Search

In photon teletherapy, the size and functional form of the photon source spot affect both the sharpness of the penumbra of treatment fields and the sharpness of portal images. Photon source spot parameters are also used in photon teletherapy dose calculation codes. A simple method for characterizing the source spot would complement the existing, more involved methods that have been described in the medical physics literature. Such a method, using a rollbar made of tungsten or other high-Z metal, is used in industrial radiography. We describe the use of a tungsten rollbar for characterizing the source spot edge spread function (and thereby the source spot size and shape) of a megavoltage bremsstrahlung photon source. We use Monte Carlo simulations to quantify anticipated experimental artifacts of the method, assuming typical spot sizes for circ-function, Gaussian, and Bennett line shapes. We illustrate the use of the rollbar method by characterizing the source spot of a typical 9 MV linac used for industrial radiography. The source spot is analyzed using two approaches: (a) fitting the rollbar image with analytic functions and (b) using Abel inversion to obtain the cylindrically symmetric spot profile consistent with the measured rollbar image. Monte Carlo simulations, based on a 6 MV photon teletherapy accelerator, suggest that aspects of the method are applicable to medical bremsstrahlung sources.

Originally designed for use at medical-imaging x-ray energies, imaging systems comprising scintillating screens and amorphous Si detectors are also used at the megavoltage photon energies typical of portal imaging and industrial radiography. While image blur at medical-imaging x-ray energies is strongly influenced both by K-shell fluorescence and by the transport of optical photons within the scintillator layer, at higher photon energies the image blur is dominated by radiation scattered from the detector housing and internal support structures. We use Monte Carlo methods to study the blurring in a notional detector: a series of semi-infinite layers with material compositions, thicknesses, and densities similar to those of a commercially available flat-panel amorphous Si detector system comprising a protective housing, a gadolinium oxysulfide scintillator screen, and associated electronics. We find that the image blurring, as described by a point-spread function (PSF), has three length scales. The first component, with a submillimeter length scale, arises from electron scatter within the scintillator and detection electronics. The second component, with a millimeter-to-centimeter length scale, arises from electrons produced in the front cover of the detector. The third component, with a length scale of tens of centimeters, arises from photon scatter by the back cover of the detector. The relative contributions of each of these components to the overall PSF vary with incident photon energy. We present an algorithm that includes the energy-dependent sensitivity and energy-dependent PSF within a ray-tracing formalism. We find quantitative agreement (approximately 2%) between predicted radiographs with radiographs of copper step wedges, taken with a 9 MV bremsstrahlung source and a commercially available flat-panel system. The measured radiographs show the blurring artifacts expected from both the millimeter-scale electron transport and from the tens-of-centimeters length scale arising from the scattered photon transport. Calculations indicate that neglect of the energy-dependent blurring would lead to discrepancies in the apparent transmission of these wedges of the order of 9%.

We have developed a new tomography code, HADES-CCG. This code uses HADES, a radiographic simulation code, to perform forward- and back-projection and is coupled to a Constrained Conjugate Gradient (CCG) optimizer. An iterative solution to the reconstruction problem is found which is optimal, given the detector noise model, a source model and the appropriate attenuation cross-sections. By explicitly including experimental effects in forward- and back-projection, these effects are not folded back into the object model.