Erring Characteristics of Deformable Image Registration-Based Auto-Propagation for Internal Target Volume in Radiotherapy of Locally Advanced Non-Small Cell Lung Cancer

Abstract

Purpose: Respiratory motion of locally advanced non-small cell lung cancer (LA-NSCLC) adds to the challenge of targeting the disease with radiotherapy (RT). One technique used frequently to alleviate this challenge is an internal gross tumor volume (IGTV) generated from manual contours on a single respiratory phase of the 4DCT via the aid of deformable image registration (DIR)-based auto-propagation. Through assessing the accuracy of DIR-based auto-propagation for generating IGTVs, this study aimed to identify erring characteristics associated with the process to enhance RT targeting in LA-NSCLC.

Methods: 4DCTs of 19 patients with LA-NSCLC were acquired using retrospective gating with 10 respiratory phases (RPs). Ground-truth IGTVs (GT-IGTVs) were obtained through manual segmentation and union of gross tumor volumes (GTVs) in all 10 phases. IGTV auto-propagation was carried out using two distinct DIR algorithms for the manually contoured GTV from each of the 10 phases, resulting in 10 separate IGTVs for each patient per each algorithm. Differences between the auto-propagated IGTVs (AP-IGTVs) and their corresponding GT-IGTVs were assessed using Dice coefficient (DICE), maximum symmetric surface distance (MSSD), average symmetric surface distance (ASSD), and percent volume difference (PVD) and further examined in relation to anatomical tumor location, RP, and deformation index (DI) that measures the degree of deformation during auto-propagation. Furthermore, dosimetric implications due to the analyzed differences between the AP-IGTVs and GT-IGTVs were assessed.

Results: Findings were largely consistent between the two algorithms: DICE, MSSD, ASSD, and PVD showed no significant differences between the 10 RPs used for propagation (Kruskal-Wallis test, ps > 0.90); MSSD and ASSD differed significantly by tumor location in the central-peripheral and superior-inferior dimensions (ps < 0.0001) while only in the central-peripheral dimension for PVD (p < 0.001); DICE, MSSD, and ASSD significantly correlated with the DI (Spearman's rank correlation test, ps < 0.0001). Dosimetric assessment demonstrated that 79% of the radiotherapy plans created by targeting planning target volumes (PTVs) derived from the AP-IGTVs failed prescription constraints for their corresponding ground-truth PTVs.

Conclusion: In LA-NSCLC, errors in DIR-based IGTV propagation present to varying degrees and manifest dependences on DI and anatomical tumor location, indicating the need for personalized consideration in designing RT internal target volume.

Keywords: 4DCT; ITV generation; LA-NSCLC; auto-propagation; deformable propagation.

Figure 1

Flowchart with overview of study procedures. Abbreviations: 4D computed tomography (4DCT), auto-propagated internal gross tumor volume (AP-IGTV), auto-propagated planning target volume (AP-PTV), average symmetric surface distance (ASSD), Dice coefficient (DICE), dose volume histogram (DVH), ground-truth gross tumor volume (GT-GTV), ground-truth internal gross tumor volume (GT-IGTV), ground-truth planning target volume (GT-PTV), locally advanced non-small cell lung cancer (LA-NSCLC), maximum symmetric surface distance (MSSD), percent volume difference (PVD), respiratory phase (RP), and volumetric modulated arc therapy (VMAT).

Figure 2

CT cross sections of one locally advanced non-small cell lung cancer (LA-NSCLC) lesion representative of the study cohort in axial (A), sagittal (B), and coronal (C) planes with variation in auto-propagated internal gross tumor volumes (AP-IGTV) derived from the two employed deformable image registration (DIR) algorithms—MIM (outlined in red) and Velocity (outlined in blue)—compared to the ground-truth internal gross tumor volume (GT-IGTV; outlined in thick yellow).

Figure 3

Histogram of Dice coefficient (DICE), maximum symmetric surface distance (MSSD), average symmetric surface distance (ASSD), and percent volume difference (PVD) from pairwise comparisons of all 190 auto-propagated internal gross tumor volumes (AP-IGTVs) relative to their corresponding ground-truth internal gross tumor volumes (GT-IGTVs) for the two employed deformable image registration (DIR) algorithms (MIM in red; Velocity in blue).

Figure 4

Boxplots comparing accuracies of auto-propagated internal gross tumor volumes (AP-IGTVs) from the two employed deformable image registration (DIR) algorithms—MIM (red) and Velocity (blue)—relative to their corresponding ground-truth internal gross tumor volumes (GT-IGTVs) in terms of Dice coefficient (DICE), maximum symmetric surface distance (MSSD), average symmetric surface distance (ASSD), and percent volume difference (PVD) between different respiratory phases for the study cohort. On each box, the central mark indicates the median, and the top and bottom edges of the box indicate the 25th and 75th percentiles, respectively.

Figure 5

Box plots comparing accuracies of auto-propagated internal gross tumor volumes (AP-IGTVs) from the two employed deformable image registration (DIR) algorithms—MIM (red) and Velocity (blue)—relative to their corresponding ground-truth internal gross tumor volumes (GT-IGTVs) in terms of Dice coefficient (DICE), maximum symmetric surface distance (MSSD), average symmetric surface distance (ASSD), and percent volume difference (PVD) along the central–peripheral, and superior–inferior dimensions. On each box, the central mark indicates the median, and the top and bottom edges of the box indicate the 25th and 75th percentiles, respectively. For each box pair comparison, NS: not significant at level of 0.05; *: p < 0.05; **: p < 0.01; ***: p < 0.001; ****: p < 0.0001.

Figure 6

Scatter plots with linear regression (95% confidence intervals of the predicted mean) outlining the distribution of Dice coefficient (DICE), maximum symmetric surface distance (MSSD), average symmetric surface distance (ASSD), and percent volume difference (PVD) versus the deformation index for the two employed deformable image registration (DIR) algorithms—MIM (red) and Velocity (blue). Each dot corresponds to a different single respiratory phase for one individual patient.

Figure 7

Top: dose volume histograms (n = 19) of ground-truth planning target volumes (GT-PTVs) created with volumetric modulated arc therapy (VMAT) plan targeting corresponding auto-propagated planning tumor volume (AP-PTV). All plans met dose prescription of V_100% ≥ 95% of AP-IGTV and nearby organ at risk (OAR) constraints. Each line colored according to auto-propagated internal gross target volume (AP-IGTV) deformation index. Bottom: scatter plot of V_100% values of GT-PTVs as a function of deformation index of the corresponding AP-IGTV. Dots in green above the blue dashed line met planning objective of V_100% ≥ 95%, whereas dots in red below the blue dashed line failed to meet the constraint. Size of the dots is proportional to the size of ground-truth internal gross target volume (GT-IGTV).