Fully Automated, Quality-Controlled Cardiac Analysis From CMR: Validation and Large-Scale Application to Characterize Cardiac Function.
Aged
Automation
Deep Learning
/ standards
Diagnosis, Computer-Assisted
/ standards
Female
Heart Diseases
/ diagnostic imaging
Humans
Image Interpretation, Computer-Assisted
/ standards
Magnetic Resonance Imaging, Cine
/ standards
Male
Middle Aged
Predictive Value of Tests
Quality Control
Quality Indicators, Health Care
/ standards
Reproducibility of Results
Stroke Volume
Ventricular Function, Left
Ventricular Function, Right
CMR feature tracking
cardiac aging
cardiac function
cardiac magnetic resonance
machine learning
quality control
Journal
JACC. Cardiovascular imaging
ISSN: 1876-7591
Titre abrégé: JACC Cardiovasc Imaging
Pays: United States
ID NLM: 101467978
Informations de publication
Date de publication:
03 2020
03 2020
Historique:
received:
30
01
2019
revised:
26
04
2019
accepted:
16
05
2019
pubmed:
22
7
2019
medline:
11
11
2020
entrez:
22
7
2019
Statut:
ppublish
Résumé
This study sought to develop a fully automated framework for cardiac function analysis from cardiac magnetic resonance (CMR), including comprehensive quality control (QC) algorithms to detect erroneous output. Analysis of cine CMR imaging using deep learning (DL) algorithms could automate ventricular function assessment. However, variable image quality, variability in phenotypes of disease, and unavoidable weaknesses in training of DL algorithms currently prevent their use in clinical practice. The framework consists of a pre-analysis DL image QC, followed by a DL algorithm for biventricular segmentation in long-axis and short-axis views, myocardial feature-tracking (FT), and a post-analysis QC to detect erroneous results. The study validated the framework in healthy subjects and cardiac patients by comparison against manual analysis (n = 100) and evaluation of the QC steps' ability to detect erroneous results (n = 700). Next, this method was used to obtain reference values for cardiac function metrics from the UK Biobank. Automated analysis correlated highly with manual analysis for left and right ventricular volumes (all r > 0.95), strain (circumferential r = 0.89, longitudinal r > 0.89), and filling and ejection rates (all r ≥ 0.93). There was no significant bias for cardiac volumes and filling and ejection rates, except for right ventricular end-systolic volume (bias +1.80 ml; p = 0.01). The bias for FT strain was <1.3%. The sensitivity of detection of erroneous output was 95% for volume-derived parameters and 93% for FT strain. Finally, reference values were automatically derived from 2,029 CMR exams in healthy subjects. The study demonstrates a DL-based framework for automated, quality-controlled characterization of cardiac function from cine CMR, without the need for direct clinician oversight.
Sections du résumé
OBJECTIVES
This study sought to develop a fully automated framework for cardiac function analysis from cardiac magnetic resonance (CMR), including comprehensive quality control (QC) algorithms to detect erroneous output.
BACKGROUND
Analysis of cine CMR imaging using deep learning (DL) algorithms could automate ventricular function assessment. However, variable image quality, variability in phenotypes of disease, and unavoidable weaknesses in training of DL algorithms currently prevent their use in clinical practice.
METHODS
The framework consists of a pre-analysis DL image QC, followed by a DL algorithm for biventricular segmentation in long-axis and short-axis views, myocardial feature-tracking (FT), and a post-analysis QC to detect erroneous results. The study validated the framework in healthy subjects and cardiac patients by comparison against manual analysis (n = 100) and evaluation of the QC steps' ability to detect erroneous results (n = 700). Next, this method was used to obtain reference values for cardiac function metrics from the UK Biobank.
RESULTS
Automated analysis correlated highly with manual analysis for left and right ventricular volumes (all r > 0.95), strain (circumferential r = 0.89, longitudinal r > 0.89), and filling and ejection rates (all r ≥ 0.93). There was no significant bias for cardiac volumes and filling and ejection rates, except for right ventricular end-systolic volume (bias +1.80 ml; p = 0.01). The bias for FT strain was <1.3%. The sensitivity of detection of erroneous output was 95% for volume-derived parameters and 93% for FT strain. Finally, reference values were automatically derived from 2,029 CMR exams in healthy subjects.
CONCLUSIONS
The study demonstrates a DL-based framework for automated, quality-controlled characterization of cardiac function from cine CMR, without the need for direct clinician oversight.
Identifiants
pubmed: 31326477
pii: S1936-878X(19)30585-6
doi: 10.1016/j.jcmg.2019.05.030
pmc: PMC7060799
mid: EMS84281
pii:
doi:
Types de publication
Journal Article
Research Support, Non-U.S. Gov't
Validation Study
Video-Audio Media
Langues
eng
Sous-ensembles de citation
IM
Pagination
684-695Subventions
Organisme : Wellcome Trust
ID : 203148
Pays : United Kingdom
Organisme : Wellcome Trust
Pays : United Kingdom
Organisme : Medical Research Council
ID : MC_QA137853
Pays : United Kingdom
Organisme : Medical Research Council
ID : MC_PC_17228
Pays : United Kingdom
Organisme : Wellcome Trust
ID : WT 203148/Z/16/Z
Pays : United Kingdom
Commentaires et corrections
Type : CommentIn
Informations de copyright
Crown Copyright © 2020. Published by Elsevier Inc. All rights reserved.
Références
J Am Coll Radiol. 2018 Mar;15(3 Pt B):504-508
pubmed: 29402533
Clin Radiol. 2015 Sep;70(9):989-98
pubmed: 26139384
Int J Cardiol. 2018 Nov 15;271:387-391
pubmed: 29885827
Nutrition. 1989 Sep-Oct;5(5):303-11; discussion 312-3
pubmed: 2520314
JACC Cardiovasc Imaging. 2015 Dec;8(12):1444-1460
pubmed: 26699113
IEEE Trans Cybern. 2017 May;47(5):1336-1349
pubmed: 28362600
J Cardiovasc Magn Reson. 2017 Sep 25;19(1):70
pubmed: 28942735
JACC Cardiovasc Imaging. 2018 Feb;11(2 Pt 1):196-205
pubmed: 28528164
Int J Cardiovasc Imaging. 2018 Feb;34(2):281-291
pubmed: 28836039
Med Image Comput Comput Assist Interv. 2011;14(Pt 3):479-86
pubmed: 22003734
Med Image Anal. 2017 Dec;42:60-88
pubmed: 28778026
J Cardiovasc Magn Reson. 2014 Jul 30;16:56
pubmed: 25160814
J Magn Reson Imaging. 2016 Nov;44(5):1186-1196
pubmed: 27143317
IEEE Trans Med Imaging. 2018 Nov;37(11):2514-2525
pubmed: 29994302
IEEE Trans Med Imaging. 2017 Jun;36(6):1221-1230
pubmed: 28391191
J Cardiovasc Magn Reson. 2018 Sep 14;20(1):65
pubmed: 30217194
J Cardiovasc Magn Reson. 2017 Feb 3;19(1):18
pubmed: 28178995
J Cardiovasc Magn Reson. 2015 Apr 18;17:29
pubmed: 25928314
J Med Eng. 2017;2017:4501647
pubmed: 28695126
Clin Chim Acta. 1984 Feb 14;137(1):97F-114F
pubmed: 6697530
J Cardiovasc Magn Reson. 2006;8(3):417-26
pubmed: 16755827
J Cardiovasc Magn Reson. 2016 Feb 01;18:8
pubmed: 26830817