed two independent proteomics datasets profiled using distinct platforms: the Johns Hopkins cohort profiled using the SomaScan platform (348 multiple sclerosis/49 age-matched controls) and the Department of Defense cohort profiled using the Olink platform (134 multiple sclerosis/79 age-matched controls), including 117 pre-symptomatic samples from people with multiple sclerosis (median lead time: 4.0 years), to estimate systemic and organ-specific proteomic age gaps using established clocks in pre-symptomatic and symptomatic phases, and assess their associations with severity. In the Johns Hopkins cohort, people with multiple sclerosis demonstrated acceleration of systemic (β=2.2, 95% CI 1.2–3.2, P &lt;0.001, FDR&lt;0.001), brain (β=1.7, 95% CI 0.6–2.7, P =0.003, FDR=0.01), muscle (β=2.5, 95% CI 1.3–3.7, P &lt;0.001, FDR&lt;0.001), and immune age (β=1.8, 95% CI 0.6–2.9, P =0.003, FDR=0.01), with findings reproduced in the Department of Defense cohort for systemic (β=0.7, 95% CI 0.0–1.4, P =0.04, FDR=0.34) and brain age (3.2 years, 95% CI 2.1–4.3, P &lt;0.001, FDR&lt;0.001). Proteomic age acceleration was evident prior to symptom onset [systemic: (β=1.0, 95% CI 0.4–1.7, P =0.002, FDR=0.02); brain: (β=2.4, 95% CI 1.2–3.7, P &lt;0.001, FDR=0.002)], whereas no immune age acceleration was detected before or after onset. Higher systemic age gap was associated with greater global Age-Related Multiple Sclerosis Severity Score (β=0.14, 95% CI 0.05–0.24, P =0.005, FDR=0.03) and slower walking speed (β=0.02, 95% CI 0.01–0.03, P =0.006, FDR=0.04), while higher muscle age gap was associated with greater global Age-Related Multiple Sclerosis Severity Score (β=0.17, 95% CI 0.10–0.24, P &lt;0.001, FDR&lt;0.001), poorer manual dexterity (β=0.28, 95% CI 0.04–0.52, P =0.03, FDR=0.30), slower walking speed (β=0.02, 95% CI 0.01–0.03, P =0.002, FDR=0.02), lower peripapillary retinal nerve fiber layer (β= −0.26, 95% CI −0.41 to −0.10, P =0.001, FDR=0.02) and ganglion cell-inner plexiform layer thicknesses (β= −0.35; 95% CI −0.65 to −0.05; P =0.02, FDR=0.30). Higher brain age gap was associated with several imaging measures, including lower whole-brain (β= −0.002, 95% CI −0.003 to −0.001, P =0.002, FDR=0.02), and lower peripapillary retinal nerve fiber layer thickness (β= −0.21, 95% CI −0.39 to −0.03, P =0.02, FDR=0.10). Proteomic age acceleration in multiple sclerosis is detectable years before symptom onset and distinct organ-specific aging signatures are associated with disease severity. Proteomic aging may provide a biologically informative marker of early disease processes and a clinically relevant readout of disease heterogeneity.

  PublisherOA PDFDOI

• Pipeline refinement on diffusion tractography and T1 tractography in the presence of multiple sclerosis lesions

  2026-02-13

  articleSenior author
  PublisherDOI

Recent grants

• 3D segmentation and registration of macular SD-OCT for application in MS

  NIH · $1.6M · 2014–2019

• Tongue muscle function after cancer surgery using 4D MRI, DTI, and MR tagging

  NIH · $1.7M · 2015–2021

• NIH Grant R01HL047405

  NIH · $3.2M · 2010

• NIH Grant R01NS037747

  NIH · $2.7M · 2009

• NIH Grant R29HL047405

  NIH · $545k · 1999

Frequent coauthors

• Aaron Carass

  275 shared

• Peter A. Calabresi

  National Institutes of Health

  265 shared

• Shiv Saidha

  Johns Hopkins Medicine

  190 shared

• Jonghye Woo

  186 shared

• Dzung L. Pham

  Uniformed Services University of the Health Sciences

  179 shared

• Fangxu Xing

  Harvard University

  165 shared

• Daniel S. Reich

  National Institutes of Health

  115 shared

• Maureen Stone

  University of Washington

  110 shared

Education

• Ph.D., Electrical Engineering and Computer Science

  Massachusetts Institute of Technology

  1988

• S.M., Electrical Engineering and Computer Science

  Massachusetts Institute of Technology

  1982

• B.S., Electrical Engineering and Computer Science

  University of Connecticut

  1979

Awards & honors

• National Science Foundation Presidential Faculty Fellows Awa…
• Maryland's Outstanding Young Engineer (1997)
• MICCAI Society Enduring Impact Award (2012)
• 2025 DSAI Demonstration Project Award