Automated detection of lung nodules and coronary artery calcium using artificial intelligence on low-dose CT scans for lung cancer screening: accuracy and prognostic value.
Artificial intelligence
Cardiothoracic imaging
Convolutional neural networks
Coronary artery disease
Deep learning
Lung cancer screening
Journal
BMC medicine
ISSN: 1741-7015
Titre abrégé: BMC Med
Pays: England
ID NLM: 101190723
Informations de publication
Date de publication:
04 03 2021
04 03 2021
Historique:
received:
10
11
2020
accepted:
26
01
2021
entrez:
4
3
2021
pubmed:
5
3
2021
medline:
7
7
2021
Statut:
epublish
Résumé
Artificial intelligence (AI) in diagnostic radiology is undergoing rapid development. Its potential utility to improve diagnostic performance for cardiopulmonary events is widely recognized, but the accuracy and precision have yet to be demonstrated in the context of current screening modalities. Here, we present findings on the performance of an AI convolutional neural network (CNN) prototype (AI-RAD Companion, Siemens Healthineers) that automatically detects pulmonary nodules and quantifies coronary artery calcium volume (CACV) on low-dose chest CT (LDCT), and compare results to expert radiologists. We also correlate AI findings with adverse cardiopulmonary outcomes in a retrospective cohort of 117 patients who underwent LDCT. A total of 117 patients were enrolled in this study. Two CNNs were used to identify lung nodules and CACV on LDCT scans. All subjects were used for lung nodule analysis, and 96 subjects met the criteria for coronary artery calcium volume analysis. Interobserver concordance was measured using ICC and Cohen's kappa. Multivariate logistic regression and partial least squares regression were used for outcomes analysis. Agreement of the AI findings with experts was excellent (CACV ICC = 0.904, lung nodules Cohen's kappa = 0.846) with high sensitivity and specificity (CACV: sensitivity = .929, specificity = .960; lung nodules: sensitivity = 1, specificity = 0.708). The AI findings improved the prediction of major cardiopulmonary outcomes at 1-year follow-up including major adverse cardiac events and lung cancer (AUC We conclude the AI prototype rapidly and accurately identifies significant risk factors for cardiopulmonary disease on standard screening low-dose chest CT. This information can be used to improve diagnostic ability, facilitate intervention, improve morbidity and mortality, and decrease healthcare costs. There is also potential application in countries with limited numbers of cardiothoracic radiologists.
Sections du résumé
BACKGROUND
Artificial intelligence (AI) in diagnostic radiology is undergoing rapid development. Its potential utility to improve diagnostic performance for cardiopulmonary events is widely recognized, but the accuracy and precision have yet to be demonstrated in the context of current screening modalities. Here, we present findings on the performance of an AI convolutional neural network (CNN) prototype (AI-RAD Companion, Siemens Healthineers) that automatically detects pulmonary nodules and quantifies coronary artery calcium volume (CACV) on low-dose chest CT (LDCT), and compare results to expert radiologists. We also correlate AI findings with adverse cardiopulmonary outcomes in a retrospective cohort of 117 patients who underwent LDCT.
METHODS
A total of 117 patients were enrolled in this study. Two CNNs were used to identify lung nodules and CACV on LDCT scans. All subjects were used for lung nodule analysis, and 96 subjects met the criteria for coronary artery calcium volume analysis. Interobserver concordance was measured using ICC and Cohen's kappa. Multivariate logistic regression and partial least squares regression were used for outcomes analysis.
RESULTS
Agreement of the AI findings with experts was excellent (CACV ICC = 0.904, lung nodules Cohen's kappa = 0.846) with high sensitivity and specificity (CACV: sensitivity = .929, specificity = .960; lung nodules: sensitivity = 1, specificity = 0.708). The AI findings improved the prediction of major cardiopulmonary outcomes at 1-year follow-up including major adverse cardiac events and lung cancer (AUC
CONCLUSION
We conclude the AI prototype rapidly and accurately identifies significant risk factors for cardiopulmonary disease on standard screening low-dose chest CT. This information can be used to improve diagnostic ability, facilitate intervention, improve morbidity and mortality, and decrease healthcare costs. There is also potential application in countries with limited numbers of cardiothoracic radiologists.
Identifiants
pubmed: 33658025
doi: 10.1186/s12916-021-01928-3
pii: 10.1186/s12916-021-01928-3
pmc: PMC7931546
doi:
Substances chimiques
Calcium
SY7Q814VUP
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
55Références
Ann Intern Med. 2014 Mar 4;160(5):330-8
pubmed: 24378917
J Thorac Imaging. 2014 Sep;29(5):310-6
pubmed: 24992501
J Thorac Imaging. 2020 May;35 Suppl 1:S21-S27
pubmed: 32317574
J Am Coll Cardiol. 1996 Apr;27(5):978-90
pubmed: 8609364
JAMA. 2003 Aug 20;290(7):891-7
pubmed: 12928465
Acad Radiol. 2009 Jan;16(1):28-38
pubmed: 19064209
Lancet Oncol. 2014 Nov;15(12):1342-50
pubmed: 25282284
World J Radiol. 2014 Jun 28;6(6):381-7
pubmed: 24976939
Thorax. 2019 Jul;74(7):643-649
pubmed: 30862725
Circulation. 1998 May 12;97(18):1837-47
pubmed: 9603539
N Engl J Med. 2011 Aug 4;365(5):395-409
pubmed: 21714641
IEEE Trans Pattern Anal Mach Intell. 2017 Jun;39(6):1137-1149
pubmed: 27295650
N Engl J Med. 2013 May 23;368(21):1980-91
pubmed: 23697514
AJR Am J Roentgenol. 2010 May;194(5):1244-9
pubmed: 20410410
J Thorac Imaging. 2020 May;35 Suppl 1:S28-S34
pubmed: 32235188
JACC Cardiovasc Imaging. 2016 Dec;9(12):1407-1416
pubmed: 27085449
CA Cancer J Clin. 2020 Jan;70(1):7-30
pubmed: 31912902
Ann Intern Med. 2012 Dec 4;157(11):776-84
pubmed: 23208167
Curr Cardiovasc Imaging Rep. 2016;9:12
pubmed: 27057268
J Cardiovasc Comput Tomogr. 2020 Jan - Feb;14(1):12-17
pubmed: 30952612
Sci Rep. 2018 Jun 18;8(1):9286
pubmed: 29915334
Lung Cancer. 2019 Dec;138:72-78
pubmed: 31654837
J Am Heart Assoc. 2020 Jan 21;9(2):e014402
pubmed: 31937196
N Engl J Med. 2006 Oct 26;355(17):1763-71
pubmed: 17065637
Sci Rep. 2017 Apr 19;7:46479
pubmed: 28422152
Am Heart J. 2000 Feb;139(2 Pt 1):272-81
pubmed: 10650300
Radiology. 2010 Nov;257(2):541-8
pubmed: 20829542
IEEE Trans Med Imaging. 2012 Dec;31(12):2322-34
pubmed: 22961297
Radiology. 2017 Apr;283(1):49-58
pubmed: 27918707
Transl Lung Cancer Res. 2018 Jun;7(3):361-367
pubmed: 30050773
J Thorac Imaging. 2020 Mar;35(2):129-135
pubmed: 31651689