Standardized Open-Source Volume Calculations for Small Voxelized Volumes
R Cardan1*, R Popple2, (1) UAB University of Alabama, Birmingham, Birmingham, AL, (2) University of Alabama Birmingham, Birmingham, AL
Presentations
PO-GePV-T-397 (Sunday, 7/10/2022) [Eastern Time (GMT-4)]
ePoster Forums
Purpose: The discrepancy of small volume calculations has been shown to be large between various systems used in radiation oncology. These discrepancies obfuscate outcome analysis and intersystem comparisons of plan quality. A technique is needed to provide a standard for volume metrics.
Methods: A spherically symmetric function was used to generate dose distributions of varying diameters projected onto a 3D voxelized grid. Using a marching cubes algorithm, a triangular mesh was created at the 75% isodose level. The mesh volume was calculated and compared to the known “golden” volume of the sphere. The relationship was plotted to determine the conversion between the calculated and true volumes.
Results: The raw volumes were found to differ by 37.8% for a 3.0 mm diameter sphere and 0.5% for the largest 100 mm sphere, with an average error of 9.72 ± 11.04%. The conversion factor from mesh volume to true volume was found to follow the power law (R2 = 0.989) by the equation V=m*(1+ 0.9574m-0.4103) where V is the true volume and m is the mesh volume. The discrepancy after applying conversion was an average difference of 0.20 ± 1.34%, with a maximum of 3.11%.
Conclusion: Using a simple correction factor, triangular meshes produced by marching cubes can produce highly accurate volume metrics for small voxelized volumes. An open-source algorithm for this technique is provided at http://github.com/rexcardan/srs-marching.git.
Funding Support, Disclosures, and Conflict of Interest: One or more authors has consulting agreements with Varian Medical Systems
Keywords
Software, Radiation Therapy, Dose
Taxonomy
Not Applicable / None Entered.
Contact Email