Shaan Khurshid1,2, Samuel Friedman3, James P Pirruccello1,2, Paolo Di Achille3, Nathaniel Diamant3, Christopher D Anderson4,5,2, Patrick T Ellinor6,2, Puneet Batra3, Jennifer E Ho1,2, Anthony A Philippakis3, Steven A Lubitz2. 1. Division of Cardiology (S.K., J.P.P., J.E.H.), Massachusetts General Hospital, Boston. 2. Cardiovascular Disease Initiative (S.K., J.P.P., C.D.A., P.T.E., J.E.H., S.A.L.), Broad Institute of Harvard and the Massachusetts Institute of Technology, Cambridge. 3. Data Sciences Platform (S.F., P.D.A., N.D., P.B., A.A.P.), Broad Institute of Harvard and the Massachusetts Institute of Technology, Cambridge. 4. Center for Genomic Medicine (C.D.A.), Massachusetts General Hospital, Boston. 5. Henry and Allison McCance Center for Brain Health (C.D.A.), Massachusetts General Hospital, Boston. 6. Cardiac Arrhythmia Service (P.T.E.), Massachusetts General Hospital, Boston.
Abstract
BACKGROUND: Classical methods for detecting left ventricular (LV) hypertrophy (LVH) using 12-lead ECGs are insensitive. Deep learning models using ECG to infer cardiac magnetic resonance (CMR)-derived LV mass may improve LVH detection. METHODS: Within 32 239 individuals of the UK Biobank prospective cohort who underwent CMR and 12-lead ECG, we trained a convolutional neural network to predict CMR-derived LV mass using 12-lead ECGs (left ventricular mass-artificial intelligence [LVM-AI]). In independent test sets (UK Biobank [n=4903] and Mass General Brigham [MGB, n=1371]), we assessed correlation between LVM-AI predicted and CMR-derived LV mass and compared LVH discrimination using LVM-AI versus traditional ECG-based rules (ie, Sokolow-Lyon, Cornell, lead aVL rule, or any ECG rule). In the UK Biobank and an ambulatory MGB cohort (MGB outcomes, n=28 612), we assessed associations between LVM-AI predicted LVH and incident cardiovascular outcomes using age- and sex-adjusted Cox regression. RESULTS: LVM-AI predicted LV mass correlated with CMR-derived LV mass in both test sets, although correlation was greater in the UK Biobank (r=0.79) versus MGB (r=0.60, P<0.001 for both). When compared with any ECG rule, LVM-AI demonstrated similar LVH discrimination in the UK Biobank (LVM-AI c-statistic 0.653 [95% CI, 0.608 -0.698] versus any ECG rule c-statistic 0.618 [95% CI, 0.574 -0.663], P=0.11) and superior discrimination in MGB (0.621; 95% CI, 0.592 -0.649 versus 0.588; 95% CI, 0.564 -0.611, P=0.02). LVM-AI-predicted LVH was associated with incident atrial fibrillation, myocardial infarction, heart failure, and ventricular arrhythmias. CONCLUSIONS: Deep learning-inferred LV mass estimates from 12-lead ECGs correlate with CMR-derived LV mass, associate with incident cardiovascular disease, and may improve LVH discrimination compared to traditional ECG rules.
BACKGROUND: Classical methods for detecting left ventricular (LV) hypertrophy (LVH) using 12-lead ECGs are insensitive. Deep learning models using ECG to infer cardiac magnetic resonance (CMR)-derived LV mass may improve LVH detection. METHODS: Within 32 239 individuals of the UK Biobank prospective cohort who underwent CMR and 12-lead ECG, we trained a convolutional neural network to predict CMR-derived LV mass using 12-lead ECGs (left ventricular mass-artificial intelligence [LVM-AI]). In independent test sets (UK Biobank [n=4903] and Mass General Brigham [MGB, n=1371]), we assessed correlation between LVM-AI predicted and CMR-derived LV mass and compared LVH discrimination using LVM-AI versus traditional ECG-based rules (ie, Sokolow-Lyon, Cornell, lead aVL rule, or any ECG rule). In the UK Biobank and an ambulatory MGB cohort (MGB outcomes, n=28 612), we assessed associations between LVM-AI predicted LVH and incident cardiovascular outcomes using age- and sex-adjusted Cox regression. RESULTS: LVM-AI predicted LV mass correlated with CMR-derived LV mass in both test sets, although correlation was greater in the UK Biobank (r=0.79) versus MGB (r=0.60, P<0.001 for both). When compared with any ECG rule, LVM-AI demonstrated similar LVH discrimination in the UK Biobank (LVM-AI c-statistic 0.653 [95% CI, 0.608 -0.698] versus any ECG rule c-statistic 0.618 [95% CI, 0.574 -0.663], P=0.11) and superior discrimination in MGB (0.621; 95% CI, 0.592 -0.649 versus 0.588; 95% CI, 0.564 -0.611, P=0.02). LVM-AI-predicted LVH was associated with incident atrial fibrillation, myocardial infarction, heart failure, and ventricular arrhythmias. CONCLUSIONS: Deep learning-inferred LV mass estimates from 12-lead ECGs correlate with CMR-derived LV mass, associate with incident cardiovascular disease, and may improve LVH discrimination compared to traditional ECG rules.
Authors: Jonathan Chrispin; Aditya Jain; Elsayed Z Soliman; Eliseo Guallar; Alvaro Alonso; Susan R Heckbert; David A Bluemke; João A C Lima; Saman Nazarian Journal: J Am Coll Cardiol Date: 2014-03-19 Impact factor: 24.094
Authors: P N Casale; R B Devereux; P Kligfield; R R Eisenberg; D H Miller; B S Chaudhary; M C Phillips Journal: J Am Coll Cardiol Date: 1985-09 Impact factor: 24.094
Authors: Nadine Kawel-Boehm; Richard Kronmal; John Eng; Aaron Folsom; Gregory Burke; J Jeffrey Carr; Steven Shea; João A C Lima; David A Bluemke Journal: Radiology Date: 2019-08-27 Impact factor: 29.146
Authors: Steffen E Petersen; Nay Aung; Mihir M Sanghvi; Filip Zemrak; Kenneth Fung; Jose Miguel Paiva; Jane M Francis; Mohammed Y Khanji; Elena Lukaschuk; Aaron M Lee; Valentina Carapella; Young Jin Kim; Paul Leeson; Stefan K Piechnik; Stefan Neubauer Journal: J Cardiovasc Magn Reson Date: 2017-02-03 Impact factor: 5.364
Authors: Steffen E Petersen; Paul M Matthews; Jane M Francis; Matthew D Robson; Filip Zemrak; Redha Boubertakh; Alistair A Young; Sarah Hudson; Peter Weale; Steve Garratt; Rory Collins; Stefan Piechnik; Stefan Neubauer Journal: J Cardiovasc Magn Reson Date: 2016-02-01 Impact factor: 5.364
Authors: Olivia L Hulme; Shaan Khurshid; Lu-Chen Weng; Christopher D Anderson; Elizabeth Y Wang; Jeffrey M Ashburner; Darae Ko; David D McManus; Emelia J Benjamin; Patrick T Ellinor; Ludovic Trinquart; Steven A Lubitz Journal: JACC Clin Electrophysiol Date: 2019-10-02
Authors: Shaan Khurshid; Samuel Friedman; James P Pirruccello; Paolo Di Achille; Nathaniel Diamant; Christopher D Anderson; Patrick T Ellinor; Puneet Batra; Jennifer E Ho; Anthony A Philippakis; Steven A Lubitz Journal: Circ Cardiovasc Imaging Date: 2021-06-15 Impact factor: 8.589
Authors: Nathaniel Diamant; Paolo Di Achille; Lu-Chen Weng; Emily S Lau; Shaan Khurshid; Samuel Friedman; Christopher Reeder; Pulkit Singh; Xin Wang; Gopal Sarma; Mercedeh Ghadessi; Johanna Mielke; Eren Elci; Ivan Kryukov; Hanna M Eilken; Andrea Derix; Patrick T Ellinor; Christopher D Anderson; Anthony A Philippakis; Puneet Batra; Steven A Lubitz; Jennifer E Ho Journal: Cardiovasc Digit Health J Date: 2022-06-24