Impact of deformable image registration on dose accumulation applied electrocardiograph-gated 4DCT in the heart and left ventricular myocardium during esophageal cancer radiotherapy

Background: The deformable image registration (DIR) technique has the potential to realize the dose accumulation during radiotherapy. This study will analyze the feasibility of evaluating dose-volume parameters for the heart and left ventricular myocardium (LVM) by applying DIR. Methods: The electrocardiograph-gated four-dimensional CT (ECG-gated 4DCT) data of 21 patients were analyzed retrospectively. The heart and LVM were contoured on 20 phases of 4DCT (0%, 5%, ... , 95%). The heart and LVM in the minimum volume/dice similarity coefficient (DSC) phase (Volume (min)/DSC (min)) were deformed to the maximum volume/DSC phase (Volume (max)/DSC (max)), which used the intensity-based free-form DIR algorithm of MIM software. The dose was deformed according to the deformation vector. The variations in volume, mean dose (D-mean), V-20, V-30 and V-40 for the heart and LVM before and after DIR were compared, and the reference phase was the Volume (max)/DSC (max) phase. Results: For the heart, the difference between the pre- and post-registration Volume min and Volume (max) were reduced from 13.87 to 1.72%; the DSC was increased from 0.899 to 0.950 between the pre- and post-registration DSC min phase relative to the DSC (max) phase. The post-registration D-mean, V-20, V-30 and V-40 of the heart were statistically significant compared to those in the Volume (max)/DSC (max) phase (p < 0.05). For the LVM, the difference between the pre- and post-registration Volume min and Volume (max) were only reduced from 18.77 to 17.38%; the DSC reached only 0.733 in the post-registration DSC min phase relative to the DSC (max) phase. The pre- and post-registration volume, D-mean, V-20, V-30 and V-40 of the LVM were all statistically significant compared to those in the Volume (max)/DSC (max) phase (p < 0.05). Conclusions: There was no significant relationship between the variation in dose-volume parameters and the variation in the volume and morphology for the heart; however, the inconsistency of the variation in the volume and morphology for the LVM was a major factor that led to uncertainty in the dose-volume evaluation. In addition, the individualized local deformation registration technology should be applied in dose accumulation for the heart and LVM.