Quantification of Cardiac Substructure Inter-fraction Displacement for MR-guided Radiation Therapy

Document Type

Conference Proceeding

Publication Date


Publication Title

International Journal of Radiation Oncology Biology Physics


Background: Radiation therapy (RT) dose to cardiac substructures has been linked to toxicities including coronary heart disease and heart failure. However, cardiac substructures are poorly visualized on traditional non-contrast treatment planning computed tomography CT) images due to limited soft tissue contrast.

Objectives: We sought to quantify inter-fraction displacements of cardiac substructures using daily magnetic resonance (MR) guided RT datasets that can be applied for robust per-structure margin design. Methods: Nine cases with 11 non-breast lesions in the thoracic region [lung, mediastinum, and esophagus] were retrospectively analyzed. All patients underwent daily MR-guided RT using a 0.35T MR-linac (6 end-inhale, 2 end-exhale using a 17-25 s TrueFISP scan, 1.5×1.5×3 mm3 resolution, 1 free breathing (∼3-minute TrueFISP scan, resolution= 1.5×1.5×1.5 mm3). To elucidate cardiac substructures, MR datasets (0.35T) for simulation (MR-SIM) and treatment fractions at the same breath-hold position were rigidly co-registered to the planning CT using bony alignment. A hybrid diagnostic MR/CT atlas was used to propagate contours for 12 cardiac substructures (e.g. chambers, coronary arteries (CA), and great vessels) to all serial MR-SIM and 3-4 fractions for each case (n=35) that were then modified by two radiation oncologists as needed. After performing a tumor-based image registration to the initial MR-SIM, inter-fraction differences were quantified via centroid analysis. Dominant axes of displacement were identified for each substructure.

Results: Over all fractions, the average inter-fraction shift of the heart from the MR-SIM was 2.3 ± 1.5 mm, 1.7 ± 1.6 mm and 2.9 ± 2.2 mm in the left-right (L-R), anterior-posterior (A-P), and superior-inferior (S-I) directions, respectively with an overall vector displacement of 4.6 ± 2.3 mm. In the L-R direction, the left atrium had the lowest mean shift (2.2 ± 1.8 mm) with the heart chambers ranging between 2.7-3.1 ± 2 mm. Great vessels (i.e., ascending aorta, superior vena cava and pulmonary vein) with a shift of 1.7 ± 1.2 mm had the minimal displacement across fractions in the A-P axis; whereas the right ventricle shifted the least (2.6 ± 2 mm) followed by other heart chambers (3.1± 2 mm) in the S-I direction. The maximum inter-fraction vector displacement occurred for the left anterior descending (8.7 ± 4.2 mm), left main (6.9 ± 2.5 mm) and right (6.8 ± 2.8 mm) CA. For two patient fractions, SI deviations were >1 cm for most substructures due to lack of compliance with breath-hold conditions.

Conclusions: Cardiac substructure displacements demonstrated variability in magnitude and dominant axis, suggesting that anisotropic substructure-specific planning organ at risk margins may be warranted. While these results suggest patient-specific margins may be necessary, more precise margin definition requires confirmation in a larger cohort and accounting for other uncertainties.





First Page


Last Page


This document is currently not available here.