Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Automated Localization of Blood Vessels in Retinal Images

Vahid Mohammadi Safarzadeh

TL;DR

This work tackles automatic localization of retinal blood vessels in the presence of bright and dark lesions. It proposes two automated methods that share a final multi-scale line-based detection stage, differing in their initial lesion suppression: Method 1 relies on K-means clustering on the green channel, while Method 2 leverages a structurally differentiable plane obtained via total-variation regularization. Both methods are evaluated on DRIVE and STARE datasets, including pathological images, and achieve performance that is competitive with state-of-the-art approaches in terms of accuracy and ROC AUC, with reasonable processing times. The study demonstrates robust vessel localization by explicitly addressing lesion-induced artifacts and employing a multi-scale line detector, supporting potential clinical screening and pre-diagnostic workflows for retinal diseases.

Abstract

Vessel structure is one of the most important parts of the retina which physicians can detect many diseases by analysing its features. Localization of blood vessels in retina images is an important process in medical image analysis. This process is also more challenging with the presence of bright and dark lesions. In this thesis, two automated vessel localization methods to handle both healthy and unhealthy (pathological) retina images are analyzed. Each method consists of two major steps and the second step is the same in the two methods. In the first step, an algorithm is used to decrease the effect of bright lesions. In Method 1, this algorithm is based on K- Means segmentation, and in Method 2, it is based on a regularization procedure. In the second step of both methods, a multi-scale line operator is used to localize the line-shaped vascular structures and ignore the dark lesions which are generally assumed to have irregular patterns. After the introduction of the methods, a detailed quantitative and qualitative comparison of the methods with one another as well as the state-of-the-art solutions in the literature based on the segmentation results on the images of the two publicly available datasets, DRIVE and STARE, is reported. The results demonstrate that the methods are highly comparable with other solutions.

Automated Localization of Blood Vessels in Retinal Images

TL;DR

This work tackles automatic localization of retinal blood vessels in the presence of bright and dark lesions. It proposes two automated methods that share a final multi-scale line-based detection stage, differing in their initial lesion suppression: Method 1 relies on K-means clustering on the green channel, while Method 2 leverages a structurally differentiable plane obtained via total-variation regularization. Both methods are evaluated on DRIVE and STARE datasets, including pathological images, and achieve performance that is competitive with state-of-the-art approaches in terms of accuracy and ROC AUC, with reasonable processing times. The study demonstrates robust vessel localization by explicitly addressing lesion-induced artifacts and employing a multi-scale line detector, supporting potential clinical screening and pre-diagnostic workflows for retinal diseases.

Abstract

Vessel structure is one of the most important parts of the retina which physicians can detect many diseases by analysing its features. Localization of blood vessels in retina images is an important process in medical image analysis. This process is also more challenging with the presence of bright and dark lesions. In this thesis, two automated vessel localization methods to handle both healthy and unhealthy (pathological) retina images are analyzed. Each method consists of two major steps and the second step is the same in the two methods. In the first step, an algorithm is used to decrease the effect of bright lesions. In Method 1, this algorithm is based on K- Means segmentation, and in Method 2, it is based on a regularization procedure. In the second step of both methods, a multi-scale line operator is used to localize the line-shaped vascular structures and ignore the dark lesions which are generally assumed to have irregular patterns. After the introduction of the methods, a detailed quantitative and qualitative comparison of the methods with one another as well as the state-of-the-art solutions in the literature based on the segmentation results on the images of the two publicly available datasets, DRIVE and STARE, is reported. The results demonstrate that the methods are highly comparable with other solutions.
Paper Structure (44 sections, 29 equations, 32 figures, 8 tables)

This paper contains 44 sections, 29 equations, 32 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Review
  3. Challenges and Incentives
  4. A Brief Overview of the Methods
  5. Retina Image
  6. Thesis Structure
  7. Literature Review
  8. Overview of Previous Works
  9. Theoretical Foundations
  10. Retina Image Datasets
  11. Binary Mask
  12. Green Channel of Retina Colour Image
  13. Pre-processing
  14. Transforming to Weber Space
  15. Expanding the Boundaries of the Field of View
  16. ...and 29 more sections

Figures (32)

  • Figure 1: The presented methods in one view. a) Method 1. b) Method 2
  • Figure 2: Retina Image. a) Retina image and its main parts (image number 16 of DRIVE dataset). b) A sample of retina having anomaly in Fovea-Macula area (image number 1 from STARE dataset).
  • Figure 3: A retina Image from DRIVE dataset (a), and its corresponding labelled image (b.)
  • Figure 4: A retina Image (a) and its corresponding mask (b.)
  • Figure 5: An original retina colour image (a), the red channel (b), green channel (c), and the blue channel (d.)
  • ...and 27 more figures