Program Information
A Novel Maximum Likelihood Method for High-Resolution Proton Radiography/proton CT
C Collins-Fekete1,3,4*, R Schulte2 , L Beaulieu1,3 , J Seco4,5 , (1) Universite Laval, Quebec, Quebec, (2) Loma Linda University, Loma Linda, CA, (3) Centre Hospitalier University de Quebec, Quebec, QC, (4) Mass General Hospital; Harvard Medical, Boston , MA, (5) Department of Medical Physics in Radiooncology, DKFZ German Cancer Research Cente
Presentations
SU-C-207A-1 (Sunday, July 31, 2016) 1:00 PM - 1:55 PM Room: 207A
Purpose: Multiple Coulomb scattering is the largest contributor to blurring in proton imaging. Here we tested a maximum likelihood least squares estimator (MLLSE) to improve the spatial resolution of proton radiography (pRad) and proton computed tomography (pCT).
Methods: The object is discretized into voxels and the average relative stopping power through voxel columns defined from the source to the detector pixels is optimized such that it maximizes the likelihood of the proton energy loss. The length spent by individual protons in each column is calculated through an optimized cubic spline estimate. pRad images were first produced using Geant4 simulations. An anthropomorphic head phantom and the Catphan line-pair module for 3-D spatial resolution were studied and resulting images were analyzed. Both parallel and conical beam have been investigated for simulated pRad acquisition. Then, experimental data of a pediatric head phantom (CIRS) were acquired using a recently completed experimental pCT scanner. Specific filters were applied on proton angle and energy loss data to remove proton histories that underwent nuclear interactions. The MTF10% (lp/mm) was used to evaluate and compare spatial resolution.
Results: Numerical simulations showed improvement in the pRad spatial resolution for the parallel (2.75 to 6.71 lp/cm) and conical beam (3.08 to 5.83 lp/cm) reconstructed with the MLLSE compared to averaging detector pixel signals. For full tomographic reconstruction, the improved pRad were used as input into a simultaneous algebraic reconstruction algorithm. The Catphan pCT reconstruction based on the MLLSE-enhanced projection showed spatial resolution improvement for the parallel (2.83 to 5.86 lp/cm) and conical beam (3.03 to 5.15 lp/cm). The anthropomorphic head pCT displayed important contrast gains in high-gradient regions. Experimental results also demonstrated significant improvement in spatial resolution of the pediatric head radiography.
Conclusion: The proposed MLLSE shows promising potential to increase the spatial resolution (up to 244%) in proton imaging.
Contact Email: