Biomedical Image Segmentation

Advances and Trends

Ayman El-Baz, Xiaoyi Jiang, Jasjit S. Suri

• Publisher: CRC Press
• Year: 2016
• ISBN:
  9781315355047
• Edition: 1
• Topic: Medicine
• Subtopic: Biotechnology in Medicine

As one of the most important tasks in biomedical imaging, image segmentation provides the foundation for quantitative reasoning and diagnostic techniques. A large variety of different imaging techniques, each with its own physical principle and characteristics (e.g., noise modeling), often requires modality-specific algorithmic treatment. In recent years, substantial progress has been made to biomedical image segmentation. Biomedical image segmentation is characterized by several specific factors. This book presents an overview of the advanced segmentation algorithms and their applications.

chapter one

Deformable model-based methods for image segmentation

Marwa Ismail, Ahmed ElTanboly, Magdi El-Azab, Ayman El-Baz, and Robert Keynton

Contents

Abstract

1.1 Introduction

1.2 Parametric deformable models

1.2.1 Solution approaches to the parametric deformable models

1.2.2 Guiding forces

1.2.2.1 Edge-based external force models

1.2.2.2 Region-based external force models

1.2.3 Summary on parametric deformable models

1.3 Geometric deformable models

1.3.1 Numerical approaches solving the level set equation

1.3.1.1 Upwind differencing

1.3.1.2 Hamilton–Jacobi

1.3.1.3 Hamilton–Jacobi WENO

1.3.1.4 TVD Runge–Kutta

1.3.2 Hamilton–Jacobi equations

1.3.2.1 Burgers’ equation

1.3.2.2 The Lax–Friedrichs scheme

1.3.2.3 The Roe–Fix scheme

1.3.2.4 Godunov’s scheme

1.3.3 Optimization schemes

1.3.3.1 Euler–Lagrange equation

1.3.4 Guiding forces

1.3.4.1 Edge-based forces

1.3.4.2 Region-based forces

1.3.4.3 Summary on geometric deformable models

1.4 Applications in medical imaging

1.5 Conclusion and future trends

References

Abstract

Medical imaging is a rapidly evolving area, where there is a strong need to understand scalar or vector-valued images. The outcome of such processing can have strong diagnostic implications and can be used as an additional tool to detect and treat different diseases in a timely and proper fashion. This chapter targets one of the most popular processes in medical imaging, which is segmentation. Segmentation methods have different classifications; they can be edge-based or region-based, and the process can be manual or fully automated. This chapter focuses on deformable-based methods, specifically, region-based, for segmentation that are mathematically represented in partial differential equations. In particular, the mathematical background for such models is given in terms of parametric- or geometric-based. The pros and cons of each type are also discussed, along with the future trends. Finally, the chapter summarizes the recent applications of deformable model-based segmentation in medical imaging. It briefly surveys recent work in the segmentation of different tissue types, such as the kidney, prostate, and lung.

1.1 Introduction

In computer vision, image segmentation is the process of partitioning an image into multiple segments. In other words, it is the process of labeling every pixel in an image such that pixels with the same label share the same characteristics, such as texture, brightness, or gray level [1]. Segmentation of medical images, in particular, has assumed immense importance as a noninvasive tool that provides physicians with a reliable way of diagnosis of the abnormalities found in different organs.

In the field of medical imaging, accurate segmentation of structures is crucial for detecting lesions and abnormalities. It has recently provided great advances to clinicians in assessing abnormalities through computer-aided diagnostic (CAD) systems. Segmentation, however, is highly challenging due to many factors, such as the low contrast between different tissue types, which makes it difficult to segment the desired object manually, and the motion artifacts associated with the scans which add...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Preface
• Editors
• Contributors
• Chapter 1 Deformable model-based methods for image segmentation
• Chapter 2 Domain knowledge for level set segmentation in medical imaging: A review
• Chapter 3 Robust image segmentation with a parametric deformable model using learned shape priors
• Chapter 4 A 3D active shape model for left ventricle segmentation in MRI
• Chapter 5 Model-based segmentation algorithms for myocardial magnetic resonance imaging sequences
• Chapter 6 Incorporating shape variability in implicit template deformation for image segmentation
• Chapter 7 Exudate detection in fundus images using active contour methods and region-wise classification
• Chapter 8 Preprocessing local features and fuzzy logic-based image segmentation
• Chapter 9 Model-based curvilinear network extraction toward quantitative microscopy
• Chapter 10 Level set-based cell segmentation using convex energy functionals
• Chapter 11 Histogram-based level set methods for medical image segmentation
• Chapter 12 An appearance-guided deformable model for 4D kidney segmentation using diffusion MRI
• Chapter 13 Prostate segmentation using deformable model-based methods: A review
• Chapter 14 A novel NMF-based CAD system for early diagnosis of prostate cancer by using 4D diffusion-weighted magnetic resonance images (DW-MRI)
• Chapter 15 Distance regularized level sets for segmentation of the left and right ventricles
• Chapter 16 Salient object segmentation with a shape-constrained level set
• Chapter 17 Tracking and segmentation of the endocardium of the left ventricle in a 2D ultrasound using deep learning architectures and Monte Carlo sampling
• Chapter 18 A shortest path approach to interactive medical image segmentation
• Chapter 19 Local statistical models for ultrasound image segmentation
• Chapter 20 Image segmentation with physical noise models
• Chapter 21 A fast lung segmentation approach
• Chapter 22 Fully automatic segmentation of hip CT images via landmark detection-based atlas selection and optimal surface detection
• Index