(Thursday, 8/2/2018) 1:00 PM - 3:00 PM
Room: Room 202
Purpose: Monte Carlo (MC) method can accurately compute scattered photon signal for scatter removal in cone beam CT (CBCT). But it suffers from low computational efficiency due to its particle-by-particle simulation scheme, which unavoidably spends computation efforts on transporting particles that do not contribute to signals at the detector. In this work, we propose a novel MC method employed path-by-path MC scheme to perform efficient and accurate scatter simulation
Methods: As opposed to sampling a photon from x-ray source, we sample an entire photon path starting from the source, scattered by the phantom, and hitting the detector. Each path carries a probability determined by x-ray source fluence distribution, energy spectrum, and photon transport physics in the phantom. A Metropolis algorithm is used to sample these paths under its probability distribution. The method is implemented on a GPU platform. To test our method, we compute scatter signal in a CBCT projection caused by a homogeneous aluminum phantom and a head-and-neck patient case (HN). We compare simulation results and computational efficiency with the computations performed under the conventional particle-by-particle MC scheme on the same GPU.
Results: In the homogeneous aluminum case, we reach an overall speed up factors of ~48.57 times and a relative difference of 1.76%. In the HN case, the speed up factor is ~ 20.78 times with a relative difference of 2.89%.
Conclusion: We have successfully developed a novel MC method with a path-by-path sampling scheme for fast scattering calculation in CBCT. The achieved efficiency and accuracy make this method promising for MC-based scatter removal in CBCT.
Funding Support, Disclosures, and Conflict of Interest: National Natural Science Foundation of China (81301940 and 81428019), National Key Research and Development Program (2016YFA0202003), Guangdong Natural Science Foundation of China (2016A030310388 and 2017A030313692), and Southern Medical University Startup fund (LX2016N003).
Keywords
Cone-beam CT, Scatter, Monte Carlo
Taxonomy
IM- Cone Beam CT: General (Most aspects)
Contact Email
