(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose:
Ionization chambers are frequently used for profile measurements due to their desired low energy dependence. However, the large size of the chamber’s sensitive volume leads to the volume-averaging effect, and the low-density air cavity causes further perturbation to secondary electron fluence. These combined effects, named as the volume effect, lead to the broadening of the penumbra regions and, at small fields, also cause underestimation of the measured output. Therefore, the volume effect associated with ionization chamber-measured dose profiles needs to be corrected.
Methods:
The neural network (NN) method introduced by Liu et al. (2018) was applied to deconvolve beam profiles of small fields between 3 x 3 mm² and 20 x 20 mm². Three ionization chambers with different sensitive volumes (PTW 31021, PTW 31022, SNC 125c) were used to collect three sets of beam profiles. Another set of beam profiles was collected with a microDiamond detector and used to train separate NNs for each ionization chamber. To test the NNs, a different set of profiles was measured and used as input to the NN. The deconvolved results were compared to the microDiamond measurements.
Results:
The comparison between the deconvolved profiles and the microdiamond-measured reference profiles was performed using gamma analysis (1 mm / 1%) for all data points above 3% of the dose maximum. For the smallest field size, the passing rate for all three ionization chambers is above 99%.
Conclusion:
The comparison of the signal profiles obtained with different detectors show the need to deconvolve the profiles measured in small fields using ionization chambers. The applicability of a NN for deconvolution in small fields has been demonstrated. The results showed that the method works equally well for various compact ionization chambers.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email