Performance of a convolutional neural network derived from an ECG database in recognizing myocardial infarction

Makimoto, Hisaki; Höckmann, Moritz; Lin, Tina; Glöckner, David; Gerguri, Shqipe; Clasen, Lukas; Schmidt, Jan; Assadi-Schmidt, Athena; Bejinariu, Alexandru; Müller, Patrick; Angendohr, Stephan; Babady, Mehran; Brinkmeyer, Christoph; Makimoto, Asuka; Kelm, Malte

doi:10.1038/s41598-020-65105-x

Cited by 68 publications

(43 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The heat maps help to visualize whether a DNN is concentrating on the same regions of the medical image that a human expert would focus attention on for a given diagnosis, rather than concentrating on a clinically irrelevant part of the medical image or even on confounders in the image [13][14][15] . However, recent work has shown that saliency methods used to validate model predictions can be misleading in some cases and may lead to increased bias and loss of user trust with concerning implications for clinical translation efforts 16 .…”

Section: Introductionmentioning

confidence: 99%

Benchmarking saliency methods for chest X-ray interpretation

Saporta

Gui

Agrawal

et al. 2021

Preprint

View full text Add to dashboard Cite

Deep learning has enabled automated medical image interpretation at a level often surpassing that of practicing medical experts. However, many clinical practices have cited a lack of model interpretability as reason to delay the use of "black-box" deep neural networks in clinical workflows. Saliency maps, which "explain" a model's decision by producing heat maps that highlight the areas of the medical image that influence model prediction, are often presented to clinicians as an aid in diagnostic decision-making. In this work, we demonstrate that the most commonly used saliency map generating method, Grad-CAM, results in low performance for 10 pathologies on chest X-rays. We examined under what clinical conditions saliency maps might be more dangerous to use compared to human experts, and found that Grad-CAM performs worse for pathologies that had multiple instances, were smaller in size, and had shapes that were more complex. Moreover, we showed that model confidence was positively correlated with Grad-CAM localization performance, suggesting that saliency maps were safer for clinicians to use as a decision aid when the model had made a positive prediction with high confidence. Our work demonstrates that several important limitations of interpretability techniques for medical imaging must be addressed before use in clinical workflows.

show abstract

Section: Introductionmentioning

confidence: 99%

Benchmarking saliency methods for chest X-ray interpretation

Saporta

Gui

Agrawal

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…In fact, DNNs are also applicable in small datasets. Makimoto et al (11) used the PTB ECG database consisting of 289 ECGs including 148 myocardial infarction (MI) cases to develop a CNN to recognize MI in ECG. The deep learning with a simple CNN for image analysis may achieve a comparable capability to physicians in recognizing MI on ECG.…”

Section: Discussionmentioning

confidence: 99%

“…With the development of deep neural network models (DNNs), artificial intelligence (AI) has made great progress and has been gradually applied to the diagnosis of echocardiography and ECG (10,11). Since DNNs can recognize patterns and learn useful features from raw input data without requiring extensive data preprocessing, feature engineering or handcrafted rules, and DNNs' performance tends to increase as the amount of training data increases, this approach is suitable for ECG analysis (12).…”

Section: Introductionmentioning

confidence: 99%

Detection of Left Atrial Enlargement Using a Convolutional Neural Network-Enabled Electrocardiogram

Jiang

Deng

Xue

et al. 2020

Front. Cardiovasc. Med.

View full text Add to dashboard Cite

Background: Left atrial enlargement (LAE) can independently predict the development of a variety of cardiovascular diseases.Objectives: This study sought to develop an artificial intelligence approach for the detection of LAE based on 12-lead electrocardiography (ECG).Methods: The study population came from an epidemiological survey of heart disease in Guangzhou. Elderly people (3,391) over 65 years old who had both 10-s 12 lead ECG and echocardiography were enrolled in this study. The left atrial (LA) anteroposterior diameter >40 mm on echocardiography was diagnosed as LAE, and the LA anteroposterior diameter was indexed by body surface area (BSA) to classify LAE into different degrees. A convolutional neural network (CNN) was trained and validated to detect LAE from normal ECGs. The performance of the model was evaluated by calculating the area under the curve (AUC), accuracy, sensitivity, specificity, and F1 score.Results: In this study, gender, obesity, hypertension, and valvular heart disease seemed to be related to left atrial enlargement. The AI-enabled ECG identified LAE with an AUC of 0.949 (95% CI: 0.911–0.987). The sensitivity, specificity, accuracy, precision, and F1 score were 84.0%, 92.0%, 88.0%, 91.3%, and 0.875, respectively. Physicians identified LAE with sensitivity, specificity, accuracy, precision, and F1 scores of 38.0%, 84.0%, 61.0%, 70.4%, and 0.494, respectively. In classifying LAE in different degrees, the AUCs of identifying normal, mild LAE, and moderate-severe LAE ECGs were 0.942 (95% CI: 0.903–0.981), 0.951 (95% CI: 0.917–0.987), and 0.998 (95% CI: 0.996–1.00), respectively. The sensitivity, specificity, accuracy, positive predictive value, and F1 scores of diagnosing mild LAE were 82.0%, 92.0%, 88.7%, 89.1%, and 0.854, while the sensitivity, specificity, accuracy, positive predictive value, and F1 scores of diagnosing moderate-severe LAE were 98.0%, 84.0%, 88.7%, 96.1%, and 0.969, respectively.Conclusions: An AI-enabled ECG acquired during sinus rhythm permits identification of individuals with a high likelihood of LAE. This model requires further refinement and external validation, but it may hold promise for LAE screening.

show abstract

“…objects in an image if the image contains multiple occurrences of the same classification (as it is the case in a 12-lead ECG) [13]. Last, Cohen-Shelly et al showed the feasibility of artificial intelligence based screening for aortic valve stenosis using standard 12-lead ECG [39].…”

Section: Plos Onementioning

confidence: 99%

Detection of myocardial ischemia by intracoronary ECG using convolutional neural networks

Bigler

Seiler

2021

PLoS ONE

View full text Add to dashboard Cite

Introduction The electrocardiogram (ECG) is a valuable tool for the diagnosis of myocardial ischemia as it presents distinctive ischemic patterns. Deep learning methods such as convolutional neural networks (CNN) are employed to extract data-derived features and to recognize natural patterns. Hence, CNN enable an unbiased view on well-known clinical phenomenon, e.g., myocardial ischemia. This study tested a novel, hypothesis-generating approach using pre-trained CNN to determine the optimal ischemic parameter as obtained from the highly susceptible intracoronary ECG (icECG). Method This was a retrospective observational study in 228 patients with chronic coronary syndrome. Each patient had participated in clinical trials with icECG recording and ST-segment shift measurement at the beginning (i.e., non-ischemic) and the end (i.e., ischemic) of a one-minute proximal coronary artery balloon occlusion establishing the reference. Using these data (893 icECGs in total), two pre-trained, open-access CNN (GoogLeNet/ResNet101) were trained to recognize ischemia. The best performing CNN during training were compared with the icECG ST-segment shift for diagnostic accuracy in the detection of artificially induced myocardial ischemia. Results Using coronary patency or occlusion as reference for absent or present myocardial ischemia, receiver-operating-characteristics (ROC)-analysis of manually obtained icECG ST-segment shift (mV) showed an area under the ROC-curve (AUC) of 0.903±0.043 (p<0.0001, sensitivity 80%, specificity 92% at a cut-off of 0.279mV). The best performing CNN showed an AUC of 0.924 (sensitivity 93%, specificity 92%). DeLong-Test of the ROC-curves showed no significant difference between the AUCs. The underlying morphology responsible for the network prediction differed between the trained networks but was focused on the ST-segment and the T-wave for myocardial ischemia detection. Conclusions When tested in an experimental setting with artificially induced coronary artery occlusion, quantitative icECG ST-segment shift and CNN using pathophysiologic prediction criteria detect myocardial ischemia with similarly high accuracy.

show abstract

Performance of a convolutional neural network derived from an ECG database in recognizing myocardial infarction

Cited by 68 publications

References 5 publications

Benchmarking saliency methods for chest X-ray interpretation

Benchmarking saliency methods for chest X-ray interpretation

Detection of Left Atrial Enlargement Using a Convolutional Neural Network-Enabled Electrocardiogram

Detection of myocardial ischemia by intracoronary ECG using convolutional neural networks

Contact Info

Product

Resources

About