Improvement of phoswich detector-based β+/γ-ray discrimination algorithm with deep learning.
Autoencoder
deep learning
phoswich detector
positron detection
pulse shape discrimination technique
Journal
Medical physics
ISSN: 2473-4209
Titre abrégé: Med Phys
Pays: United States
ID NLM: 0425746
Informations de publication
Date de publication:
Oct 2023
Oct 2023
Historique:
revised:
27
06
2023
received:
28
09
2022
accepted:
29
06
2023
medline:
23
10
2023
pubmed:
20
7
2023
entrez:
20
7
2023
Statut:
ppublish
Résumé
Positron probes can accurately localize malignant tumors by directly detecting positrons emitted from positron-emitting radiopharmaceuticals that accumulate in malignant tumors. In the conventional method for direct positron detection, multilayer scintillator detection and pulse shape discrimination techniques are used. However, some γ-rays cannot be distinguished by conventional methods. Accordingly, these γ-rays are misidentified as positrons, which may increase the error rate of positron detection. To analyze the energy distribution in each scintillator of the multilayer scintillator detector to distinguish true positrons and γ-rays and to improve the positron detection algorithm by discriminating true and false positrons. We used Autoencoder, an unsupervised deep learning architecture, to obtain the energy distribution data in each scintillator of the multilayer scintillator detector. The Autoencoder was trained to separate the combined signals generated from the multilayer scintillator detector into two signals of each scintillator. An energy window was then applied to the energy distribution obtained using the trained Autoencoder to distinguish true positrons from false positrons. Finally, the performance of the proposed method and conventional positron detection algorithm was evaluated in terms of the sensitivity and error rate for positron detection. The energy distribution map obtained using the trained Autoencoder was proven to be similar to that of the simulated results. Furthermore, the proposed method demonstrated a 29.79% (+0.42%p) increase in positron detection sensitivity compared to the conventional method, both having an equal error rate of 0.48%. However, when both methods were set to have the same sensitivity of 1.83%, the proposed method had an error rate that was 25.0% (-0.16%p) lower than that of the conventional method. We proposed and developed an Autoencoder-based positron detection algorithm that can discriminate between true and false positrons with a smaller error rate than conventional methods. We verified that the proposed method could increase the positron detection sensitivity while maintaining a low error rate compared to the conventional method. If the proposed algorithm is implemented in handheld positron detection probes or cameras, diseases such as cancers can be more accurately localized in a shorter time compared with using traditional methods.
Sections du résumé
BACKGROUND
BACKGROUND
Positron probes can accurately localize malignant tumors by directly detecting positrons emitted from positron-emitting radiopharmaceuticals that accumulate in malignant tumors. In the conventional method for direct positron detection, multilayer scintillator detection and pulse shape discrimination techniques are used. However, some γ-rays cannot be distinguished by conventional methods. Accordingly, these γ-rays are misidentified as positrons, which may increase the error rate of positron detection.
PURPOSE
OBJECTIVE
To analyze the energy distribution in each scintillator of the multilayer scintillator detector to distinguish true positrons and γ-rays and to improve the positron detection algorithm by discriminating true and false positrons.
METHODS
METHODS
We used Autoencoder, an unsupervised deep learning architecture, to obtain the energy distribution data in each scintillator of the multilayer scintillator detector. The Autoencoder was trained to separate the combined signals generated from the multilayer scintillator detector into two signals of each scintillator. An energy window was then applied to the energy distribution obtained using the trained Autoencoder to distinguish true positrons from false positrons. Finally, the performance of the proposed method and conventional positron detection algorithm was evaluated in terms of the sensitivity and error rate for positron detection.
RESULTS
RESULTS
The energy distribution map obtained using the trained Autoencoder was proven to be similar to that of the simulated results. Furthermore, the proposed method demonstrated a 29.79% (+0.42%p) increase in positron detection sensitivity compared to the conventional method, both having an equal error rate of 0.48%. However, when both methods were set to have the same sensitivity of 1.83%, the proposed method had an error rate that was 25.0% (-0.16%p) lower than that of the conventional method.
CONCLUSIONS
CONCLUSIONS
We proposed and developed an Autoencoder-based positron detection algorithm that can discriminate between true and false positrons with a smaller error rate than conventional methods. We verified that the proposed method could increase the positron detection sensitivity while maintaining a low error rate compared to the conventional method. If the proposed algorithm is implemented in handheld positron detection probes or cameras, diseases such as cancers can be more accurately localized in a shorter time compared with using traditional methods.
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
6118-6129Subventions
Organisme : National Research Foundation of Korea (NRF)
ID : RS-2023-00234651
Organisme : National Research Foundation of Korea (NRF)
ID : RS-2022-00165164
Organisme : National Research Foundation of Korea (NRF)
ID : NRF-2023R1A2C2007545
Organisme : National Research Foundation of Korea (NRF)
ID : NRF-2022R1I1A1A01065484
Informations de copyright
© 2023 American Association of Physicists in Medicine.
Références
Piert M, Burian M, Meisetschläger G, et al. Positron detection for the intraoperative localisation of cancer deposits. Eur J Nucl Med Mol Imaging. 2007;34(10):1534-1544. doi:10.1007/s00259-007-0430-5
Collamati F, Bocci V, Castellucci P, et al. Radioguided surgery with β radiation: a novel application with Ga68. Sci Rep. 2018;8(1):1-9.
Yamamoto S, Matsumoto K, Sakamoto S, Tarutani K, Minato K, Senda M. An intra-operative positron probe with background rejection capability for FDG-guided surgery. Ann Nucl Med. 2005;19(1):23-28.
Strong VE, Humm J, Russo P, et al. A novel method to localize antibody-targeted cancer deposits intraoperatively using handheld PET beta and gamma probes. Surg Endosc. 2008;22(2):386-391.
Van Oosterom MN, Rietbergen DDD, Welling MM, Van Der Poel HG, Maurer T, Van Leeuwen FWB. Recent advances in nuclear and hybrid detection modalities for image-guided surgery. Expert Rev Med Devices. 2019;16(8):711-734.
Kitagawa Y, Kitano S, Kubota T, et al. Minimally invasive surgery for gastric cancer-toward a confluence of two major streams: a review. Gastric Cancer. 2005;8(2):103-110.
Knoll GF. Radiation Detection and Measurement. John Wiley & Sons; 2010.
Raylmann RR. A solid-state intraoperative beta probe system. IEEE Trans Nucl Sci. 2000;47(4):1696-1703.
Raylman RR. Performance of a dual, solid-state intraoperative probe system with 18F, 99mTc, and 111In. J Nucl Med. 2001;42(2):352-360.
Levin CS, Tornai MP, MacDonald LR, Hoffman EJ. Annihilation gamma ray background characterization and rejection for a positron camera. In: 1996 IEEE Nuclear Science Symposium. Conference Record. Vol 2. IEEE; 1996:1044-1048.
Polack JK, Flaska M, Enqvist A, Sosa CS, Lawrence CC, Pozzi SA. An algorithm for charge-integration, pulse-shape discrimination and estimation of neutron/photon misclassification in organic scintillators. Nucl Instruments Methods Phys Res Sect A Accel Spectrometers, Detect Assoc Equip. 2015;795:253-267.
Farsoni AT, Alemayehu B, Alhawsawi A, Becker EM. A phoswich detector with compton suppression capability for radioxenon measurements. IEEE Trans Nucl Sci. 2012;60(1):456-464.
Tornai MP, Levin CS, MacDonald LR, Holdsworth CH, Hoffman EJ. A miniature phoswich detector for gamma-ray localization and beta imaging. IEEE Trans Nucl Sci. 1998;45(3):1166-1173.
Verdier MA, Spadola S, Pinot L, et al. Gamma-background rejection method for a dual scintillator positron probe dedicated to radio-guided surgery. Nucl Instruments Methods Phys Res Sect A Accel Spectrometers, Detect Assoc Equip. 2018;912:315-319.
Yamamoto S, Higashi T, Matsumoto K, Senda M. Development of a positron-imaging detector with background rejection capability. Ann Nucl Med. 2006;20:655-662.
Sabet H, Stack BC, Nagarkar VV. A hand-held, intra-operative positron imaging probe for surgical applications. IEEE Trans Nucl Sci. 2015;62(5):1927-1934.
Chandrikamohan P, DeVol TA. Comparison of pulse shape discrimination methods for phoswich and CsI: Tl detectors. IEEE Trans Nucl Sci. 2007;54(2):398-403.
Pozzi SA, Bourne MM, Clarke SD. Pulse shape discrimination in the plastic scintillator EJ-299-33. Nucl Instruments Methods Phys Res Sect A Accel Spectrometers, Detect Assoc Equip. 2013;723:19-23.
Kim HS, Smith MB, Koslowsky MR, Kwak SW, Ye SJ, Kim G. Characterization of a CLYC detector and validation of the Monte Carlo simulation by measurement experiments. J Radiat Prot Res. 2017;42(1):48-55.
Kim HS, Kim G, Ye SJ. Dual-particle imaging performance of a Cs2LiYCl6: Ce (CLYC)-based Rotational Modulation Collimator (RMC) system. IEEE Trans Nucl Sci. 2022;69(6):1389-1396.
Boo J, Hammig MD, Jeong M. Compact lightweight imager of both gamma rays and neutrons based on a pixelated stilbene scintillator coupled to a silicon photomultiplier array. Sci Rep. 2021;11(1):1-14.
Kim C, Yeom JY, Kim G. Digital n-γ pulse shape discrimination in organic scintillators with a high-speed digitizer. J Radiat Prot Res. 2019;44(2):53-63. 10.14407/jrpr.2019.44.2.53
Ullah MN, Park C, Pratiwi E, Kim C, Choi H, Yeom JY. A new positron-gamma discriminating phoswich detector based on wavelength discrimination (WLD). Nucl Instruments Methods Phys Res Sect A Accel Spectrometers, Detect Assoc Equip. 2019;946:162631. doi:10.1016/j.nima.2019.162631
Kim C, Lee W, Melis A, et al. A review of inorganic scintillation crystals for extreme environments. Cryst. 2021;11(6):669.
Kim C, Kim J, Hong W, Yeom JY, Kim G. Development of an energy and efficiency calibration method for stilbene scintillators. Nucl Eng Technol. 2022;54(10):3833-3840.
Hurtado S, Garcıa-León M, Garcıa-Tenorio R. GEANT4 code for simulation of a germanium gamma-ray detector and its application to efficiency calibration. Nucl Instruments Methods Phys Res Sect A Accel Spectrometers, Detect Assoc Equip. 2004;518(3):764-774.
Hurtado S, Garcia-Tenorio R, Garcia-Leon M. Coincidence summing corrections in gamma-ray spectrometry using GEANT4 code. IEEE Trans Nucl Sci. 2009;56(3):1531-1536.
Hauf S, Kuster M, Batič M, et al. Radioactive decays in Geant4. IEEE Trans Nucl Sci. 2013;60(4):2966-2983.
Hinton GE, Salakhutdinov RR. Reducing the dimensionality of data with neural networks. Science. 2006;313(5786):504-507.
Rumelhart DE, Hinton GE, Williams RJ. Learning Internal Representations by Error Propagation. California Univ San Diego La Jolla Inst for Cognitive Science; 1985.
Lu X, Tsao Y, Matsuda S, Hori C. Speech enhancement based on deep denoising autoencoder. In: Interspeech. ISCA; 2013:436-440.
Gondara L. Medical image denoising using convolutional denoising autoencoders. In: 2016 IEEE 16th International Conference on Data Mining Workshops (ICDMW). IEEE; 2016:241-246.
Ryu S, Kim M, Kim H. Denoising autoencoder-based missing value imputation for smart meters. IEEE Access. 2020;8:40656-40666.
Nair V, Hinton GE. Rectified linear units improve restricted boltzmann machines. In: Proceedings of the 27th International Conference on International Conference on Machine Learning. Omnipress; 2010:807-814.
Hall NC, Povoski SP, Murrey DA, Knopp MV, Martin EW. Combined approach of perioperative 18 F-FDG PET/CT imaging and intraoperative 18 F-FDG handheld gamma probe detection for tumor localization and verification of complete tumor resection in breast cancer. World J Surg Oncol. 2007;5(1):1-8.
Ullah MN, Park JH, Pratiwi E, Kim GB, Yeom JY. Wavelength discrimination (WLD) detector optimization for time-of-flight positron emission tomography with depth of interaction information. Nucl Instruments Methods Phys Res Sect A Accel Spectrometers, Detect Assoc Equip. 2020;982:164498. doi:10.1016/J.NIMA.2020.164498
Ullah MN, Pratiwi E, Park JH, Lee K, Choi H, Yeom JY. Wavelength discrimination (WLD) TOF-PET detector with DOI information. Phys Med Biol. 2020;65(5):55003.
Govorkova E, Puljak E, Aarrestad T, et al. Autoencoders on field-programmable gate arrays for real-time, unsupervised new physics detection at 40 MHz at the Large Hadron Collider. Nat Mach Intell. 2022;4(2):154-161.