Fully automated 2D and 3D convolutional neural networks pipeline for video segmentation and myocardial infarction detection in echocardiography

Abstract

Myocardial infarction (MI) is a life-threatening disorder that occurs due to a prolonged limitation of blood supply to the heart muscles, and which requires an immediate diagnosis to prevent death. To detect MI, cardiologists utilize in particular echocardiography, which is a non-invasive cardiac imaging that generates real-time visualization of the heart chambers and the motion of the heart walls. These videos enable cardiologists to identify almost immediately regional wall motion abnormalities (RWMA) of the left ventricle (LV) chamber, which are highly correlated with MI. However, data acquisition is usually performed during emergency which results in poor-quality and noisy data that can affect the accuracy of the diagnosis. To address the identified problems, we propose in this paper an innovative, real-time and fully automated model based on convolutional neural networks (CNN) to early detect MI in a patient’s echocardiography. Our model is a pipeline consisting of a 2D CNN that performs data preprocessing by segmenting the LV chamber from the apical four-chamber (A4C) view, followed by a 3D CNN that performs a binary classification to detect MI. The pipeline was trained and tested on the HMC-QU dataset consisting of 162 echocardiography. The 2D CNN achieved 97.18% accuracy on data segmentation, and the 3D CNN achieved 90.9% accuracy, 100% precision, 95% recall, and 97.2% F1 score. Our detection results outperformed existing state-of-the-art models that were tested on the HMC-QU dataset for MI detection. This work demonstrates that developing a fully automated system for LV segmentation and MI detection is efficient and propitious.