A novel concept to include uncertainties in the evaluation of stereotactic body radiation therapy after 4D dose accumulation using deformable image registration

Abstract

Purpose To use four‐dimensional (4D) dose accumulation based on deformable image registration (DIR) to assess dosimetric uncertainty in lung stereotactic body radiation therapy (SBRT) treatment planning. A novel concept, the Evaluation Target Volume (ETV), was introduced to achieve this goal. Methods The internal target volume (ITV) approach was used for treatment planning for 11 patients receiving lung SBRT. Retrospectively, 4D dose calculation was done in Pinnacle v9.10. Total dose was accumulated in the reference phase using DIR with MIM. DIR was validated using landmarks introduced by an expert radiation oncologist. The 4D and three‐dimensional (3D) dose distributions were compared within the gross tumor volume (GTV) and the planning target volume (PTV) using the D95 and Dmin (calculated as Dmin,0.035cc) metrics. For lung involvement, the mean dose and V20, V10, and V5 were used in the 3D to 4D dose comparison, and Dmax (D0.1cc) was used for all other organs at risk (OAR). The new evaluation target volume (ETV) was calculated by expanding the GTV in the reference phase in order to include geometrical uncertainties of the DIR, interobserver variability in the definition of the tumor, and uncertainties of imaging and delivery systems. D95 and Dmin,0.035cc metrics were then calculated on the basis of the ETV for 4D accumulated dose distributions, and these metrics were compared with those calculated from the PTV for 3D planned dose distributions. Results The target registration error (TRE) per spatial component was below 0.5 ± 2.1mm for all our patients. For five patients, dose degradation above 2% (>4% in 2 patients) was found in the PTV after 4D accumulation and attributed to anatomical variations due to breathing. Comparison of D95 and Dmin,0.035cc metrics showed that the ETV (4D accumulated dose) estimated substantially lower coverage than the PTV (3D planning dose): in six out of the 11 cases, and for at least for one of the two metrics, coverage estimated by ETV was at least 5% lower than that estimated by PTV. Furthermore, the ETV approach revealed hot and cold spots within its boundaries. Conclusions A workflow for 4D dose accumulation based on DIR has been devised. Dose degradation was attributed to respiratory motion. To overcome limitations in the PTV for the purposes of evaluating DIR‐based 4D accumulated dose distributions, a new concept, the ETV, was proposed. This concept appears to facilitate more reliable dose evaluation and a better understanding of dosimetric uncertainties due to motion and deformation.

Using 4D dose accumulation to calculate organ‐at‐risk dose deviations from motion‐synchronized liver and lung tomotherapy treatments

“…The results showed that the dose to the GTV is maintained more closely than the PTV, which is expected as the purpose of the PTV is to expand the GTV to account for uncertainties and ensure the prescription is delivered to the GTV 30 . Underdosings of the target (1% for GTV and ∼4% for PTV) have been reported for other studies comparing 4D‐accumulated, motion‐compensated tracking doses to planned dose 10,31 . The differences in target dose may be from the dose deformation effect, 11,31–33 the unflattened profile of the Radixact beam (causing maximum decrease in output of 3%–4% for the 1‐cm jaw and 1%–2% for the 2.5‐cm jaw as the target moves off‐axis), 3,7 the discrete nature of MLC compensation (which allows for axial motion less than ±3.125 cm to be uncompensated), and uncertainties in the framework.…”

“…For the example cases in this work, DIR was performed in MIM, which uses an intensity‐based free‐form deformation algorithm with a sum of squared differences as a similarity measurement 20 . The resolution of all 4DCTs was 1 × 2 × 1 mm 3 or smaller and each registration for all subjects was reviewed and refined with the Reg Reveal and Reg Refine tools in MIM, which allow the user to modify and improve the quality of the registration in different regions 10 …”

“…The recommendations of Task Group 132 should be followed to verify the accuracy of the DIR, which includes commissioning/verification of the algorithm and patient‐specific verification of each registration 21 . The MIM DIR algorithm used for the examples in this work has been verified in the literature 10,20,22,23 . Azcona et al.…”

“…The deformation information is commonly obtained using four‐dimensional computed tomography (4DCT) datasets. Methods of dose accumulation with 4DCT datasets have been published in the literature, 10,11 but not applied to Radixact Synchrony deliveries or used to calculate OAR dose deviations. Chao et al.…”

Integrating CVH and LVH metrics into an optimization strategy for the selection of Iris collimator for Cyberknife Xsight lung tracking treatment

“…The CVH/ LVH evaluation needs to be replaced using DVH/normal tissue complication probabilities models in order to evaluate the actual dose delivered to the CTV and surrounding lung tissues. The entire treatment delivery process needs to be taken into account in the temporal interplay between CTV and the temporal aspect of actual dose delivery 25–27 . Additionally, the PTV border volume, which is usually lung tissues rather than tumor tissues, used to compensate for treatment uncertainties, the Monte Calro dose calculation accuracy is needed.…”