Toward automated quality assurance for intensity- modulated radiation therapy

To investigate whether high-quality, relatively inexpensive, document and transparency scanners used as densitometers are sufficiently quantitative for routine quality assurance (QA).

The scanner we investigated used a linear amplifier, digitizing gray-scale images to 12-bit resolution with a user-selected spatial resolution of 0.170 mm(2) pixels. To reduce Newton's rings artifacts, the standard glass platen was replaced by glass with an antireflective coating. Conversion of reading to transmission was conducted by permanently placing a calibrated photographic step tablet on the scanner platen. After conversion to light transmission, a zero-phase two-dimensional Wiener filter was used to reduce pixel-to-pixel signal variation. Light-scatter artifacts were removed by deconvolution of a measured light-spread kernel. The light-spread kernel artifacts were significant along the scanner's detector axis, but were insignificant along the scanning axis.

Pixel-to-pixel noise was better than 2% for optical densities, ranging from 0.4 to 2.0 and 0 to 2.7 for the unfiltered and filtered images, respectively. The document scanning system response was compared against a confocal scanning laser densitometer. A series of IMRT dose distribution and dose calibration film sets were scanned using the two scanners, and the measured dose was compared. The maximum mean and standard deviation of the measured dose difference between the document scanner and confocal scanner was 1.48% and 1.06%, respectively.

While the document scanners are not as flexible as dedicated film densitometers, these results indicate that, using the intensity and scatter corrections, the system provides accurate and precise measurements up to an optical density of 2.0, sufficient for routine IMRT film QA. For some film types, this requires the reduction in monitor units to limit the dose delivered to the film. The user must be cautious that the delivered IMRT dose is scaled appropriately. This inexpensive and accurate system is being integrated into an automated QA program.