On the use of machine learning methods for mPSD calibration in HDR brachytherapy

Purpose: We sought to evaluate the feasibility of using machine learning algorithms for multipoint plastic scintillator detector calibration in high-dose-rate brachytherapy. Methods: The dosimetry system consisted of an optimized 1-mm-core mPSD and a compact assembly of photomultiplier tubes coupled with dichroic mirrors and filters. An $^{192}$Ir source was remotely controlled and sent to various positions in a homemade PMMA holder. Dose measurements covering a range of 0.5 to 12 cm of source displacement were carried out according to TG-43 recommendations. Individual scintillator doses were decoupled using a linear regression model, a random forest estimator, and artificial neural network algorithms. The performance of the different algorithms was evaluated using different sample sizes and distances to the source for the mPSD system calibration. Results: The decoupling methods deviations from the expected TG-43 dose generally remained below 20%. However, the dose prediction with the three algorithms was accurate to within 7% relative to the dose predicted by the TG-43 formalism for measurements performed in the same range of distances used for calibration. The performance random forest was compromised when the predictions were done beyond the range of distances used for calibration. The dose prediction by the linear regression was less influenced by the calibration conditions than random forest, but with more significant deviations. The number of available measurements for training purposes influenced the random forest and neural network models the most. Their accuracy tended to converge toward deviation values close to 1% from a number of dwell positions greater than 100. Conclusions: In performing HDR brachytherapy dose measurements with an optimized mPSD system, ML algorithms are good alternatives for precise dose reporting and treatment assessment.