Artificial Intelligence for Computational Modeling of the Heart

Tommaso Mansi,Tiziano Passerini,Dorin Comaniciu

• Editorial: Academic Press
• Año: 2019
• ISBN:
  9780128168950
• Categoría: Biowissenschaften
• Categoría: Biotechnologie

Artificial Intelligence for Computational Modeling of the Heart presents recent research developments towards streamlined and automatic estimation of the digital twin of a patient's heart by combining computational modeling of heart physiology and artificial intelligence. The book first introduces the major aspects of multi-scale modeling of the heart, along with the compromises needed to achieve subject-specific simulations. Reader will then learn how AI technologies can unlock robust estimations of cardiac anatomy, obtain meta-models for real-time biophysical computations, and estimate model parameters from routine clinical data. Concepts are all illustrated through concrete clinical applications.

• Presents recent advances in computational modeling of heart function and artificial intelligence technologies for subject-specific applications
• Discusses AI-based technologies for robust anatomical modeling from medical images, data-driven reduction of multi-scale cardiac models, and estimations of physiological parameters from clinical data
• Illustrates the technology through concrete clinical applications and discusses potential impacts and next steps needed for clinical translation

Part 1

Modeling of the beating heart: approaches and implementation

Outline

1. Multi-scale models of the heart for patient-specific simulations

2. Implementation of a patient-specific cardiac model

1

Multi-scale models of the heart for patient-specific simulations

Viorel Mihalef; Tiziano Passerini; Tommaso Mansi Siemens Healthineers, Princeton, NJ, United States

Abstract

This chapter presents a general overview on computational models of the heart. It focuses on mathematical modeling of the various physiological aspects involved in cardiac function, namely anatomy, electrophysiology, biomechanics, fluid dynamics and fluid structure interaction. Their implementation is described in the next chapter. The goal of this chapter is to give a brief introduction of each physiological system, and examples of mathematical models that have been developed so far by the community, along with their advantages and potential limitations in view of patient-specific simulations.

Keywords

Computational Modeling; Anatomy; Electrophysiology; Biomechanics; Hemodynamics

Modeling heart function consists in translating the main mechanisms that govern heart function into mathematical laws and computational algorithms. Multi-scale models describe the interactions at different time scales (from nanoseconds to minutes to years), spatial scales (from nanometers to centimeters), and functional scales (from molecular pathways to circulatory system). Such models can therefore pave the way to predictive medicine by computing advanced measurements, planning therapies in-silico before the intervention, evaluating the effects of molecular changes on the global cardiac function. Scientific and technological development of such models are enabling comprehensive and detailed understanding of heart physiology. At the same time, clinical translation requires a clear clinical focus on selecting the models, controlling the underlying assumptions and simplifications, and ensuring their usefulness to address real-world clinical problems while being integrated into the day-to-day workflow. This chapter introduces the typical components of a multi-scale heart model that includes personalized anatomy, electrophysiology, biomechanics, valves, hemodynamics, and common approaches to parameter estimation from routinely available clinical data.

1.1 Models of cardiac anatomy

Let us first review the major anatomical elements to consider for cardiac modeling. The reader is invited to refer to specialized literature such as [42] for a more detailed description of cardiac anatomy and function. The very first step in modeling heart function consists in defining a realistic model of its anatomy, which includes shape, tissue types (also called substrate) and tissue micro-architecture. At a high level, the heart can be described as a system of four blood-filled chambers separated by valves (Fig. 1.1). Functionally and anatomically the heart can be divided into the atria and ventricles, as well as a left (systemic) and right (pulmonary) side. The oxygen-poor blood enters first the right atrium, then the right ventricle through the tricuspid valve. The right ventricle then ejects the blood into the lungs through the pulmonary valve. The blood gets oxygenated by the lungs and then goes back to the heart in the left atrium chamber, from where it reaches the left ventricle through the mitral valve. Finally, the left ventricle ejects the blood through the aortic valve into the circulatory system. The muscle, called myocardium, of the atria is much thinner than ventricular myocardium as they only need to transfer the blood to the ventricles. The ventricles have to provide a more significant amount of work to push blood in the peripheral circulation. The heart is enclosed into a relatively stiff bag called pericardium, which is fixed in the thoracic cavity. A thin layer of pericardial fluid bet...

Índice

• Cover image
• Title page
• Table of Contents
• Copyright
• List of figures
• List of contributors
• Foreword
• Preface
• List of abbreviations
• Part 1: Modeling of the beating heart: approaches and implementation
• 1: Multi-scale models of the heart for patient-specific simulations
• 2: Implementation of a patient-specific cardiac model
• Part 2: Artificial intelligence methods for cardiac modeling
• 3: Learning cardiac anatomy
• 4: Data-driven reduction of cardiac models
• 5: Machine learning methods for robust parameter estimation
• 6: Additional clinical applications
• Bibliography
• Index