Toward Reduction in False-Positive Thyroid Nodule Biopsies with a Deep Learning-based Risk Stratification System Using US Cine-Clip Images.
Abdomen/GI
Computer Applications–3D
Convolutional Neural Network (CNN)
Diagnosis
Head/Neck
Neural Networks
Oncology
Supervised Learning
Thyroid
Transfer Learning
US
Journal
Radiology. Artificial intelligence
ISSN: 2638-6100
Titre abrégé: Radiol Artif Intell
Pays: United States
ID NLM: 101746556
Informations de publication
Date de publication:
May 2022
May 2022
Historique:
received:
21
06
2021
revised:
16
01
2022
accepted:
19
04
2022
entrez:
2
6
2022
pubmed:
3
6
2022
medline:
3
6
2022
Statut:
epublish
Résumé
To develop a deep learning-based risk stratification system for thyroid nodules using US cine images. In this retrospective study, 192 biopsy-confirmed thyroid nodules (175 benign, 17 malignant) in 167 unique patients (mean age, 56 years ± 16 [SD], 137 women) undergoing cine US between April 2017 and May 2018 with American College of Radiology (ACR) Thyroid Imaging Reporting and Data System (TI-RADS)-structured radiology reports were evaluated. A deep learning-based system that exploits the cine images obtained during three-dimensional volumetric thyroid scans and outputs malignancy risk was developed and compared, using fivefold cross-validation, against a two-dimensional (2D) deep learning-based model (Static-2DCNN), a radiomics-based model using cine images (Cine-Radiomics), and the ACR TI-RADS level, with histopathologic diagnosis as ground truth. The system was used to revise the ACR TI-RADS recommendation, and its diagnostic performance was compared against the original ACR TI-RADS. The system achieved higher average area under the receiver operating characteristic curve (AUC, 0.88) than Static-2DCNN (0.72, The risk stratification system using US cine images had higher diagnostic performance than prior models and improved specificity of ACR TI-RADS when used to revise ACR TI-RADS recommendation.
Identifiants
pubmed: 35652118
doi: 10.1148/ryai.210174
pmc: PMC9152684
doi:
Types de publication
Journal Article
Langues
eng
Pagination
e210174Subventions
Organisme : NCI NIH HHS
ID : K00 CA234954
Pays : United States
Organisme : NCI NIH HHS
ID : T32 CA009695
Pays : United States
Informations de copyright
© 2022 by the Radiological Society of North America, Inc.
Déclaration de conflit d'intérêts
Disclosures of conflicts of interest: R.Y. No relevant relationships. T.K. No relevant relationships. M.N.A. No relevant relationships. A.G. No relevant relationships. S.A.S. No relevant relationships. M.U.A. No relevant relationships. E.A. Support for this article from National Institutes of Health (NIH) (award nos. PADLY, PADPM, PAEPI, PAFFL, PAWAO, PCOXK, PCQUD, PCRMR); grants or contracts from National Science Foundation (NSF) (award no. QCBFG), American College of Radiology (AMRAD) (award no. UAPJI), AstraZeneca (award no. UBERP), and Philips Electronics North America Corp (award no. UBGKW). A.L.W. No relevant relationships. N.M. Primary investigator in study of MRI contrast agents, supported by Bracco Diagnostics. D.G. No relevant relationships. V.B. No relevant relationships. N.D.S. Support for this manuscript from NIH fellowship NCI F99/K00 Predoctoral-to-Postdoctoral Fellow Transition Award (K00CA234954) (payment made to Stanford); consulting fees paid to author from Focused Ultrasound Foundation. H.S. No relevant relationships. D.L.R. Grant from NSF in support for this article; grants from NIH, Food and Drug Administration, GE Medical Systems, and Philips; royalty, CRC press for book Radiomics and Radiogenomics; consulting fees from Roche Genentech; provisional patents for Evaluating Artificial Intelligence Applications in Clinical Practice (62/812,905), Automatic Organ Segmentation and Lesion Detection (62/749,053), and Automated Annotation of Text Reports to Enable Developing AI Applications (62/814,225); associate editor for Radiology: Artificial Intelligence. T.S.D. No relevant relationships.
Références
Radiology. 2016 Feb;278(2):563-77
pubmed: 26579733
J Ultrasound Med. 2012 Feb;31(2):197-204
pubmed: 22298862
AJR Am J Roentgenol. 2018 Jul;211(1):162-167
pubmed: 29702015
Radiology. 2018 Apr;287(1):185-193
pubmed: 29498593
Tomography. 2019 Mar;5(1):170-183
pubmed: 30854455
Neural Netw. 2018 Oct;106:249-259
pubmed: 30092410
Thyroid. 2016 Jan;26(1):1-133
pubmed: 26462967
Cancers Head Neck. 2017 Jan 11;2:1
pubmed: 31093348
J Surg Res. 2013 Oct;184(2):761-5
pubmed: 23623584
Stat Med. 2013 Mar 15;32(6):964-77
pubmed: 22912343
Sci Rep. 2018 Apr 26;8(1):6600
pubmed: 29700427
Ultrasound Med Biol. 2015 Dec;41(12):3096-101
pubmed: 26411668
IEEE Trans Pattern Anal Mach Intell. 2020 Feb;42(2):318-327
pubmed: 30040631
Front Oncol. 2020 Nov 12;10:591846
pubmed: 33282741
Lancet Oncol. 2019 Feb;20(2):193-201
pubmed: 30583848
JAMA Oncol. 2016 Aug 1;2(8):1023-9
pubmed: 27078145
AJR Am J Roentgenol. 2020 Apr;214(4):885-892
pubmed: 31967504
Eur Radiol. 2021 Apr;31(4):2405-2413
pubmed: 33034748
Cancer Res. 2017 Nov 1;77(21):e104-e107
pubmed: 29092951
Radiology. 2019 Sep;292(3):695-701
pubmed: 31287391
Sci Rep. 2019 Nov 28;9(1):17843
pubmed: 31780753
Front Oncol. 2021 Apr 27;11:575166
pubmed: 33987082
Eur Radiol. 2016 Oct;26(10):3353-60
pubmed: 26795614
Eur Radiol. 2006 Feb;16(2):428-36
pubmed: 16155720
Medicine (Baltimore). 2019 Apr;98(15):e15133
pubmed: 30985680
Thyroid. 2020 Jun;30(6):878-884
pubmed: 32013775
Lancet Digit Health. 2021 Apr;3(4):e250-e259
pubmed: 33766289
Thyroid Res. 2011 Jan 07;4(1):1
pubmed: 21211056
Ultrasound Med Biol. 2021 May;47(5):1299-1309
pubmed: 33583636
J Am Coll Radiol. 2017 May;14(5):587-595
pubmed: 28372962