Deep Learning Based Cardiac Phase Detection Using Echocardiography Imaging

Moomal Farhad, Mohammad M. Masud, Azam Beg

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Echocardiography is a widely used, affordable, and high-throughput test to evaluate heart conditions. There are two cardiac phases known as end-systolic (ES) and end-diastolic (ED), which are observed by cardiologists and technicians during echocardiography. These phases are used to perform critical calculations during echocardiography, such as calculating heart chamber size and the calculation of ejection fraction. Typically, ED and ES phase detection is performed manually by a technician or cardiologist, which makes echocardiography interpretation time-consuming and error-prone. Therefore, it is crucial to develop an automated and efficient technique to detect the cardiac phases and minimize diagnostic errors. In this paper, we propose a deep learning model, namely DeepPhase, for accurate and automated interpretation of the ES and ED phases to assist cardiology personnel. Our proposed convolutional neural network (CNN) is designed to learn from echocardiography images and identify the cardiac phase from the given image without segmentation of left ventrical or the use of electrocardiograms. We have performed an extensive evaluation on two-real world echocardiography image datasets, namely, the benchmark Cardiac Acquisitions for Multistructure Ultrasound Segmentation (CAMUS) dataset and a new dataset (referred to as CardiacPhase) that we collected from a cardiac hospital. The proposed model outperformed relevant state-of-the-art techniques by achieving 0.98 and 0.92 area under the curve (AUC) on the CAMUS dataset, and the CardiacPhase dataset, respectively. We have proved the generalizability of our proposed work by training and testing with different cardiac views.

Original languageEnglish
Title of host publicationAdvanced Data Mining and Applications - 17th International Conference, ADMA 2021, Proceedings
EditorsBohan Li, Lin Yue, Jing Jiang, Weitong Chen, Xue Li, Guodong Long, Fei Fang, Han Yu
PublisherSpringer Science and Business Media Deutschland GmbH
Pages3-17
Number of pages15
ISBN (Print)9783030954048
DOIs
Publication statusPublished - 2022
Event17th International Conference on Advanced Data Mining and Applications, ADMA 2021 - Sydney, Australia
Duration: Feb 2 2022Feb 4 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13087 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference17th International Conference on Advanced Data Mining and Applications, ADMA 2021
Country/TerritoryAustralia
CitySydney
Period2/2/222/4/22

Keywords

  • Artificial intelligence
  • Convolutional neural network
  • Deep learning
  • Echocardiography
  • Medical imaging

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Deep Learning Based Cardiac Phase Detection Using Echocardiography Imaging'. Together they form a unique fingerprint.

Cite this