Registration Based Retrieval using Texture Measures

Abstract

The aim of the study presented in this manuscript was to develop and analyze registration based retrieval of medical image using texture measures. Three main methods are implemented in this work. The first method includes Affine transformation, Demons and Affine with B-spline. The second method implemented is medical image retrieval system using content based medical image retrieval. GLCM, LBP and GLCM with LBP are used for texture based retrieval. Shape based retrieval is processed using canny edge with the Otsu method. From registration based retrieval, Affine with B-Spline performs well and produces best result by increasing the retrieval rate and the next better performances are given by Demons and Affine registration. The results showed that the best results for registration based retrieval are given by Affine with B-Spline registration based retrieval using GLCM+LBP with (100/50). Based on more relevant retrieved images, Brain (100/50) and Knee (100/50) observed to have more relevant retrieved images.