Calibration of an amorphous-silicon flat panel portal imager for exit-beam dosimetry

Amorphous-silicon flat panel detectors are currently used to acquire digital portal images with excellent image quality for patient alignment before external beam radiation therapy. As a first step towards interpreting portal images acquired during treatment in terms of the actual dose delivered to the patient, a calibration method is developed to convert flat panel portal images to the equivalent water dose deposited in the detector plane and at a depth of 1.5 cm. The method is based on empirical convolution models of dose deposition in the flat panel detector and in water. A series of calibration experiments comparing the response of the flat panel imager and ion chamber measurements of dose in water determines the model parameters. Kernels derived from field size measurements account for the differences in the production and detection of scattered radiation in the two systems. The dissimilar response as a function of beam energy spectrum is characterized from measurements performed at various off-axis positions and for increasing attenuator thickness in the beam. The flat panel pixel inhomogeneity is corrected by comparing a large open field image with profiles measured in water. To verify the accuracy of the calibration method, calibrated flat panel profiles were compared with measured dose profiles for fields delivered through solid water slabs, a solid water phantom containing an air cavity, and an anthropomorphic head phantom. Open rectangular fields of various sizes and locations as well as a multileaf collimator-shaped field were delivered. For all but the smallest field centered about the central axis, the calibrated flat panel profiles matched the measured dose profiles with little or no systematic deviation and approximately 3% (two standard deviations) accuracy for the in-field region. The calibrated flat panel profiles for fields located off the central axis showed a small -1.7% systematic deviation from the measured profiles for the in-field region. Out of the field, the differences between the calibrated flat panel and measured profiles continued to be small, approximately 0%-2% of the mean in-field dose. Further refinement of the calibration model should increase the accuracy of the procedure. This calibration method for flat panel portal imagers may be used as part of a validation scheme to verify the dose delivered to the patient during treatment.