The Impact of Artificial Intelligence CNN Based Denoising on FDG PET Radiomics
BackgroundWith a constantly increasing number of diagnostic images performed each year, Artificial Intelligence (AI) denoising methods offer an opportunity to respond to the growing demand. However, it may affect information in the image in an unknown manner. This study quantifies the effect of AI-b...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2021-08-01
|
| Series: | Frontiers in Oncology |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fonc.2021.692973/full |
| _version_ | 1818923134669029376 |
|---|---|
| author | Cyril Jaudet Kathleen Weyts Alexis Lechervy Alain Batalla Stéphane Bardet Aurélien Corroyer-Dulmont Aurélien Corroyer-Dulmont |
| author_facet | Cyril Jaudet Kathleen Weyts Alexis Lechervy Alain Batalla Stéphane Bardet Aurélien Corroyer-Dulmont Aurélien Corroyer-Dulmont |
| author_sort | Cyril Jaudet |
| collection | DOAJ |
| description | BackgroundWith a constantly increasing number of diagnostic images performed each year, Artificial Intelligence (AI) denoising methods offer an opportunity to respond to the growing demand. However, it may affect information in the image in an unknown manner. This study quantifies the effect of AI-based denoising on FDG PET textural information in comparison to a convolution with a standard gaussian postfilter (EARL1).MethodsThe study was carried out on 113 patients who underwent a digital FDG PET/CT (VEREOS, Philips Healthcare). 101 FDG avid lesions were segmented semi-automatically by a nuclear medicine physician. VOIs in the liver and lung as reference organs were contoured. PET textural features were extracted with pyradiomics. Texture features from AI denoised and EARL1 versus original PET images were compared with a Concordance Correlation Coefficient (CCC). Features with CCC values ≥ 0.85 threshold were considered concordant. Scatter plots of variable pairs with R2 coefficients of the more relevant features were computed. A Wilcoxon signed rank test to compare the absolute values between AI denoised and original images was performed.ResultsThe ratio of concordant features was 90/104 (86.5%) in AI denoised versus 46/104 (44.2%) with EARL1 denoising. In the reference organs, the concordant ratio for AI and EARL1 denoised images was low, respectively 12/104 (11.5%) and 7/104 (6.7%) in the liver, 26/104 (25%) and 24/104 (23.1%) in the lung. SUVpeak was stable after the application of both algorithms in comparison to SUVmax. Scatter plots of variable pairs showed that AI filtering affected more lower versus high intensity regions unlike EARL1 gaussian post filters, affecting both in a similar way. In lesions, the majority of texture features 79/100 (79%) were significantly (p<0.05) different between AI denoised and original PET images.ConclusionsApplying an AI-based denoising on FDG PET images maintains most of the lesion’s texture information in contrast to EARL1-compatible Gaussian filter. Predictive features of a trained model could be thus the same, however with an adapted threshold. Artificial intelligence based denoising in PET is a very promising approach as it adapts the denoising in function of the tissue type, preserving information where it should. |
| first_indexed | 2024-12-20T02:04:37Z |
| format | Article |
| id | doaj.art-91982b7465194e598780b283a8f32fa5 |
| institution | Directory Open Access Journal |
| issn | 2234-943X |
| language | English |
| last_indexed | 2024-12-20T02:04:37Z |
| publishDate | 2021-08-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Oncology |
| spelling | doaj.art-91982b7465194e598780b283a8f32fa52022-12-21T19:57:14ZengFrontiers Media S.A.Frontiers in Oncology2234-943X2021-08-011110.3389/fonc.2021.692973692973The Impact of Artificial Intelligence CNN Based Denoising on FDG PET RadiomicsCyril Jaudet0Kathleen Weyts1Alexis Lechervy2Alain Batalla3Stéphane Bardet4Aurélien Corroyer-Dulmont5Aurélien Corroyer-Dulmont6Medical Physics Department, CLCC François Baclesse, Caen, FranceNuclear Medicine Department, CLCC François Baclesse, Caen, FranceUMR GREYC, Normandie Univ, UNICAEN, ENSICAEN, CNRS, Caen, FranceMedical Physics Department, CLCC François Baclesse, Caen, FranceNuclear Medicine Department, CLCC François Baclesse, Caen, FranceMedical Physics Department, CLCC François Baclesse, Caen, FranceNormandie Univ, UNICAEN, CEA, CNRS, ISTCT/CERVOxy group, GIP CYCERON, Caen, FranceBackgroundWith a constantly increasing number of diagnostic images performed each year, Artificial Intelligence (AI) denoising methods offer an opportunity to respond to the growing demand. However, it may affect information in the image in an unknown manner. This study quantifies the effect of AI-based denoising on FDG PET textural information in comparison to a convolution with a standard gaussian postfilter (EARL1).MethodsThe study was carried out on 113 patients who underwent a digital FDG PET/CT (VEREOS, Philips Healthcare). 101 FDG avid lesions were segmented semi-automatically by a nuclear medicine physician. VOIs in the liver and lung as reference organs were contoured. PET textural features were extracted with pyradiomics. Texture features from AI denoised and EARL1 versus original PET images were compared with a Concordance Correlation Coefficient (CCC). Features with CCC values ≥ 0.85 threshold were considered concordant. Scatter plots of variable pairs with R2 coefficients of the more relevant features were computed. A Wilcoxon signed rank test to compare the absolute values between AI denoised and original images was performed.ResultsThe ratio of concordant features was 90/104 (86.5%) in AI denoised versus 46/104 (44.2%) with EARL1 denoising. In the reference organs, the concordant ratio for AI and EARL1 denoised images was low, respectively 12/104 (11.5%) and 7/104 (6.7%) in the liver, 26/104 (25%) and 24/104 (23.1%) in the lung. SUVpeak was stable after the application of both algorithms in comparison to SUVmax. Scatter plots of variable pairs showed that AI filtering affected more lower versus high intensity regions unlike EARL1 gaussian post filters, affecting both in a similar way. In lesions, the majority of texture features 79/100 (79%) were significantly (p<0.05) different between AI denoised and original PET images.ConclusionsApplying an AI-based denoising on FDG PET images maintains most of the lesion’s texture information in contrast to EARL1-compatible Gaussian filter. Predictive features of a trained model could be thus the same, however with an adapted threshold. Artificial intelligence based denoising in PET is a very promising approach as it adapts the denoising in function of the tissue type, preserving information where it should.https://www.frontiersin.org/articles/10.3389/fonc.2021.692973/fulldenoisingAIPETradiomicsmedical imagingconvolutional neural network |
| spellingShingle | Cyril Jaudet Kathleen Weyts Alexis Lechervy Alain Batalla Stéphane Bardet Aurélien Corroyer-Dulmont Aurélien Corroyer-Dulmont The Impact of Artificial Intelligence CNN Based Denoising on FDG PET Radiomics Frontiers in Oncology denoising AI PET radiomics medical imaging convolutional neural network |
| title | The Impact of Artificial Intelligence CNN Based Denoising on FDG PET Radiomics |
| title_full | The Impact of Artificial Intelligence CNN Based Denoising on FDG PET Radiomics |
| title_fullStr | The Impact of Artificial Intelligence CNN Based Denoising on FDG PET Radiomics |
| title_full_unstemmed | The Impact of Artificial Intelligence CNN Based Denoising on FDG PET Radiomics |
| title_short | The Impact of Artificial Intelligence CNN Based Denoising on FDG PET Radiomics |
| title_sort | impact of artificial intelligence cnn based denoising on fdg pet radiomics |
| topic | denoising AI PET radiomics medical imaging convolutional neural network |
| url | https://www.frontiersin.org/articles/10.3389/fonc.2021.692973/full |
| work_keys_str_mv | AT cyriljaudet theimpactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT kathleenweyts theimpactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT alexislechervy theimpactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT alainbatalla theimpactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT stephanebardet theimpactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT aureliencorroyerdulmont theimpactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT aureliencorroyerdulmont theimpactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT cyriljaudet impactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT kathleenweyts impactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT alexislechervy impactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT alainbatalla impactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT stephanebardet impactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT aureliencorroyerdulmont impactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics AT aureliencorroyerdulmont impactofartificialintelligencecnnbaseddenoisingonfdgpetradiomics |