Correction of systematic setup errors in prostate radiation therapy: How many images to perform?

The purpose of this study was to develop an evidence-based off-line setup correction protocol for systematic errors in prostate radiation therapy. Daily orthogonal electronic portal images were acquired from 30 patients. Field displacements were measured in the medial-lateral (ML), superior-inferior (SI), and anterior-posterior (AP) directions for each treatment fraction. The off-line protocol corrects the mean field displacement found from n consecutive images, starting at a particular fraction of treatment, with a fixed tolerance level. Simulations were performed with the measured data to determine (1) how many images (n) should be averaged to determine the systematic error; (2) on which treatment fraction should the protocol be initiated; and (3) what tolerance level should be applied to determine whether the patient position should be corrected. Uncorrected systematic errors in the ML, SI, and AP directions were (mean position ± 1 standard deviation [SD]): −0.7 ± 2.2 mm, −1.5 ± 1.3 mm, and 1.4 ± 2.6 mm, respectively. Random errors (1 SD and range) were 1.9 mm (1.3-3.3), 1.5 mm (0.-4.1), and 1.8 mm (1.0-2.6), respectively. A correction based on a single image taken on the first fraction actually increased the systematic errors in the ML and SI directions compared with no correction. More accurate correction of systematic errors was achieved with increasing number of images averaged, with only small benefit after 5 images. With fewer images averaged, delaying the start of the protocol resulted in more accurate correction because of the influence of unrepresentative positions at early fractions. The number of corrections made on patients with small (<2 mm) systematic errors was minimized for tolerance values of 2 mm and n ≥ 5 images averaged. The optimal off-line setup correction protocol would be to shift the patient by the mean displacement of the first 5 portal images of a radical course of radiation therapy. A small tolerance level should be utilized with 2 mm giving good accuracy with minimal unnecessary shifts.