Optimizing voxel-wise dose calculations in cone-beam computed tomography

An analytical algorithm for the estimation of the patient-specific dose distributions in cone-beam computed tomography is introduced. The developed dose estimation method requires the reconstructed voxel data in values of linear attenuation coefficients and the scanning protocol. The algorithm first calculates the dose distribution due to the primary beam attenuation along the beam path between the source and each reconstructed voxel in conjunction with the solid angle subtended by given voxel. Then, this primary dose voxel map becomes the source for the dose distribution due to the scattered photons. For the pre-calculated primary dose value in a given voxel, the scatter dose values to all the other voxels are similarly calculated as the primary. The developed algorithm shows a good agreement with the Monte Carlo (MC) simulation for an anthropomorphic head-and-neck phantom. The accuracy of the analytical method is investigated by comparing estimates with the MC estimates and the strategy for computational acceleration is discussed in terms of the number of projections used for reconstruction, the number of spectral bins of the incident x-ray spectrum, the number of voxels, and the extent of scattering ranges.