3D Convolutional Neural Network-Based Denoising of Low-Count Whole-Body <sup>18</sup>F-Fluorodeoxyglucose and <sup>89</sup>Zr-Rituximab PET Scans
Acquisition time and injected activity of <sup>18</sup>F-fluorodeoxyglucose (<sup>18</sup>F-FDG) PET should ideally be reduced. However, this decreases the signal-to-noise ratio (SNR), which impairs the diagnostic value of these PET scans. In addition, <sup>89</sup&g...
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-02-01
|
| Series: | Diagnostics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-4418/12/3/596 |
| _version_ | 1797472082128273408 |
|---|---|
| author | Bart M. de Vries Sandeep S. V. Golla Gerben J. C. Zwezerijnen Otto S. Hoekstra Yvonne W. S. Jauw Marc C. Huisman Guus A. M. S. van Dongen Willemien C. Menke-van der Houven van Oordt Josée J. M. Zijlstra-Baalbergen Liesbet Mesotten Ronald Boellaard Maqsood Yaqub |
| author_facet | Bart M. de Vries Sandeep S. V. Golla Gerben J. C. Zwezerijnen Otto S. Hoekstra Yvonne W. S. Jauw Marc C. Huisman Guus A. M. S. van Dongen Willemien C. Menke-van der Houven van Oordt Josée J. M. Zijlstra-Baalbergen Liesbet Mesotten Ronald Boellaard Maqsood Yaqub |
| author_sort | Bart M. de Vries |
| collection | DOAJ |
| description | Acquisition time and injected activity of <sup>18</sup>F-fluorodeoxyglucose (<sup>18</sup>F-FDG) PET should ideally be reduced. However, this decreases the signal-to-noise ratio (SNR), which impairs the diagnostic value of these PET scans. In addition, <sup>89</sup>Zr-antibody PET is known to have a low SNR. To improve the diagnostic value of these scans, a Convolutional Neural Network (CNN) denoising method is proposed. The aim of this study was therefore to develop CNNs to increase SNR for low-count <sup>18</sup>F-FDG and <sup>89</sup>Zr-antibody PET. Super-low-count, low-count and full-count <sup>18</sup>F-FDG PET scans from 60 primary lung cancer patients and full-count <sup>89</sup>Zr-rituximab PET scans from five patients with non-Hodgkin lymphoma were acquired. CNNs were built to capture the features and to denoise the PET scans. Additionally, Gaussian smoothing (GS) and Bilateral filtering (BF) were evaluated. The performance of the denoising approaches was assessed based on the tumour recovery coefficient (TRC), coefficient of variance (COV; level of noise), and a qualitative assessment by two nuclear medicine physicians. The CNNs had a higher TRC and comparable or lower COV to GS and BF and was also the preferred method of the two observers for both <sup>18</sup>F-FDG and <sup>89</sup>Zr-rituximab PET. The CNNs improved the SNR of low-count <sup>18</sup>F-FDG and <sup>89</sup>Zr-rituximab PET, with almost similar or better clinical performance than the full-count PET, respectively. Additionally, the CNNs showed better performance than GS and BF. |
| first_indexed | 2024-03-09T19:57:12Z |
| format | Article |
| id | doaj.art-a126295b2b554a6f80b62a95d68b22e0 |
| institution | Directory Open Access Journal |
| issn | 2075-4418 |
| language | English |
| last_indexed | 2024-03-09T19:57:12Z |
| publishDate | 2022-02-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Diagnostics |
| spelling | doaj.art-a126295b2b554a6f80b62a95d68b22e02023-11-24T00:54:35ZengMDPI AGDiagnostics2075-44182022-02-0112359610.3390/diagnostics120305963D Convolutional Neural Network-Based Denoising of Low-Count Whole-Body <sup>18</sup>F-Fluorodeoxyglucose and <sup>89</sup>Zr-Rituximab PET ScansBart M. de Vries0Sandeep S. V. Golla1Gerben J. C. Zwezerijnen2Otto S. Hoekstra3Yvonne W. S. Jauw4Marc C. Huisman5Guus A. M. S. van Dongen6Willemien C. Menke-van der Houven van Oordt7Josée J. M. Zijlstra-Baalbergen8Liesbet Mesotten9Ronald Boellaard10Maqsood Yaqub11Cancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsCancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsCancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsCancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsCancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsCancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsCancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsCancer Center Amsterdam, Department of Medical Oncology, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsCancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsFaculty of Medicine and Life Sciences, Hasselt University, Agoralaan Building D, B-3590 Diepenbeek, BelgiumCancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsCancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The NetherlandsAcquisition time and injected activity of <sup>18</sup>F-fluorodeoxyglucose (<sup>18</sup>F-FDG) PET should ideally be reduced. However, this decreases the signal-to-noise ratio (SNR), which impairs the diagnostic value of these PET scans. In addition, <sup>89</sup>Zr-antibody PET is known to have a low SNR. To improve the diagnostic value of these scans, a Convolutional Neural Network (CNN) denoising method is proposed. The aim of this study was therefore to develop CNNs to increase SNR for low-count <sup>18</sup>F-FDG and <sup>89</sup>Zr-antibody PET. Super-low-count, low-count and full-count <sup>18</sup>F-FDG PET scans from 60 primary lung cancer patients and full-count <sup>89</sup>Zr-rituximab PET scans from five patients with non-Hodgkin lymphoma were acquired. CNNs were built to capture the features and to denoise the PET scans. Additionally, Gaussian smoothing (GS) and Bilateral filtering (BF) were evaluated. The performance of the denoising approaches was assessed based on the tumour recovery coefficient (TRC), coefficient of variance (COV; level of noise), and a qualitative assessment by two nuclear medicine physicians. The CNNs had a higher TRC and comparable or lower COV to GS and BF and was also the preferred method of the two observers for both <sup>18</sup>F-FDG and <sup>89</sup>Zr-rituximab PET. The CNNs improved the SNR of low-count <sup>18</sup>F-FDG and <sup>89</sup>Zr-rituximab PET, with almost similar or better clinical performance than the full-count PET, respectively. Additionally, the CNNs showed better performance than GS and BF.https://www.mdpi.com/2075-4418/12/3/596low-countCNNdenoising<sup>18</sup>F-FDG<sup>89</sup>Zr-antibody |
| spellingShingle | Bart M. de Vries Sandeep S. V. Golla Gerben J. C. Zwezerijnen Otto S. Hoekstra Yvonne W. S. Jauw Marc C. Huisman Guus A. M. S. van Dongen Willemien C. Menke-van der Houven van Oordt Josée J. M. Zijlstra-Baalbergen Liesbet Mesotten Ronald Boellaard Maqsood Yaqub 3D Convolutional Neural Network-Based Denoising of Low-Count Whole-Body <sup>18</sup>F-Fluorodeoxyglucose and <sup>89</sup>Zr-Rituximab PET Scans Diagnostics low-count CNN denoising <sup>18</sup>F-FDG <sup>89</sup>Zr-antibody |
| title | 3D Convolutional Neural Network-Based Denoising of Low-Count Whole-Body <sup>18</sup>F-Fluorodeoxyglucose and <sup>89</sup>Zr-Rituximab PET Scans |
| title_full | 3D Convolutional Neural Network-Based Denoising of Low-Count Whole-Body <sup>18</sup>F-Fluorodeoxyglucose and <sup>89</sup>Zr-Rituximab PET Scans |
| title_fullStr | 3D Convolutional Neural Network-Based Denoising of Low-Count Whole-Body <sup>18</sup>F-Fluorodeoxyglucose and <sup>89</sup>Zr-Rituximab PET Scans |
| title_full_unstemmed | 3D Convolutional Neural Network-Based Denoising of Low-Count Whole-Body <sup>18</sup>F-Fluorodeoxyglucose and <sup>89</sup>Zr-Rituximab PET Scans |
| title_short | 3D Convolutional Neural Network-Based Denoising of Low-Count Whole-Body <sup>18</sup>F-Fluorodeoxyglucose and <sup>89</sup>Zr-Rituximab PET Scans |
| title_sort | 3d convolutional neural network based denoising of low count whole body sup 18 sup f fluorodeoxyglucose and sup 89 sup zr rituximab pet scans |
| topic | low-count CNN denoising <sup>18</sup>F-FDG <sup>89</sup>Zr-antibody |
| url | https://www.mdpi.com/2075-4418/12/3/596 |
| work_keys_str_mv | AT bartmdevries 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT sandeepsvgolla 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT gerbenjczwezerijnen 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT ottoshoekstra 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT yvonnewsjauw 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT marcchuisman 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT guusamsvandongen 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT willemiencmenkevanderhouvenvanoordt 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT joseejmzijlstrabaalbergen 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT liesbetmesotten 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT ronaldboellaard 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans AT maqsoodyaqub 3dconvolutionalneuralnetworkbaseddenoisingoflowcountwholebodysup18supffluorodeoxyglucoseandsup89supzrrituximabpetscans |