Experimental analysis of the efficiency of gamma-ray spectrum smoothing with LS-SVM when using a compact scintillation detector.

  • Egor V. Mischenko Institute of Radiobiology of the National Academy of Sciences of Belarus
  • Aleksander N. Nikitin Institute of Radiobiology of the National Academy of Sciences of Belarus
  • Evgenia V. Solonenko Institute of Radiobiology of the National Academy of Sciences of Belarus

Abstract

Improving methods for processing gamma-ray spectrograms is a highly promising direction for further advancements in determining the content of radionuclides in environmental objects. A significant challenge lies in smoothing spectra during in situ measurements due to time constraints and low signal-to-noise ratio. This study evaluates the effectiveness of using Least Squares Support Vector Machine (LS-SVM) regression for smoothing spectra obtained with a NaI(Tl) scintillation detector, comparing it with moving average and exponential smoothing methods. The evaluation of spectrum alignment effectiveness was conducted for the entire energy measurement range. Semi-synthetic gamma-ray spectra with measurement times 60, 300, 900, 1800, 3600, 7200, and 72000 seconds, generated by channel summing of randomly selected real spectra converted into count rates, were employed for analysis. Real spectra were acquired using an experimental setup consisting of an ATOMTEX detection unit, a radionuclide source, a computer with specialized software, and auxiliary devices. The analysis demonstrated that the LS-SVM method is most effective for smoothing spectrograms obtained at measurement times ranging from 1·103 to 7·103 seconds. Separate optimization of smoothing model hyperparameters is required for three partially overlapping energy subranges. For more longer measurements, applying tested spectrogram smoothing methods does not reduce the signal-to-noise ratio. For short intervals (up to 100 seconds), the moving average method exhibits the highest efficiency, allowing for a 2–2.5 dB improvement in the signal-to-noise ratio. In the measurement time range of 102–103 seconds, the most effective suppression of statistical noise is achieved with exponential smoothing with α = 0.75.

Author Biographies

Egor V. Mischenko, Institute of Radiobiology of the National Academy of Sciences of Belarus

researcher at the laboratory of radioecology.

Aleksander N. Nikitin, Institute of Radiobiology of the National Academy of Sciences of Belarus

PhD (agriculture); deputy director for research.

Evgenia V. Solonenko, Institute of Radiobiology of the National Academy of Sciences of Belarus

researcher at the laboratory of radio- ecology.

References

  1. Rouze G, Morgana C, McBratney A. Understanding the utility of aerial gamma radiometrics for mapping soil properties through proximal gamma surveys. Geoderma. 2017;289:185–195. DOI: 10.1016/j.geoderma.2016.12.004.
  2. Maacha L, Jaffal M, Jarni A. A contribution of airborne magnetic, gamma ray spectrometric data in understanding the structure of the Central Jebilet Hercynian massif and implications for mining. Journal of African Earth Sciences. 2017;134:389–403. DOI: 10.1016/j. jafrearsci.2017.07.012.
  3. Söderström M, Eriksson J. Gamma-ray spectrometry and geological maps as tools for cadmium risk assessment in arable soils. Geoderma. 2013;192:323–334. DOI: 10.1016/j.geoderma.2012.07.014.
  4. Khan H, Chaudhry Z, Ismail MK. Assessment of Radionuclides, Trace Metals and Radionuclide Transfer from Soil to Food of Jhangar Valley (Pakistan) Using Gamma-Ray Spectrometry. Water Air Soil Pollution. 2010;213:353–362. DOI: 10.1007/s11270-010- 0390-4.
  5. International atomic energy agency. Guidelines for Radioelement Mapping Using Gamma Ray Spectrometry Data, IAEA- TECDOC-1363. Vienna: IAEA; 2003. 184 p.
  6. Blackburn JA. Statistical noise and spectral analysis. Nuclear Instruments and Methods. 1968;63(1):66–70. DOI: 10.1016/0029- 554X(68)90302-9.
  7. Minty B, Hovgaard J. Spectral methods for reducing noise in gamma-ray spectrometry. ASEG Extended Abstracts. 2001;4:1–4. DOI: 10.1071/ASEG2001ab089.
  8. Varley A, Tyler A, Smith L, Dale P & Davies M. Remediating radium contaminated legacy sites: Advances made through machine learning in routine monitoring of «hot» particles. Science of The Total Environment. 2015;521:270–279. DOI: 10.1016/j. scitotenv.2015.03.131.
  9. Dragović S, Onjia A. Classification of soil samples according to their geographic origin using gamma-ray spectrometry and principal component analysis. Journal of Environmental Radioactivity. 2006;89(2):150–158. DOI: 10.1016/j.jenvrad.2006.05.002.
  10. Zhu MH, Liu LG, Qi DX, You Z, Xu AA. Smoothing noisy spectroscopic data with many-knot spline method. Nuclear Instruments and Methods in Physics Research. 2008;589:484–486. DOI: 10.1016/j.nima.2008.03.008.
  11. Tirosh S, Van De Ville D, Unser M. Polyharmonic smoothing splines and the multidimensional Wiener filtering of fractal-like signals. IEEE Trans Image Process. 2006;15 (9):2616–2630. DOI: 10.1109/tip.2006.877390.
  12. Claeskens G, Carroll RJ. Automatic estimation of multivariate spectra via smoothing splines. Biometrika. 2007;94(2):335–345. DOI: 10.1093/biomet/asm022.
  13. Fan CL, Pang MY. Reconstructing smooth curve from noise sampled data. Proceedings. The 2009 International Workshop on Information Security and Application (IWISA 2009). [Place unknown]: Academy Publisher; 2009. p. 218–222.
  14. J. B. Lee, A. S. Woodyatt and M. Berman. Enhancement of high spectral resolution remote-sensing data by a noise-adjusted principal components transform. IEEE Transactions on Geoscience and Remote Sensing. 1990;28(3):295–304. DOI: 10.1109/36.54356.
  15. Hovgaard J. A new processing technique for airborne gamma-ray spectrometer data (Noise adjusted singular value decomposition). Sixth topical meeting on Emergency Preparedness and Response. San Francisco: American Nuclear Society; 1997. p. 123–127.
  16. Jiancheng S, Chongxun Z, Yatong Z. Nonlinear Noise Reduction of Chaotic Time Series Based on Multidimensional Recurrent LS-SVM. Neurocomputing. 2008;71(16–18):3675–3679. DOI: 10.1016/j.neucom.2008.02.006.
  17. Zhang F, O’Donnell LJ. Support vector regression. Machine Learning. Academic Press. 2020;12:123–140. DOI: 10.1016/B978- 0-12-815739-8.00007-9.
  18. Liu J, Guan X, et al. Airborne γ spectrum analysis based on LS-SVM segmented noise reduction method. Nuclear Electronics and Detection Technology. 2017;37(1):24–28.
  19. Varley A, Tyler A, Dowdall M, Bondar Y, Zabrotski V. An in situ method for the high resolution mapping of 137Cs and estimation of vertical depth penetration in a highly contaminated environment. Science of the Total Environment. 2017;605–606:957–966. DOI: 10.1016/j.scitotenv.2017.06.067.
  20. Kim J, Lim KT, Kim J, Kim C, Jeon B, Park K, Kim G. Quantitative analysis of NaI(Tl) gamma-ray spectrometry using an artificial neural network. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment. 2019;944:162549. DOI: 10.1016/j.nima.2019.162549.
  21. Ying C, Jun L, Lei C. Aeronautical γ spectrum noise reduction method based on LS-SVM segmentation regression. Journal of Physics. 2019;2:022141. DOI: 10.1088/1742-6596/1237/2/022141.
Published
2024-05-21
Keywords: gamma-spectrometry, spectrum unfolding, ignal-to-noise ratio, support vector regression
Supporting Agencies This study was supported by the Belarussian state scientific research program «Natural Resources and Environment» (2021–2025) task 3.05.4.
How to Cite
Mischenko, E., Nikitin, A., & Solonenko, E. (2024). Experimental analysis of the efficiency of gamma-ray spectrum smoothing with LS-SVM when using a compact scintillation detector. Journal of the Belarusian State University. Ecology, 1, 32-45. Retrieved from https://journals.bsu.by/index.php/ecology/article/view/6406
Issue
No 1 (2024): Journal of the Belarusian state university. Ecology
Section
Radioecology and Radiobiology, Radiation Safety