Deep Echocardiography: A First Step toward Automatic Cardiac Disease Diagnosis Using Machine Learning

Abstract

Echocardiography, the use of ultrasound waves to investigate the action of the heart, is the primary physiological test for cardiovascular disease diagnoses. Firstly, this article discusses the common diagnostic procedures of echocardiography, meanwhile emphasizes and elaborates that view recognition is the first essential step, and then explicates issues concerning manual view identification based on two main aspects i.e. echo images’ properties and sonographers’ task difficulties. Secondly, the published articles within the past five years relating to how artificial intelligence is applied to the echocardiographic view recognition are selected, compared and summarized. It is found that compared with previous machine learning algorithm, deep learning has the ability to boost the analysis and interpretation of ultrasonic images into a new level. Finally, the challenges and limitations existed during the development of AI in health care are highlighted and discussed.