Session Item

In proton therapy of lung cancer patients, the need for plan adaptation is monitored by weekly repeat CTs (reCTs). To ease the clinical workflow, daily cone beam CTs (CBCTs), used for patient setup, could be used for this purpose. However, the image quality of CBCTs is typically insufficient for accurate proton dose computation. Currently, several methods exist for image quality improvement of CBCTs. In this study, we will evaluate the clinical applicability of CBCT-based synthetic-CTs (sCTs) for monitoring the need for adaptation.

Forty-two lung cancer patients treated with intensity modulated proton therapy were retrospectively included, consisting of three groups: 1) 16 patients with no anatomical changes and no adaptation, 2) 11 patients with anatomical changes but no clinical adaptation, and 3) 15 patients with anatomical changes and a plan adaptation. All patients had a reCT and a same-day CBCT. Two commercial solutions for CBCT-based sCT generation were applied (RayStation 11B, RaySearch Laboratories). The first method was an iterative CBCT correction algorithm, which uses image correlation to the planning CT (pCT) to correct the CBCT numbers and remove low-frequency artifacts (Cor-sCT). The second method applies deformable image registration (DIR) between the pCT and CBCT to deform the pCT to the daily anatomy of the CBCT and additionally corrects the anatomy for large differences in air/lung regions between the deformed pCT and the CBCT (DIR-sCT).

A plan re-evaluation was performed on the reCT and the two sCTs according to our clinical adaptive protocol: First, the contours were deformably propagated from the pCT to the reCT/sCTs, whereafter the clinical plan was robustly evaluated on the reCT/sCTs. The adaptation criteria were based on the clinical target volume (CTV) coverage (V95%≥95%) and maximum dose to the mediastinum and spinal cord. We compared adaptation decisions made based on the reCT and the sCT. If the plan violated clinical constraints on the reCT but was acceptable on the sCT a false negative was scored.

Figure 1 shows an example of the pCT, reCT, CBCT and the two sCTs for a patient with anatomical changes (group 3). For this patient, an adaptation was needed due to target under-dosage on the reCT as well as both sCTs (true positive). Only one and two patients were scored as false negatives, and five and four patients were scored as false positives for the Cor-sCT and DIR-sCT, respectively (Figure 2). The false negative and false positive patients were caused by minor differences in CTV coverage (just around the constraint), or anatomical differences between reCT and CBCT.

The results regarding adaptation decision that were obtained from the Cor-sCT and DIR-sCT were both consistent with the reCT in the majority (86%) of the patients. The false negative rate was low and could be explained. Therefore, the quality of both synthetic CT methods is suitable for use in clinical evaluations to determine the need for plan adaptation.