Chapter 5 - Generative adversarial network for cardiovascular imaging

Abstract

Analysis of cardiac function plays an important role in clinical cardiology for patient management, disease diagnosis, risk evaluation, and treatment decision. Delineation of right- and left-sided cavities, as well as major vessels, is an important step in clinical cardiology for cardiac disease diagnosis. Medical imaging provides a noninvasive diagnostic tool to study cardiac anatomy and detect pathological changes that occur in disease states such as dilated cardiomyopathy, hypertrophic cardiomyopathy, and right ventricular dysfunction. The accurate automation of the corresponding task can accelerate the diagnostic part and assist in therapy decisions. In this chapter, we overview recent machine learning approaches that have been applied for automatic cardiac diagnosis based on deep learning and generative adversarial networks and explain the recent application of generative adversarial networks (GANs) for classification, detection, segmentation, registration, image reconstruction, and synthetization mostly in cardiovascular imaging. In continuum, we review the public cardiac dataset with open challenges and conclude the chapter with GAN limitations and future research directions.