Tree-based Ridge Extractor for image analysis

Introduction

Accurate and reliable analyses of medical images have become a key step to determine the presence or severity of diseases, directly impacting the clinical care for patients.

Importance of Biomedical Image Analysis

Among the numerous and various biomedical image analyses, the identification of blood vessels and the associated vascular trees is one of the most central topics for diagnosis of the coronary, lungs, liver, or retina, which require detection methods to assist physicians.

Limitations of Current Methods

Many such methods show intrinsic limitations due to:

• The required pre-processing of the images
• The approach used, e.g., limited training databases, etc.

Complex Patterns in Nature

Complex filamentary patterns, such as the vascular trees, are common and widely observed in various fields including geology and astrophysics. In astrophysics, elongated structures made of the alignment of galaxies, called filaments of the cosmic web (CW), reveal the nature of matter in the Universe and the way it assembles over the ~14 billion years’ history of the cosmos.

Relevance to Astrophysics

For these reasons, detecting CW filaments and studying them is an increasingly important topic in astrophysics and is one of the aims of the ERC-Advanced project ByoPiC (Baryon Picture of the Cosmos).

Project Overview

This proposal is a direct offspring of ByoPiC. It is the first step of a multi-step technology transfer strategy which aims at transferring an innovative algorithm to detect filaments in the distribution of galaxies to the analysis of biomedical images. This lays the ground for applications in scores of other domains, including:

1. Satellite imagery
2. Geophysical tomographic data
3. Engineering
4. And more.