Is Accurate 3D Dosimetry Possible When Using Optical-CT Readout with Low (or Without) Refractive Index Matching Fluids?
L Rankine1*, M Oldham1,2, (1) Medical Physics Graduate Program, Duke University, Durham, NC, (2) Department of Radiation Oncology, Duke University Medical Center, Durham, NCSU-E-T-40 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
Purpose: Achieving accurate optical-CT 3D dosimetry without the use of viscous refractive index (RI) matching fluids would greatly increase convenience. This work evaluates various potential 'dry-scan' optical-CT configurations (including parallel, point, and converging light-ray geometries) through computer simulations.
Methods: Three surrounding refractive media were investigated: air, water, and a fluid closely matched to PRESAGE™ (n=1.00,1.33,1.49). Reconstructions performed using both filtered-back-projection (FBP) and algebraic-reconstruction-technique (ART). For each media, the efficacy of the three configurations and the two algorithms was evaluated by calculating usable radius (i.e. the outermost radius where data was accurate to within 2%), and gamma analysis. This definition recognizes that, for optical-CT imaging, errors are greatest near the edge of the dosimeter where refraction can be most pronounced. Simulations were performed on three types of dose distribution: uniform, volumetric modulated arc therapy (VMAT), and brachytherapy (Cs-137).
Results: Uniformly irradiated dosimeter useable radius recovered using FBP was 68% for water-matching, and 31% for air-matching (dry-scanning). ART gave useable radii of 99% for both water and dry-scanning, indicating greater recovery of useful data for the uniform distribution. FBP and ART performed equally well for a VMAT dose distribution where less dose is delivered near the edge of the dosimeter (useable radius of 86% and 53% for water and dry-scanning respectively). Brachytherapy useable radius recovered using FBP was 99% and 98% for water and dry-scanning respectively, and a major decrease was seen with ART. Point-source geometry provided 1-2% larger usable radii than parallel. Converging geometry recovered up to 10% reduced useable radii than point and parallel.
Conclusion: For applications where dose information is not required in the periphery of the dosimeter, some dry and low-viscous matching configurations may be feasible. When dosimetry is required on the periphery, best results were obtained using close refractive matching and ART.