The Impact of Resampling and Denoising Deep Learning Algorithms on Radiomics in Brain Metastases MRI.
MRI
deep learning
denoising
radiomics
resampling
Journal
Cancers
ISSN: 2072-6694
Titre abrégé: Cancers (Basel)
Pays: Switzerland
ID NLM: 101526829
Informations de publication
Date de publication:
22 Dec 2021
22 Dec 2021
Historique:
received:
29
09
2021
revised:
14
12
2021
accepted:
18
12
2021
entrez:
11
1
2022
pubmed:
12
1
2022
medline:
12
1
2022
Statut:
epublish
Résumé
Magnetic resonance imaging (MRI) is predominant in the therapeutic management of cancer patients, unfortunately, patients have to wait a long time to get an appointment for examination. Therefore, new MRI devices include deep-learning (DL) solutions to save acquisition time. However, the impact of these algorithms on intensity and texture parameters has been poorly studied. The aim of this study was to evaluate the impact of resampling and denoising DL models on radiomics. Resampling and denoising DL model was developed on 14,243 T1 brain images from 1.5T-MRI. Radiomics were extracted from 40 brain metastases from 11 patients (2049 images). A total of 104 texture features of DL images were compared to original images with paired When two times shorter image acquisition shows strong disparities with the originals concerning the radiomics, with significant differences and loss of correlation of 79.81% and 48.08%, respectively. Interestingly, DL models restore textures with 46.15% of unstable parameters and 25.96% of low CCC and without difference for the first-order intensity parameters. Resampling and denoising DL models reconstruct low resolution and noised MRI images acquired quickly into high quality images. While fast MRI acquisition loses most of the radiomic features, DL models restore these parameters.
Sections du résumé
BACKGROUND
BACKGROUND
Magnetic resonance imaging (MRI) is predominant in the therapeutic management of cancer patients, unfortunately, patients have to wait a long time to get an appointment for examination. Therefore, new MRI devices include deep-learning (DL) solutions to save acquisition time. However, the impact of these algorithms on intensity and texture parameters has been poorly studied. The aim of this study was to evaluate the impact of resampling and denoising DL models on radiomics.
METHODS
METHODS
Resampling and denoising DL model was developed on 14,243 T1 brain images from 1.5T-MRI. Radiomics were extracted from 40 brain metastases from 11 patients (2049 images). A total of 104 texture features of DL images were compared to original images with paired
RESULTS
RESULTS
When two times shorter image acquisition shows strong disparities with the originals concerning the radiomics, with significant differences and loss of correlation of 79.81% and 48.08%, respectively. Interestingly, DL models restore textures with 46.15% of unstable parameters and 25.96% of low CCC and without difference for the first-order intensity parameters.
CONCLUSIONS
CONCLUSIONS
Resampling and denoising DL models reconstruct low resolution and noised MRI images acquired quickly into high quality images. While fast MRI acquisition loses most of the radiomic features, DL models restore these parameters.
Identifiants
pubmed: 35008198
pii: cancers14010036
doi: 10.3390/cancers14010036
pmc: PMC8750741
pii:
doi:
Types de publication
Journal Article
Langues
eng
Références
PeerJ. 2019 Oct 4;7:e7702
pubmed: 31592346
Front Oncol. 2019 Jun 26;9:552
pubmed: 31293979
Yearb Med Inform. 2019 Aug;28(1):16-26
pubmed: 31419814
Magn Reson Imaging. 2012 Nov;30(9):1323-41
pubmed: 22770690
Magn Reson Med Sci. 2020 Feb 10;19(1):64-76
pubmed: 31061259
Radiology. 2020 May;295(2):328-338
pubmed: 32154773
Magn Reson Med. 1996 Sep;36(3):494-7
pubmed: 8875425
Cancer Res. 2017 Nov 1;77(21):e104-e107
pubmed: 29092951
IEEE Trans Image Process. 2017 Jul;26(7):3142-3155
pubmed: 28166495
Front Oncol. 2021 Aug 24;11:692973
pubmed: 34504782
Clin Radiol. 2021 Aug;76(8):629.e11-629.e18
pubmed: 34092362
IEEE Trans Image Process. 2020 Aug 12;PP:
pubmed: 32784137
Radiother Oncol. 2017 Oct;125(1):147-153
pubmed: 28797700
Biometrics. 1989 Mar;45(1):255-68
pubmed: 2720055
Radiol Artif Intell. 2020 Jan 29;2(1):e190007
pubmed: 32076662
Front Neurol. 2020 Feb 07;11:1
pubmed: 32116995
Magn Reson Med. 1993 Jul;30(1):51-9
pubmed: 8371675
Tomography. 2016 Dec;2(4):361-365
pubmed: 30042967
Eur J Nucl Med Mol Imaging. 2017 Aug;44(Suppl 1):17-31
pubmed: 28623376
Semin Nucl Med. 2021 Mar;51(2):126-133
pubmed: 33509369
Magn Reson Med. 1999 Nov;42(5):952-62
pubmed: 10542355
Acta Oncol. 2013 Oct;52(7):1391-7
pubmed: 24047337
Eur J Cancer. 2012 Mar;48(4):441-6
pubmed: 22257792
Br J Radiol. 2015;88(1056):20150487
pubmed: 26402216
Magn Reson Med. 1986 Aug;3(4):604-18
pubmed: 3747821
EJNMMI Phys. 2019 Dec 26;6(1):28
pubmed: 31879795
Nat Commun. 2014 Jun 03;5:4006
pubmed: 24892406
Nat Methods. 2012 Jul;9(7):671-5
pubmed: 22930834
AJNR Am J Neuroradiol. 2018 Oct;39(10):1776-1784
pubmed: 29419402
Sci Rep. 2013;3:1364
pubmed: 23455483
Cancers (Basel). 2021 Aug 10;13(16):
pubmed: 34439186
Sci Rep. 2019 Mar 18;9(1):4800
pubmed: 30886309