Soft Computing Based Medical Image Analysis

Nilanjan Dey, Amira S. Ashour, Fuquian Shi, Valentina Emilia Balas, Valentina E. Balas

• Publisher: Academic Press
• Year: 2018
• Print ISBN: 9780128130872
• eBook ISBN: 9780128131749
• Topic: Technology & Engineering
• Subtopic: Biotechnology
• Index: Technology & Engineering

Soft Computing Based Medical Image Analysis presents the foremost techniques of soft computing in medical image analysis and processing. It includes image enhancement, segmentation, classification-based soft computing, and their application in diagnostic imaging, as well as an extensive background for the development of intelligent systems based on soft computing used in medical image analysis and processing. The book introduces the theory and concepts of digital image analysis and processing based on soft computing with real-world medical imaging applications. Comparative studies for soft computing based medical imaging techniques and traditional approaches in medicine are addressed, providing flexible and sophisticated application-oriented solutions. - Covers numerous soft computing approaches, including fuzzy logic, neural networks, evolutionary computing, rough sets and Swarm intelligence - Presents transverse research in soft computing formation from various engineering and industrial sectors in the medical domain - Highlights challenges and the future scope for soft computing based medical analysis and processing techniques

Section D

Machine Learning in Medical Image Segmentation and Classification

Chapter 10

State-of-the-Art of Level-Set Methods in Segmentation and Registration of Spectral Domain Optical Coherence Tomographic Retinal Images

Natarajan Padmasini^⁎; Rengasamy Umamaheswari^†; Mohamed Yacin Sikkandar^{‡ ⁎} Rajalakshmi Engineering College, Chennai, India
^† Velammal Engineering College, Chennai, India
^‡ CAMS, Majmaah University, Riyadh, Saudi Arabia

Abstract

In the quantitative assessment of Diabetic maculopathy from Spectral Domain Optical Coherence Tomography (SDOCT) images, analysis of intraretinal fluid filled regions plays a vital role because of its comparative superiority in providing tissue-level anatomical information. The detailed study on efficacy and performance of soft computing techniques-based automatic detection and diagnosis for SDOCT retinal images is still in the preliminary stage. Although some automatic algorithms have been proposed to segment retinal layers in recent times, full accuracy in edge detection continues to be a challenging problem. Some researchers have developed different versions of automatic algorithms for segmenting intraretinal fluid based on region-based level-set method and the retinal layers by dual gradient method. This particular level-set implementation is carried out using a fast front propagation algorithm. A valid search region is then defined to identify layer boundaries. The features of the segmented region are analyzed volumetrically and based on these temporal characteristics, the severity of the disease can then be estimated. Some operational algorithms have also been developed for registration of Peripapillary OCT and fundus image for the identification of Neovascularization in the early stage of diabetic proliferative retinopathy.

This chapter offers the reader a comprehensive review of the soft computing techniques applied to SDOCT retinal image analysis, particularly for image segmentation and registration techniques.

Keywords

Retinal layers; Retinal disease; Segmentation; Registration; Early diagnosis

Acknowledgments

The authors wish to express their sincere thanks to the University Grants Commission (UGC), Government of India, and acknowledge their assistance of providing grants to carry out this project (File No. MRP-6119/15(SERO/UGC)). The authors would also like to thank Dr. Manavi D Sindal, Senior consultant, Vitreo-Retina services, and her team at Aravind Eye Hospital, Pondicherry, India, for their guidance and generous support in providing the SDOCT data for this work. The authors convey their gratitude to the physicians of DRR Hospital, Chennai, India, for readily providing fundus and OCT data.

1 Introduction

In digital image-processing work, image segmentation is the process of partitioning a digital image into multiple segments in terms of pixels. The prime objective of segmentation is to simplify and/or change the representation of an image into something that is more meaningful and easier to analyze. Such segmentation of anatomical and pathological structures in ophthalmic images is crucial for the diagnosis and study of ocular diseases. The most difficult part of optical coherence tomography (OCT) image analysis is the automated localization and delineation of structures of interest in the retina. The success of OCT in the investigation and treatment of retinal diseases is perhaps best illustrated by the progress in automated analysis. The segmentation part of the retinal structure is a challenging task that faces major problems. OCT images suffer from intrinsic speckle noise, which decreases the image quality and complicates the precise identification of boundaries of the various layers of the retina. Moreover, since the intensity pattern in OCT images results from absorption and scattering of light in the retinal tissue, intensity parameters of a homogeneous area decrease with increasing imaging depth deterministically. This complicates segmentation algorithms, which are commonly based on the basic assumption that intensity variations of homogeneous regions are only due to noise and not intrinsic to the imaging modality. Further, motion artifacts and suboptimal ...

Table of contents

• Cover image
• Title page
• Table of Contents
• Copyright
• Contributors
• Preface
• Acknowledgments
• Section A: Medical Image Analysis and Processing
• Section B: Medical Image Enhancement
• Section C: Detection and Prediction in Medical Imaging
• Section D: Machine Learning in Medical Image Segmentation and Classification
• Index