Quantifying Cardiac Deformable Image Registration Accuracy and Its Dosimetric Uncertainty for 4D Dose Accumulation In Stereotactic Arrhythmia Radioablation
Abstract
Purpose
Stereotactic arrhythmia radioablation (STAR) offers a non-invasive treatment option for refractory tachycardia; however, precise dose delivery remains challenging due to the complexity of cardiorespiratory motion. This study evaluated the geometric and dosimetric performance of multiple deformable image registration (DIR) algorithms using ECG-gated four-dimensional CT (ECG-4DCT) in both phantom and clinical datasets.
Methods
ECG-4DCT data from the XCAT phantom and 20 patients were analyzed across ten cardiac phases using six DIR algorithms; an additional deep learning–based algorithm, TransMorph, was also included for comparison with the clinical datasets. Registration accuracy was assessed using the Dice similarity coefficient (DSC), Hausdorff distance (HD95), and average surface distance (ASD), while dosimetric accuracy was evaluated using dosimetric metrics and γ analysis. Four-dimensional dynamic dose (4DDD) uncertainty was quantified using the coefficient of variation (CV) and maximum pairwise absolute dose difference (MPADD).
Results
Registration accuracy was lowest between end-systolic and end-diastolic phases in both phantom and clinical datasets. In the phantom study, MIM achieved the highest γ passing rate (89.6% at 1%/1 mm) and the smallest deviation from the reference deformation vector fields, with differences of −0.02 Gy in D95 and −0.2% in V25. In the clinical datasets, patients without metallic implants exhibited reduced geometric accuracy and increased 4DDD uncertainty (mean CV = 0.01 for V25 and 0.02 for D95; MPADD up to 11.3% and 1.15 Gy). TransMorph achieved the highest geometric accuracy, with mean DSC values of 0.86 in patients without implants and 0.89 in those with implants; however, this improved geometry was accompanied by steeper deformation gradients and pronounced localized dose discrepancies.
Conclusion
This study demonstrates that existing DIR algorithms remain limited in capturing complex cardiac motion. Furthermore, geometric accuracy alone does not guarantee physiologically valid deformation in STAR. Cardiac-specific, physiology-constrained DIR frameworks are therefore required to achieve robust and clinically reliable 4DDD evaluation.