Article Sidebar

PDF
Published: 2021-06-30
Abstract Views: 165
Views PDF: 61
DOI: https://doi.org/10.53747/nst.v11i2.356
Issue
Vol. 11 No. 2 (2021)
Section
Articles
How to Cite
Phan , V. C., Ho , T. T., Le , T. A., Nguyen , D. T., Nguyen , D. K., & Nguyen , H. H. (2021). A convolutional neural network for Y90 SPECT/CT scatter estimation. Nuclear Science and Technology, 11(2), 9-13. https://doi.org/10.53747/nst.v11i2.356

A convolutional neural network for Y90 SPECT/CT scatter estimation

Phan Viet Cuong1, Ho Thi Thao2, Le Tuan Anh3, Nguyen Duc Ton3, Nguyen Dinh Khai4, Nguyen Hong Ha5
1 Research and Development Center for Radiation Technology, Vietnam Atomic Energy Institute
2 School of Mechanical Engineering, Kyungpook National University, South Korea
3 Institute for Nuclear Science and Technology, Vietnam Atomic Energy Institute
4 Military Institute of Medical Radiology and Oncology, Vietnam
5 M1 General Physics, Paris-Saclay University, 91405 Orsay Cedex, France

Main Article Content

Abstract

Monte Carlo-based scatter modeling in SPECT has demonstrated the ability on improving image quality and quantitative accuracy but high computational cost. In this study, we describe a deep learning method based on a convolutional neural network (CNN) to increase the image quality, decrease the computation time for SPECT/CT reconstruction. Monte Carlo (MC) simulation and true scatter data are used for training and validation phase and the CNN network is trained to match the MC scatter estimation. In the testing step with a liver subject, visual image quality by CNN was better than MC scatter estimation method. Besides, the CNN scatter estimate was generated over a much shorter period of time than MC model (about 15 seconds for CNN vs ~2 hours for MC). The short processing time with CNN while maintaining quality has high clinical significance for quantitative SPECT imaging.

Article Details

References

[1]. B. F. Hutton, et al. Review and current status of SPECT scatter correction, Phys. Med. Biol. 56, R85-R112, 2011.
[2]. Dewaraja, Y.K., et al. Improved quantitative (90) Y bremsstrahlung SPECT/CT reconstruction with Monte Carlo scatter modeling. Med Phys 44, 6364-6376, 2017.
[3]. Elschot, M., Lam, M.G., van den Bosch, M.A., Viergever, M.A. & de Jong, H.W. Quantitative Monte Carlo-based 90Y SPECT reconstruction. J Nucl Med 54, 1557-1563, 2013.
[4]. Smits, M.L., et al. Radioembolization dosimetry: the road ahead. Cardiovasc Intervent Radiol 38, 261-269, 2015.
[5]. Xiang, H., Lim, H., Fessler, J.A. & Dewaraja, Y.K. SPECT/CT scatter estimation using a deep convolutional neural network: implementation in Y-90 imaging. in 2019 IEEE Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC) 1-3, 2019.
[6]. M. Ljungberg, S.S.a.M.K. The SIMIND Monte Carlo program, (Bristol:IOP Publishing, 1998.
[7]. Lim, H., Dewaraja, Y. Y-90 patients PET/CT & SPECT/CT and corresponding contours dataset. (University of Michigan - Deep Blue, 2019.
[8]. Xiang, H., Lim, H., Fessler, J.A. & Dewaraja, Y.K. A deep neural network for fast and accurate scatter estimation in quantitative SPECT/CT under challenging scatter conditions. Eur J Nucl Med Mol Imaging 47, 2956-2967, 2020.