Investigation on Performance Analysis of SVM ELM KNN Classifiers for Detection of Abnormal Region from Electrical Impedance Tomography Images

Abstract

Electrical Impedance Tomography (EIT) is the most potential, noninvasive,radiation-free, bedsides imaging technique which is broadly used inthe medical field. The electrical conductivity is reconstructed in twodimensional or three dimensional (2D or 3D) based on potentialmeasurements from the surface of the object. Real-time EIT is used toperform functional imaging of regional lung volumes and regional ventilationdistribution. Also used for detect the changes in pulmonary ventilation andperfusion distributions. In EIT, the conductivity distribution is discontinuousdue to tissue boundaries which can be conserved by using Tikhonov priormodel in the EIT image reconstruction.This research work comprises of four stages, namely reconstructingthe EIT images based on the potential measurements from the surface of theobject, extracting the desire features from images, selecting the appropriatefeatures by using two different feature selection techniques and finallyclassification of images using three different classifiers. One step Gauss-Newton (GN) algorithm and Back projection algorithm are implemented forthe EIT image reconstruction along with prior model. newline