Classification of Cardiomyopathies from MR Cine Images Using Convolutional Neural Network with Transfer Learning.

Philippe Germain,Soraya El Ghannudi,Catherine Roy,Thomas Hellmut Schindler,Armine Vardazaryan,Aissam Labani,Nicolas Padoy

doi:10.3390/diagnostics11091554

Abstract

The automatic classification of various types of cardiomyopathies is desirable but has never been performed using a convolutional neural network (CNN). The purpose of this study was to evaluate currently available CNN models to classify cine magnetic resonance (cine-MR) images of cardiomyopathies. Method: Diastolic and systolic frames of 1200 cine-MR sequences of three categories of subjects (395 normal, 411 hypertrophic cardiomyopathy, and 394 dilated cardiomyopathy) were selected, preprocessed, and labeled. Pretrained, fine-tuned deep learning models (VGG) were used for image classification (sixfold cross-validation and double split testing with hold-out data). The heat activation map algorithm (Grad-CAM) was applied to reveal salient pixel areas leading to the classification. Results: The diastolic–systolic dual-input concatenated VGG model cross-validation accuracy was 0.982 ± 0.009. Summed confusion matrices showed that, for the 1200 inputs, the VGG model led to 22 errors. The classification of a 227-input validation group, carried out by an experienced radiologist and cardiologist, led to a similar number of discrepancies. The image preparation process led to 5% accuracy improvement as compared to nonprepared images. Grad-CAM heat activation maps showed that most misclassifications occurred when extracardiac location caught the attention of the network. Conclusions: CNN networks are very well suited and are 98% accurate for the classification of cardiomyopathies, regardless of the imaging plane, when both diastolic and systolic frames are incorporated. Misclassification is in the same range as inter-observer discrepancies in experienced human readers.

Highlights

Machine learning, computer vision, and deep learning are areas that have grown explosively over the last 10 years
Study selection was made on the basis of visual diagnosis, by a practitioner with 25 years of experience, reviewing and retaining cine magnetic resonance (cine-MR) images with the characteristic appearance of normal, hypertrophic, or hypokinetic dilated cardiomyopathy
The current study provides a unique framework of the concept that applying convolutional neural network (CNN) in cine-MR may contribute to optimizing the identification and characterization of different cardiomyopathy disease entities, because CNN features likely carry important prognostic and therapeutic information

Summary

Introduction

Computer vision, and deep learning are areas that have grown explosively over the last 10 years. Most of the work concerns supervised learning with a convolutional neural network (CNN) [1]. An exhaustive review of image-based cardiac diagnosis with machine learning was recently published [3]. Most deep learning (DL) studies performed on computed tomography (CT) data were devoted to calcium scoring, coronary artery disease prognosis, and functional coronary stenosis detection using plaque or fractional flow reserve quantification [4]. Cine-MR studies focused on left-ventricular (LV) segmentation allow the automatic quantification of cardiac volumes and function and are aimed at replacing traditional tedious manual contouring by fully conventional networks with encoder–decoder structure (e.g., U-Net) [5,6]

Objectives

Methods

Discussion

Conclusion