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

This work describes a rapid and objective method of determining repositioning error during the course of precision radiation therapy using off-line CT imaging and automated mutual-information image fusion. The technique eliminates the variability associated with manual identification of anatomical landmarks by observers. A phantom study was conducted to quantify the accuracy of the image co-registration-based analysis itself. For CT voxel dimensions of , the method is shown to detect translations with an accuracy of 0.5 mm in the anterior-posterior and lateral dimensions and 0.8 mm in the superior-inferior dimension. Phantom rotation in the coronal plane was detected to within 0.5° of expected values. The analysis has been applied to eight radiotherapy patients at two independent clinics, each immobilized by the same system for cranial stereotactic radiotherapy and CT-imaged once per week over the five- to six-week course of treatment. Among all patients, the ranges of translation in the anterior-posterior, lateral, and superior-inferior dimensions were , and , respectively. Considering all patients and CT scans, the standard deviations of translation were 0.42 mm, 0.47 mm, and 1.36 mm in the anterior-posterior, lateral, and superior-inferior dimensions, respectively. The ranges of patient rotation about the superior-inferior, left-right, and anterior-posterior axes were , and , respectively. PACS numbers: 87.53.-j, 87.53.Kn, 87.53.Ly, 87.53.Xd