Clinical value of radiomics and machine learning in breast ultrasound: a multicenter study for differential diagnosis of benign and malignant lesions.
Breast cancer
Machine learning
Ultrasound
Journal
European radiology
ISSN: 1432-1084
Titre abrégé: Eur Radiol
Pays: Germany
ID NLM: 9114774
Informations de publication
Date de publication:
Dec 2021
Dec 2021
Historique:
received:
18
11
2020
accepted:
22
04
2021
revised:
06
04
2021
pubmed:
22
5
2021
medline:
17
11
2021
entrez:
21
5
2021
Statut:
ppublish
Résumé
We aimed to assess the performance of radiomics and machine learning (ML) for classification of non-cystic benign and malignant breast lesions on ultrasound images, compare ML's accuracy with that of a breast radiologist, and verify if the radiologist's performance is improved by using ML. Our retrospective study included patients from two institutions. A total of 135 lesions from Institution 1 were used to train and test the ML model with cross-validation. Radiomic features were extracted from manually annotated images and underwent a multistep feature selection process. Not reproducible, low variance, and highly intercorrelated features were removed from the dataset. Then, 66 lesions from Institution 2 were used as an external test set for ML and to assess the performance of a radiologist without and with the aid of ML, using McNemar's test. After feature selection, 10 of the 520 features extracted were employed to train a random forest algorithm. Its accuracy in the training set was 82% (standard deviation, SD, ± 6%), with an AUC of 0.90 (SD ± 0.06), while the performance on the test set was 82% (95% confidence intervals (CI) = 70-90%) with an AUC of 0.82 (95% CI = 0.70-0.93). It resulted in being significantly better than the baseline reference (p = 0.0098), but not different from the radiologist (79.4%, p = 0.815). The radiologist's performance improved when using ML (80.2%), but not significantly (p = 0.508). A radiomic analysis combined with ML showed promising results to differentiate benign from malignant breast lesions on ultrasound images. • Machine learning showed good accuracy in discriminating benign from malignant breast lesions • The machine learning classifier's performance was comparable to that of a breast radiologist • The radiologist's accuracy improved with machine learning, but not significantly.
Identifiants
pubmed: 34018057
doi: 10.1007/s00330-021-08009-2
pii: 10.1007/s00330-021-08009-2
pmc: PMC8589755
doi:
Types de publication
Journal Article
Multicenter Study
Langues
eng
Sous-ensembles de citation
IM
Pagination
9511-9519Informations de copyright
© 2021. The Author(s).
Références
Ultrasound Med Biol. 2011 Apr;37(4):539-48
pubmed: 21420580
Annu Int Conf IEEE Eng Med Biol Soc. 2016 Aug;2016:3342-3345
pubmed: 28269019
Insights Imaging. 2020 Aug 12;11(1):91
pubmed: 32785796
Radiographics. 2017 Mar-Apr;37(2):505-515
pubmed: 28212054
Front Neuroinform. 2014 Feb 21;8:14
pubmed: 24600388
BMC Med Imaging. 2019 Jul 1;19(1):51
pubmed: 31262255
Biomed Res Int. 2018 Oct 8;2018:6120703
pubmed: 30402486
Insights Imaging. 2018 Aug;9(4):449-461
pubmed: 30094592
Cancer Res. 2017 Nov 1;77(21):e104-e107
pubmed: 29092951
Sci Rep. 2018 Sep 10;8(1):13546
pubmed: 30202040
Breast. 2020 Feb;49:74-80
pubmed: 31739125
Comput Med Imaging Graph. 2002 Sep-Oct;26(5):303-7
pubmed: 12204234
Br J Radiol. 2018 Feb;91(1083):20170576
pubmed: 29215311
Radiol Artif Intell. 2019 Jul 10;1(4):e180075
pubmed: 33937796
Nat Rev Clin Oncol. 2017 Dec;14(12):749-762
pubmed: 28975929
Nat Rev Cancer. 2018 Aug;18(8):500-510
pubmed: 29777175
J Chiropr Med. 2016 Jun;15(2):155-63
pubmed: 27330520
Cancer Lett. 2020 Jul 1;481:55-62
pubmed: 32251707
Radiology. 2018 Aug;288(2):318-328
pubmed: 29944078
Sci Rep. 2019 Aug 15;9(1):11921
pubmed: 31417138
Sci Rep. 2019 Jul 1;9(1):9441
pubmed: 31263116
Oman Med J. 2018 Sep;33(5):380-386
pubmed: 30210716
Eur Radiol Exp. 2018 Nov 14;2(1):36
pubmed: 30426318
Eur Radiol Exp. 2019 Aug 5;3(1):34
pubmed: 31385114
Eur Radiol. 2020 May;30(5):3023-3033
pubmed: 32006174
J Glob Oncol. 2019 Aug;5:1-17
pubmed: 31454282