An Identification of Efficient Vessel Feature for Endoscopic Analysis

Abstract

An indispensable chore in computer–aided Minimally Invasive Surgery (MIS) is the typical feature detection. In some of the distinctive conditions in MIS imaging such as texture homogeneous areas specular reflections uses general feature point detectors for drawing out the feature points which are less distinct and repeatable in MIS images. We examine that a new set of image feature can be taken out from the copious blood vessels which are obtainable from the tissue surfaces. Two types of blood vessel feature are put forward for endoscopic image analysis in this proposed paper: branching segments and branching points. Ridgeness Based Circle Test (RBCT) and Ridgeness Based Branching Segment Detection (RBSD) are the two new proposed methods used to obtain the branching points and branching segments from the blood vessels. To estimate the working effectiveness of the proposed method extensive in vivo experiments are carried out and the state-of-the-art method is used to evaluate them. In MIS images, the blood vessels from the tissue surfaces can yield enormous large number of points, which is proved from the numerical results. The points produced by the blood vessels are more robust and repeatable when compared with other types of feature points. This type of feature detection can be used as a new tool for the analysis of MIS images in endoscopy. Thus, the proposed method for detecting feature point is very efficient and robust in nature.