Cardiovascular Computed Tomography Imaging

Modern CT scanners image the entire heart throughout the cardiac cycle with high spatial resolution. Our group uses this capability to measure regional left and right ventricular function, to reduce motion artifacts, and to estimate blood flow from routine contrast-enhanced scans.

We also work to improve how CT detects and characterizes coronary artery disease.

Blood Flow Estimation from CT

Contrast-enhanced CT captures how contrast moves through the heart and vessels, but CT is not conventionally used to measure blood flow. We use physics-informed neural networks (PINNs) to estimate flow fields directly from CT sinograms and images, and we study how scanner parameters such as gantry rotation speed, X-ray fluence, and pulsed acquisition affect accuracy and dose.

With collaborators, we have applied dynamic contrast 4DCT to measure left atrial appendage blood transport in patients with atrial fibrillation, a step toward image-based estimates of thrombosis risk.

Guo et al., Medical Physics 2026, Figure 3

Severance et al., Journal of Cardiovascular Computed Tomography 2026, Figure 1

Recent publications include:



Preprints & conference abstracts:


Sinogram-based flow estimation in CT: impact of X-ray fluence and pulsed acquisition on accuracy and dose. Guo J, Gonzalo Grande A, del Álamo JC, Contijoch F. SPIE Medical Imaging 2026 (oral)

Physics-informed neural networks for estimating fluid flow from CT sinograms. Khurana G, Guo J, et al. 8th International Conference on Image Formation in X-ray CT, 2024 (two abstracts)

Reducing Motion Artifacts in Cardiac CT

Even fast CT scanners need a few hundred milliseconds to acquire an image, and heart motion during that window creates artifacts that can mimic or hide disease. We characterize where these artifacts come from and develop reconstruction methods to resolve them.

DiFiR-CT represents moving cardiac structures with neural signed distance functions, which allows motion-corrected reconstruction directly from the acquired projections. We are also developing data-consistency metrics that automatically map and score motion-induced errors.

Gupta et al., Medical Physics 2023, Figure 6

Recent publications include:



Preprints & conference abstracts:

Automatic quantification and mapping of temporal inconsistencies that lead to motion artifacts: a pilot study. Craine A, McVeigh E, Contijoch F. SCCT 2026 Annual Scientific Meeting (poster)

Regional Estimates of Cardiac Function using cineCT

Low-dose CT acquired across the cardiac cycle allows visualization of both ventricles in 3D. We combine these acquisitions with SQUEEZ, our endocardial surface-tracking method, to map regional strain and shortening of the left and right ventricles.

We have used these maps to detect wall motion abnormalities, to follow patients after structural heart interventions, and to evaluate right ventricular performance in heart failure, pulmonary hypertension, and congenital heart disease.

Both ventricles can be visualized throughout the cardiac cycle

Regional estimates of RV strain in an adult patient with repaired Tetralogy of Fallot

Craine et al., European Heart Journal - Imaging Methods and Practice 2026, Figure 2