Reproducibility of frame positioning for fractionated stereotactic radiosurgery

Fractionated stereotactic radiation therapy is a useful new approach for treating a number of intracranial neoplasms including meningiomas, pituitary adenomas, craniopharyngiomas, and recurrent gliomas. For the majority of these we employ a conventional fractionation scheme of 180 cGy per fraction for 25 to 30 fractions, using a modified Gill-Thomas-Cosman (GTC) relocatable frame to accommodate fractionated delivery. The GTC system uses a custom acrylic dental appliance to set the frame position and an occipital plate and Velcro straps fix the head in place. Dally reproducibility is evaluated through use of a 'depth helmet,' a plastic hemispherical shell containing 25 holes at regularly spaced intervals. The depth helmet attaches to the GTC frame and the distance from the shell to the patient's head is recorded at each of the 25 positions. This paper describes a new simplified approach to the quantitative assessment of day-to-day variability in head fixation using the depth helmet measurements. This approach avoids the need to try and decide on the relative merit of 25 numerical differences at each fitting and provides a straightforward mathematical and conceptual framework for the description of fit and clinical decision making. The mathematical analysis and computer program we have developed uses all 25 measurements to provide a single three-dimensional displacement vector as well as displacement values in the three principal patient dimensions. Measurements at each of the 25 depth helmet positions are automatically separated into three principal axes corresponding to the patients left/right (x), anterior/posterior (y), and superior/inferior (z) using the spherical relations: x = r sin(φ) cos(θ), y = r sin(φ) sin(θ), z = r cos(φ), where θ and φ are the polar and azimuthal angles respectively and r is the distance from the center of the depth helmet to the surface of the patient's head. For each patient, a set of initial measurements is taken at the CT scanner with the patient in the treatment (supine) position. Because treatment planning is based on the CT scan, this serves as the baseline from which subsequent deviations are recorded. In an analysis of our first 30 patients representing over 750 fractions, the mean RMS deviation, that is, the mean three-dimensional displacement from baseline, was 0.468 ± 0.296 mm. Among individual patients the range was 0.169 mm to 1.438 mm. A closer analysis suggests that in-plane (AP/PA-lateral) deviations occur randomly. Deviations along the superior/inferior direction are greater than those in-plane, and in several patients a small shift along this axis, possibly due to a loosening or stretching of the Velcro straps, has been noted over time. We have found our method to be a useful indicator of day-to-day reproducibility, allowing ready identification and correction of three-dimensional shifts relative to the patient axes. Based on our initial analysis, we can now define quantitative limits of acceptability in repositioning for subsequent fractionated delivery.