Dosimetric- and Geometric Evaluation of Adaptive H&N IMRT Using Deformable Image Registration

Purpose/Objective: Anatomical changes can occur during RT treatment of H&N cancer patients. This can lead to a difference between planned- and delivered dose. Adaptive RT has the potential to overcome this, utilizing deformable image registration (DIR). The purpose of this study was to evaluate the performance of a DIR algorithm, using geometric and dosimetric measures.

Materials and Methods: Seven patients treated with IMRT were included in this study, each with a planning- and midterm CT (pCT, ReCT) as well as a CBCT, acquired on the same day as the ReCT. ReCT served as the ground truth for evaluation of the DIR. A deformed CT (dCT) with contours was created by deforming the pCT and associated manually drawn contours to the CBCT.

A commercially software using the Demons DIR algorithm (SmartAdapt, Varian Medical Systems, v.11.0) was utilized. A geometrical comparison was based on the estimated volumes from the contours on the dCT, and the manually drawn contours on the ReCT. Center of mass shifts (CMS) and dice similarity coefficients (DSC) were found between contours on dCT and ReCT.

In the treatment planning system (Eclipse, Varian Medical system, v.10.0) the initial treatment plan was copied to the dCT and ReCT and the dose recalculated. DVH points (D50 for parotid glands and Dmax for spinal cord) were evaluated. Conformity index (CI), lesion coverage fraction (LCF) and normal tissue overdose fraction (NTOF) was evaluated with regard to target coverage.

Results: The PTV volume was estimated larger for dCT than ReCT with a median of 4.4% (range -4.0; 69.3). In four of seven patients, the volume difference was <5%. Six patients had a median CMS for PTV of 0.28 cm (range 0.05; 0.43). The median DSC was 0.88 (range 0.60; 0.95). Similar results were obtained for GTV and CTV.

The median relative volume deviation from ReCT was -11.2% (range - 28.0; 16.7), -26.2% (range -42.1; 3.4) and -10.9% (range -33.3; 32.3) for parotid dxt, parotid sin and spinal cord, respectively. The median CMS was 0.51 cm (range 0.19; 2.22). DSC had a median of 0.47 (range 0.45; 0.85). The median relative deviation from ReCT in DVH points for parotid dxt, parotid sin and spinal cord was 8.3% (range -8.4; 25.3), -12.7% (range -28.6; 31.0), and 1.3% (range -5.4; 31.8), respectively.

CI, LCF and NTOF are visualized in the  figure. Ideal values of CI and LCF are unity and zero for NTOF. Conclusions: The DIR produced geometrical results similar to the ReCT in four of seven patients with regard to the target. Larger geometrical variations were observed for organs at risk (OAR). OAR contours obtained with the DIR were for nearly all patients estimated smaller than in the ReCT whereas target contours were estimated larger.

The dosimetric results for OAR showed some variation between dCT and ReCT, especially for the parotid glands. The LCF were similar for dCT and ReCT, whereas NTOF were larger for ReCT than for dCT. Despite variation in volume and dose, between dCT and ReCT, the differences were within acceptable limits for most of the patients.