Purpose: To accurately and efficiently estimate individualized patient dose in CT scans with Monte Carlo (MC) simulations.
Methods: A GPU-based MC tool with the ability to model a non-isotropic source was developed from MC-GPU v1.3. Unlike the isotropic source in the original MC-GPU, the anode heel effect and bowtie filtration are realized in this tool. The heel effect leads to photon intensity variation for azimuthal angles that can be modeled as a probability function, while the bowtie model is based on analytic attenuation. The bowtie is represented by two basis materials so that with inputs of material attenuation coefficients and thickness combinations at different fan angles, photon transmission and filtered spectra for each fan angle can be calculated for a given source spectrum. Using a precomputed lookup table of the bowtie, the photon’s direction and energy can be efficiently randomly selected using the alias method. An air scan was used to validate the simulation tool by comparing against the analytically calculated ground truth image. For individualized patient dose maps, we generated spatial maps of both material type and mass density based on the CT numbers as the inputs for the MC tool along with necessary scanner-specific parameters.
Results: The simulation was performed on a workstation with two Nvidia Titan RTX GPUs. The average processing speed was ~3.5e7 photons/sec (original MC-GPU with isotropic source: 4.7e7 photons/sec). In the air scan validation, the average relative error in a spectral test was 0.86% for 25-80 keV and the average relative errors in energy-integrated detection were 1.69% and 0.92% for bowtie filtration and heel effect, respectively.
Conclusion: In this work, a near real-time MC dose tool is introduced which has the potential to provide realistic and individualized dose map from non-isotropic sources. Ongoing work will include tube current modulation and automatic organ segmentation.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by GE Healthcare. Sen Wang received Zijing Scholarship from Tsinghua University.
Not Applicable / None Entered.