Analysis of patient repositioning accuracy in precision radiation therapy using automated image fusion

Abstract

The accuracy of inter-fraction patient repositioning is essential to target coverage and sparing of normal structures in precision radiotherapy. This work describes a rapid and objective method of determining repositioning error with reference to an initial Computed Tomography (CT) simulation. The approach uses a mutual-information image fusion algorithm that is automated and contained within the treatment planning system (BrainSCAN 5.2, BrainLAB AG). The method eliminates the variability associated with observer-identification of anatomical landmarks and returns the three-dimensional repositioning error for any chosen point of interest within the anatomy. A phantom study is used to quantify the accuracy of the fusion-based analysis itself. For CT voxel dimensions of 0.65 ´ 0.65 ´ 1.0 mm3, the method is shown to be accurate to within 0.5 mm in anterior-posterior and lateral dimensions, and to within 0.8 mm in the superior-inferior dimension. The analysis has been applied to eight radiotherapy patients at two independent clinics, immobilized with the same system for cranial stereotactic radiotherapy (BrainLAB cranial SRT mask system, BrainLAB AG) and CT-scanned once per week during the course of treatment. Among all patients, the largest repositioning errors averaged over the course of treatment were 0.6 mm, 0.7 mm and 1.8 mm in the anterior-posterior, lateral and superior-inferior dimensions, respectively. Considering all patients and CT scans, the average repositioning errors were 0.14 mm, 0.06 mm and 0.01 mm in the anterior-posterior, lateral and superior-inferior dimensions, respectively.