Deep Learning for Segmentation Using an Open Large-Scale Dataset in 2D Echocardiography (2019-02-22T00:00:00.000000Z)

TL;DR

Results show that encoder-decoder-based architectures outperform state-of-the-art non-deep learning methods and faithfully reproduce the expert analysis for the end-diastolic and end-systolic left ventricular volumes, which open the door for accurate and fully-automatic analysis of 2D echocardiographic images.

Abstract

Delineation of the cardiac structures from 2D echocardiographic images is a common clinical task to establish a diagnosis. Over the past decades, the automation of this task has been the subject of intense research. In this paper, we evaluate how far the state-of-the-art encoder-decoder deep convolutional neural network methods can go at assessing 2D echocardiographic images, i.e., segmenting cardiac structures and estimating clinical indices, on a dataset, especially, designed to answer this objective. We, therefore, introduce the cardiac acquisitions for multi-structure ultrasound segmentation dataset, the largest publicly-available and fully-annotated dataset for the purpose of echocardiographic assessment. The dataset contains two and four-chamber acquisitions from 500 patients with reference measurements from one cardiologist on the full dataset and from three cardiologists on a fold of 50 patients. Results show that encoder-decoder-based architectures outperform state-of-the-art non-deep learning methods and faithfully reproduce the expert analysis for the end-diastolic and end-systolic left ventricular volumes, with a mean correlation of 0.95 and an absolute mean error of 9.5 ml. Concerning the ejection fraction of the left ventricle, results are more contrasted with a mean correlation coefficient of 0.80 and an absolute mean error of 5.6%. Although these results are below the inter-observer scores, they remain slightly worse than the intra-observer’s ones. Based on this observation, areas for improvement are defined, which open the door for accurate and fully-automatic analysis of 2D echocardiographic images.

References39 items

UNet++: A Nested U-Net Architecture for Medical Image Segmentation

Deep Learning Techniques for Automatic MRI Cardiac Multi-Structures Segmentation and Diagnosis: Is the Problem Solved?

A fully automatic and multi-structural segmentation of the left ventricle and the myocardium on highly heterogeneous 2D echocardiographic data

2D left ventricle segmentation using deep learning

Fast and Fully Automatic Left Ventricular Segmentation and Tracking in Echocardiography Using Shape-Based B-Spline Explicit Active Surfaces

Anatomically Constrained Neural Networks (ACNNs): Application to Cardiac Image Enhancement and Segmentation

Left Ventricular Myocardial Segmentation in 3-D Ultrasound Recordings: Effect of Different Endocardial and Epicardial Coupling Strategies.

A survey on deep learning in medical image analysis

3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation

Standardized Evaluation System for Left Ventricular Segmentation Algorithms in 3D Echocardiography

Stacked Hourglass Networks for Human Pose Estimation

Quality Control and Reproducibility in M‐Mode, Two‐Dimensional, and Speckle Tracking Echocardiography Acquisition and Analysis: The CARDIA Study, Year 25 Examination Experience

U-Net: Convolutional Networks for Biomedical Image Segmentation

Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging.

Segmentation of Multi-Center 3D Left Ventricular Echocardiograms by Active Appearance Models

Left Ventricle Segmentation in Cardiac Ultrasound Using Hough-Forests With Implicit Shape and Appearance Priors

Endocardial 3D Ultrasound Segmentation using Autocontext Random Forests

Model-based left ventricle segmentation in 3D ultrasound using phase image

Endocardial Segmentation using Structured Random Forests in 3D Echocardiography

Learning Shape Representations for Multi-Atlas Endocardium Segmentation in 3D Echo Images

Real-time Tracking of the Left Ventricle in 3D Ultrasound Using Kalman Filter and Mean Value Coordinates

Fast Tracking of the Left Ventricle Using Global Anatomical Affine Optical Flow and Local Recursive Block Matching

Automatized Evaluation of the Left Ventricular Ejection Fraction from Echocardiographic Images Using Graph Cut

Fast Edge Detection Using Structured Forests

The Segmentation of the Left Ventricle of the Heart From Ultrasound Data Using Deep Learning Architectures and Derivative-Based Search Methods

Automated border detection in three-dimensional echocardiography: principles and promises.

Basic Skills Training to Disabled and Adult Learners Through an Accessible e-Learning Platform

Ultrasound image segmentation: a survey

Assessment of Left Ventricular Ejection Fraction and Volumes by Real-time, Two-dimensional Echocardiography: A Comparison of Cineangiographic and Radionuclide Techniques

Annotation

Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging.

Fast and fully automatic 3-d echocardiographic segmentation using B-spline explicit active surfaces: feasibility study and validation in a clinical setting.

B-Spline Explicit Active Surfaces: An Efficient Framework for Real-Time 3-D Region-Based Segmentation

Segmentation results obtained by the SRF method on Patient 27

How many patients are needed to train a CNN to get highly accurate results in 2D echocardiographic image segmentation ?

What improvement can be brought by sophisticated architectures compared to simpler CNN models for 2D echocardiographic segmentation ?

How accurate can the volumes and ejection fraction be estimated from the segmentation of CNNs compared to the inter/intra-expert variability ?

dm-epi = 2.3 , d H -endo = 11.1 , d H -epi = 14.2 mm. (d) 4CH-ES: dm-endo = 3.2 , dm-epi = 1.7 , d H -endo = 6

2CH-ED : dm-endo = 1.3 , dm-epi = 1.7 , d H -endo = 3.7 , d H -epi = 4.6 mm. (b) 2CH-ES : dm-endo = 1.8 , dm-epi