Automatic Whole Heart Segmentation Using a Two-Stage U-Net Framework and an Adaptive Threshold Window

Tao Liu,Qingjun Wang,Xiaoying Huang,Shifeng Zhao,Yun Tian

doi:10.1109/access.2019.2923318

Abstract

Whole heart segmentation is an important medical imaging method used to enable clinical applications. However, automatic segmentation of the heart is still a challenging task due to the complexity and particularity of medical images, especially when the heart is segmented into substructures. In this study, we present a training strategy that relies on a two-stage U-Net framework and an adaptive threshold window to automatically segment a whole heart and its substructures. The two-stage U-Net framework consists of a region of interest (ROI) detection of the whole heart and accurate segmentation of the heart substructures. The adaptive threshold window method is used to remove the noisy parts of the data while preserving the anatomical relationships between local regions. Experiments were performed on a dataset from the MM-WHS Challenge 2017. The proposed approach resulted in a high segmentation accuracy with a 79.3% and 95.5% Dice similarity coefficient for the whole heart and ascending aorta segmentation, respectively, using limited GPU computing resources and small amounts of annotated data. The full implementation and configuration files in this paper are available at https://github.com/liut969/Whole-Heart-Segmentation.

Highlights

In recent years, increasing numbers of people are dying from cardiovascular diseases (CVDs)
In this study, we extended the U-Net framework, so that it retains the original advantages while obtaining refined heart segmentation results
To simplify the problem and reduce the complexity, we propose the two-stage U-Net framework, which consists of a whole heart segmentation and a heart substructures segmentation

Summary

Introduction

In recent years, increasing numbers of people are dying from cardiovascular diseases (CVDs). Noninvasive cardiac imaging assessments, such as computed tomography (CT) and magnetic resonance imaging (MRI), that provide a clear anatomy of the heart, have been used extensively in diagnoses of heart disorders. Trained physicians capable of analyzing medical data are not readily available, and the quantity of medical data requiring analysis has increased exponentially. As [1] illustrates, the growth rate of trained radiologists is half of the growth rate of medical images. Whole heart image segmentation is an essential step for a wide range of clinical applications, such as pathology localizations, and accurate ventricular measurements. Segmenting an entire heart from medical images is challenging due to the shape variations of the cardiac anatomy

Methods

Results

Conclusion