ESTRO 2024 - Abstract Book

S5983

RTT - Treatment planning, OAR and target definitions

ESTRO 2024

Vickie Kong 1,2 , Tara Rosewall 1,2 , Alejandro Berlin 1,2 , Srinivas Raman 1,2 , Jennifer Dang 1 , Jeff Winter 1,2 , Peter Chung 1,2

1 Princess Margaret Cancer Centre, Radiation Medicine Program, Toronto, Canada. 2 University of Toronto, Department of Radiation Oncology, Toronto, Canada

Purpose/Objective:

Bladder cancer treatment involving whole-bladder irradiation and a simultaneous integrated boost (SIB) to the bladder tumor or the post-surgical tumor bed is currently being investigated. To improve precision, our institution has adopted the patient-specific planning target volume (PS-PTV) strategy, which requires manual delineation of bladder and post-surgical tumor bed on the first 4 Conebeam-CTs (CBCT). This process is time-consuming and increases workload. However, autosegmentation based on deformable image registration (DIR) between CBCT and CT may improve efficiency and reduce interobserver variability. This study aimed to assess the accuracy of DIR propagated SIB volume for bladder cancer radiotherapy.

Material/Methods:

The Institutional Quality Improvement Committee approved access to imaging data from 29 patients with bladder cancer who received treatment to the whole bladder (45-50Gy/20) and SIB to the post-surgical tumor bed (55Gy/20). Of the 29 patients, 23 were male and 6 were female, and all were treated with a comfortably full bladder. For each patient, two DIR-propagated SIBs were generated on each of the 4 CBCTs. The SIB_Intensity was generated based on DIR using image intensity alone, while the SIB_Bladder was created based on DIR with the inclusion of both image intensity and anatomical information from the manually delineated bladder as input. The DIR-propagated SIBs were compared to the corresponding manually delineated SIBs using the Dice Similarity Coefficient (DSC) and Mean Distance to Agreement (MDA). A DSC of < 0.8 and an MDA of > 2mm were considered unacceptable for the PS-PTV strategy. A two-tailed paired t-test was performed to compare the mean DSC and MDA of the SIB_Intensity and SIB_Bladder with the manually delineated SIB, with statistical significance defined as p < 0.05. The location of each SIB was categorized into 3 regions (Zone 1: Bladder dome, Zone 2: Lateral/Central wall and Zone 3: Trigone) to assess its association with DSC and MDA.

Results:

A total of 113 CBCTs were available to generate the DIR-propagated SIBs. The geometry of SIB_Intensity was significantly different from the manually delineated SIB, with a mean DSC of 0.71 (Range: 0.42 – 0.88), and 49% of MDA was >2mm. On the other hand, SIB_Bladder was more similar to the manually delineated SIB, achieving a significantly higher mean DSC of 0.83 (Range: 0.74 – 0.91, p < 0.05), with only 13% of MDA being >2mm. Out of 113 volumes, 78 SIB_Intensity volumes and 39 SIB_Bladder volumes had a DSC of <0.8. The performance of deformable contour propagation was worse when either the entire or a portion of SIB was located in the bladder dome, as indicated by a higher frequency of DSC < 0.8 and MDA > 2mm (Table 1). However, the difference between the manually delineated and either SIB_Intensity or SIB_Bladder was not significant for SIB located in the trigone region of the bladder, with DSC ranging from 0.82 – 0.91 and MDA ranging from 0.7 – 1.8mm.