Objective: We aim to provide an algorithm for the detection of myocardial infarction that operates directly on ECG data without any preprocessing and to investigate its decision criteria. Approach: We train an ensemble of fully convolutional neural networks on the PTB ECG dataset and apply state-of-the-art attribution methods. Main results: Our classifier reaches 93.3% sensitivity and 89.7% specificity evaluated using 10-fold cross-validation with sampling based on patients. The presented method outperforms state-of-the-art approaches and reaches the performance level of human cardiologists for detection of myocardial infarction. We are able to discriminate channel-specific regions that contribute most significantly to the neural network’s decision. Interestingly, the network’s decision is influenced by signs also recognized by human cardiologists as indicative of myocardial infarction. Significance: Our results demonstrate the high prospects of algorithmic ECG analysis for future clinical applications considering both its quantitative performance as well as the possibility of assessing decision criteria on a per-example basis, which enhances the comprehensibility of the approach.
%0 Journal Article
%1 strodthoff2019detecting
%A Strodthoff, Nils
%A Strodthoff, Claas
%D 2019
%I IOP Publishing
%J Physiological Measurement
%K convnet deep ecg infarction learning myocardial
%N 1
%P 015001
%R 10.1088/1361-6579/aaf34d
%T Detecting and interpreting myocardial infarction using fully convolutional neural networks
%U https://doi.org/10.1088%2F1361-6579%2Faaf34d
%V 40
%X Objective: We aim to provide an algorithm for the detection of myocardial infarction that operates directly on ECG data without any preprocessing and to investigate its decision criteria. Approach: We train an ensemble of fully convolutional neural networks on the PTB ECG dataset and apply state-of-the-art attribution methods. Main results: Our classifier reaches 93.3% sensitivity and 89.7% specificity evaluated using 10-fold cross-validation with sampling based on patients. The presented method outperforms state-of-the-art approaches and reaches the performance level of human cardiologists for detection of myocardial infarction. We are able to discriminate channel-specific regions that contribute most significantly to the neural network’s decision. Interestingly, the network’s decision is influenced by signs also recognized by human cardiologists as indicative of myocardial infarction. Significance: Our results demonstrate the high prospects of algorithmic ECG analysis for future clinical applications considering both its quantitative performance as well as the possibility of assessing decision criteria on a per-example basis, which enhances the comprehensibility of the approach.
@article{strodthoff2019detecting,
abstract = {Objective: We aim to provide an algorithm for the detection of myocardial infarction that operates directly on ECG data without any preprocessing and to investigate its decision criteria. Approach: We train an ensemble of fully convolutional neural networks on the PTB ECG dataset and apply state-of-the-art attribution methods. Main results: Our classifier reaches 93.3% sensitivity and 89.7% specificity evaluated using 10-fold cross-validation with sampling based on patients. The presented method outperforms state-of-the-art approaches and reaches the performance level of human cardiologists for detection of myocardial infarction. We are able to discriminate channel-specific regions that contribute most significantly to the neural network’s decision. Interestingly, the network’s decision is influenced by signs also recognized by human cardiologists as indicative of myocardial infarction. Significance: Our results demonstrate the high prospects of algorithmic ECG analysis for future clinical applications considering both its quantitative performance as well as the possibility of assessing decision criteria on a per-example basis, which enhances the comprehensibility of the approach.},
added-at = {2020-04-23T15:52:44.000+0200},
author = {Strodthoff, Nils and Strodthoff, Claas},
biburl = {https://www.bibsonomy.org/bibtex/2e6fc3839b9d159194e5eac2d074a8b92/nosebrain},
doi = {10.1088/1361-6579/aaf34d},
interhash = {3620fd4092d685dde400e33f2e0a789a},
intrahash = {e6fc3839b9d159194e5eac2d074a8b92},
journal = {Physiological Measurement},
keywords = {convnet deep ecg infarction learning myocardial},
month = jan,
number = 1,
pages = 015001,
publisher = {{IOP} Publishing},
timestamp = {2020-04-23T15:53:42.000+0200},
title = {Detecting and interpreting myocardial infarction using fully convolutional neural networks},
url = {https://doi.org/10.1088%2F1361-6579%2Faaf34d},
volume = 40,
year = 2019
}