Author: Carri K. Glide-Hurst, Thomas M Grist, Kevin M. Johnson, Prashant Nagpal, Tarun Naren, Chase Ruff, Oliver Wieben, Jiwei Zhao 👨🔬
Affiliation: Department of Biostatistics and Medical Informatics, University of Wisconsin - Madison, Department of Radiology, University of Wisconsin-Madison, Department of Medical Physics, University of Wisconsin-Madison, Departments of Human Oncology and Medical Physics, University of Wisconsin-Madison 🌍
Purpose: Cardiotoxicity is a devastating side effect for thoracic radiotherapy (RT). Currently, standard RT imaging is insufficient to decouple cardiorespiratory motion, limiting substructure-specific margin assessment to yield improved sparing strategies. We demonstrate comprehensive substructure-specific motion analysis via a novel 5D-MRI workflow, for future extension to margin calculation.
Methods: 5D-MRI datasets were acquired for 10 healthy subjects (23-65 years old), using a bSSFP sequence with radial sampling and scan time <5 minutes. Reconstructions were performed for 10 cardiac/4 respiratory phases (end/active exhale/inhale) via offline, iterative reconstruction methods, using retrospective gating with respiratory bellows and pulse oximeter signals. Coronary arteries (CAs), chambers, and great vessels (GVs) were delineated (40 image volumes/subject) and verified by a cardiovascular radiologist. Regions corresponding to conduction nodes and valves were delineated via geometric models. Centroid/bounding box excursion was calculated for cardiac, respiratory, and hysteresis motion, and statistical testing via the Kruskal-Wallis test was performed for 8 distance metrics across substructure pairings.
Results: 5D-MRI datasets were of sufficient quality for contouring and motion analysis in all subjects. CAs had the greatest cardiac motion, followed by chambers and GVs. Right heart substructures had cardiac bounding box motion >1cm. Cardiac excursion was dominant in the left, anterior, and inferior directions and was highly substructure-dependent, while respiratory motion was greatest in the inferior direction, >5mm for all substructures, and less substructure-dependent. Hysteresis motion exceeded 5mm for three subjects. Statistically significant differences (p<0.05) were observed between CAs/chambers/GVs and right/left heart for cardiac motion, and between base/apex for respiratory motion.
Conclusion: Cardiorespiratory motion was successfully decoupled and quantified via 5D-MRI. Cardiac motion was >5mm for the CAs/chambers, and respiratory motion was >5mm for all substructures. Significant differences (p<0.05) in cardiac/respiratory motion between substructures were observed. Future work includes extension to thoracic cancer patient cohorts for comprehensive substructure-specific margin evaluation for advanced cardiac sparing.