Comparative analysis of machine learning-based dose assessment algorithms for TL dosimetry
2
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This paper explores the implementation of machine learning-based algorithms for TL dose assessment. It focuses on the radiation field classification, performance quotient evaluation, and shallow and deep dose equivalent assessment of ANN and LGBM, in comparison to the traditional method of DT. We evaluate these algorithms based on the element response data measured by TLD. A data set was built for training, and the base element responses of test categories were amplified, and normalized to 1 mSv Cs-137 within the range of ±3 %. Both algorithms consist of five subset models for classifying radiation fields and identifying ratios of mixed fields. The LGBM showed the best accuracy in classifying considered radiation fields and the lowest performance quotients. By comparing the tolerance levels of deep dose and shallow dose equivalents among the three algorithms, the LGBM yields the smallest difference between the predicted and true dose equivalents. This smaller difference implies the LGBM offers the least bias and standard deviation in the expected value, giving higher accuracy and precision in dose assessment over the traditional DT method. The findings from this study further contribute to the adoption of ML-based algorithms for TL dose assessment, underscoring its importance in the field.
Abstract
This paper explores the implementation of machine learning-based algorithms for TL dose assessment. It focuses on the radiation field classification, performance quotient evaluation, and shallow and deep dose equivalent assessment of ANN and LGBM, in comparison to the traditional method of DT. We evaluate these algorithms based on the element response data measured by TLD. A data set was built for training, and the base element responses of test categories were amplified, and normalized to 1 mSv Cs-137 within the range of ±3 %. Both algorithms consist of five subset models for classifying radiation fields and identifying ratios of mixed fields. The LGBM showed the best accuracy in classifying considered radiation fields and the lowest performance quotients. By comparing the tolerance levels of deep dose and shallow dose equivalents among the three algorithms, the LGBM yields the smallest difference between the predicted and true dose equivalents. This smaller difference implies the LGBM offers the least bias and standard deviation in the expected value, giving higher accuracy and precision in dose assessment over the traditional DT method. The findings from this study further contribute to the adoption of ML-based algorithms for TL dose assessment, underscoring its importance in the field.
참고문헌 (36)
- E. G. Yukihara, "Luminescence dosimetry", Nature Rev. Methods Prim, 2(1), pp.26~ (2022)[DOI]
- Determination of dose equivalents resulting from external radiation sources, "Intertional Commission on Radiation Units and Measurement", ICRU, ~, pp.ICRU Rep. 39 (1985)
- Quantities and units in radiation protection dosimetry, "Intertional Commission on Radiation Units and Measurement", ICRU, ~, pp.ICRU Rep. 51 (1993)
- I. O. Polo, "Response of a TLD badge to the new operational quantity Hp(θ): Monte Carlo approach", Radiat. Phys. Chem, 191, pp.109869~ (2022)[DOI]
- P. Olko, "Advantages and disadvantages of luminescence dosimetry", Radiat. Meas, 45(3–6), pp.506~511 (2010)
- O. E. Belhaj, "Dose metrology : TLD", OSL dose accuracy and energy response performance, 2023, pp.55(2), 717~724
- Z. ˇ Kneˇzevi´c, "Photon dosimetry methods outside the target volume in radiation therapy: optically stimulated luminescence (OSL), thermoluminescence (TL) and radiophotoluminescence (RPL) dosimetry", Radiat. Meas, 57, pp.9~18 (2013)
- S. Yokoyama, "Radiation exposure to the lens of the eye for Japanese nuclear power plant workers", J. Radiol. Prot, 42(3), pp.031504~ (2022)
- S. F. Alanazi, "Response evaluation of two commercial thermoluminescence dosimeters(TLDs)against different parameters", BJR Open, 5(1), pp.20220035~ (2023)[DOI]
- L. Sousa Santiago de Oliveira, "Challenges in personal and clinical dosimetry using Li2B4O7 and MgB4O7 as TLD and OSLD", Brazil. J. Radiat. Sci, 10(2A), pp.15392~ (2022)[DOI]
- M. Moscovitch, "Dose algorithms for personal thermoluminescence dosimetry", Radiat. Protect. Dosim, 47(1–4), pp.373~380 (1993)[DOI]
- S. Y. Lee, "Feasibility of using neural Networks to unfold the response of multi-element TLD for mixed field dosimetry", Int. Radiat. Protect. Associat (2000)
- H. Y. Jung, "Development of Personal Radiation Dosimeter and Dose Assessment Algorithm Based on LiF Thermoluminescence Material and Artificial Neural Network", Ph.D, pp.2003
- S. -Y. Lee, "An application of artificial neural intelligence for personal dose assessment using a multi-area OSL dosimetry system", Radiat. Meas, 33(3), pp.293~304 (2001)[DOI]
- H. Jung, "Si TL detectors for photon fields", Radiat. Meas, 38(1), pp.71~80 (2004)[DOI]
- ANSI, "American National Standard for Dosimetry-Personnel Dosimetry Performance Creiteria for Testing, N13. 11", ANSI, ~ (1993)
- M. Oberhofer, "Applied Thermoluminescence Dosimetry", pp.173~175 (1981), Adam Hilger Ltd
- J. I. Lee, "Developments in the synthesis of LiF:Mg,Cu,Na,Si TL material", Radiat. Protect. Dosim, 108(1), pp.79~83 (2004)[DOI]
- W. -H. Lee, "Comparison on the dosimetry of TLD and OSLD used in nuclear medicine", J. Korea Contents Associat, 12(12), pp.329~334 (2012)[DOI]
- S. Kim, "LiF TLD in TLD holder for in vivo dosimetry", Radiat. Oncol. J, 19(3), pp.293~299 (2001)
- Establishment of technical standards for personal dosimetry performance test by new ANSI criteria, "I. -S. Jang", KAERI, 19~20, pp.TRKO202100006285 (2021)
- M. -J. Song, "development of algorithm for evaluation of the dose with thermoluminescence dosimeter, radiation protection", J. Kor. Associat. Radiat. Protect, 17(2), pp.1~13 (1992)
- D. -H. Cho, "A study on the dose assessment methodology using the probabilistic characteristics of TL element response", J. Radiat. Protect. Res, 23(3), pp.123~138 (1998)
- R. M. West, "Best practice in statistics : the use of log transformation", Ann. Clin. Biochem, 59(3), pp.162~165 (2022)[DOI]
- S. Mr Dongare, "Basic of artificial neural network", Int. J. Adv. Res. Sci. Commun. Technol, pp.299~303 (2023)[DOI]
- M. Islam, "An overview of neural network", Am. J. Neural Network Appl, 5(1) (2019)[DOI]
- J. H. Friedman, "Greedy function approximation: a gradient boosting machine", Ann. Stat, pp.1189~1232 (2001)
- G. Ke, "Lightgbm: a highly efficient gradient boosting decision tree, Adv", Neural Inf. Process. Syst, 30 (2017)
- (2015)[URL]
- "Lightgbm: light gradient boosting machine" (2023)[URL]
- "KerasTuner" (2019)[URL]
- "Bayesian Optimization, Open Source Constrained Global Optimization Tool for Python" (2014)[URL]
- Daniel Silva-Palacios, "Improving performance of Multiclass classification by Inducing class Hierarchies", Proc. Comput. Sci, 108, pp.1692~1701 (2017)[DOI]
- Yu Zhang, "An overview of multi-task learning", Natl. Sci. Rev, 5(1), pp.30~43 (2018)[DOI]