QUANTIFYING UNCERTAINTY IN A MEASUREMENT-BASED ASSESSMENT OF RELATIVE BIOLOGICAL EFFECTIVENESS IN CARBON ION RADIOTHERAPY
Author ORCID Identifier
Date of Graduation
Masters of Science (MS)
Stephen Kry, Ph.D.
Fada Guan, Ph.D.
Oleg Vassiliev, Ph.D.
Christine Peterson, Ph.D.
Paige Taylor, M.S.
One of the largest inconsistencies in dose delivered during carbon ion therapy is due to uncertainties in relative biological effectiveness (RBE), a value that is calculated via one of several clinically implemented algorithms. This study investigates the uncertainty in measured microdosimetric parameters for RBE calculation by the Microdosimetric Kinetic Model (MKM), Repair Misrepair Fixation model (RMF), and Local Effect Model I (LEM) using a Tissue Equivalent Proportional Counter (TEPC).
Microdosimetric spectra, kinetic energy spectra, and dose fragment contributions were calculated using Monte Carlo (GEANT IV) for monoenergetic and SOBP carbon beams of clinical energy. From microdosimetric spectra, lineal energy values were calculated as functions of beam energy and depth and used to calculate RBE based on MKM and RMF. From kinetic energy spectra and dose fragment contributions, RBE was calculated with LEM. To allow the assessment of RBE by RMF and LEM using microdosimetry, a method of estimating RBE from microdosimetric input values was then created with less than 5% error, on average, across all clinical energies and SOBPs.
The impact on the RBE from eight unique random or systematic sources of uncertainty associated with TEPC measurements were then simulated including electronic uncertainty, gas pressure, W-value, energy calibration, low energy cut-off, counting statistics, wall effects and pulse pile-up. These sources were quantified by statistically introducing uncertainty into the simulated measurements 200 times and sampling the resultant RBE associated with each of the 200 perturbations. The uncertainty introduced by the sources of physical noise varied depending on the model used, measurement depth, and beam energy.
The largest source of uncertainty was associated with the W-value (i.e., detector calibration), which had an uncertainty of typically 2% (1σ). Overall, the total 1σ uncertainty in the MKM based on uncertainty in TEPC measurements ranged from 2-4%. Uncertainty ranging from 2-12% was seen for RMF and LEM, incorporating error due to microdosimetric estimation. While the true RBE has extensive uncertainty associated with it, the modeled RBE can be measured with good accuracy, within a 5% deviation for MKM, which meets the reasonable tolerance goal for assessing delivered dose. For RMF and LEM, this threshold is exceeded in several individual cases (i.e. certain depths within specific beams), but is met on average. The number of cases in which this threshold is met can be increased by applying a common correction factor for both measurement and estimation based bias.
Carbon RBE, carbon radiotherapy, RBE measurement, TEPC uncertainty, microdosimetric uncertainty
Biological and Chemical Physics Commons, Medicine and Health Sciences Commons, Other Physics Commons