Tb Diagnosis System using Genetic Particle Swarm Optimization Based Neural Network Classifier
P. Prasanna Kumari1, B. Prabhakara Rao2

1Mrs. P. Prasanna Kumari, Pursing PhD in JNTUK, Kakinada, Andhra Pradesh, India.
2Dr. B. Prabhakara Rao, Program Director of IST, JNTUK, Kakinada, India.
Manuscript received on 08 August 2019 | Revised Manuscript received on 15 August 2019 | Manuscript published on 30 August 2019 | PP: 3833-3839 | Volume-8 Issue-10, August 2019 | Retrieval Number: J99930881019/2019©BEIESP | DOI: 10.35940/ijitee.J9993.0881019
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Classification of medical image is an important task in the diagnosis of any disease. It even helps doctors in their diagnosis decisions. Tuberculosis (TB) is a disease caused by bacteria called Mycobacterium tuberculosis. Recently, several techniques are applied to diagnosis the TB diseases. Unfortunately, diagnosing TB is still a major challenge. Therefore, in this paper an efficient Tuberculosis diagnosis system is proposed using Multi Kernel Fuzzy C Means Rough Set (MKFCMRS) based feature selection and optimal neural network classifier. Our proposed method comprised of four stages namely, feature extraction, feature selection, classification and Region identification. Initially the TB images are extracted from the given input database and that each of the input images are given to feature extraction process, in which statistical, structural and gray level dependent features are extracted. After that, the feature selection scheme is carried out through multi-kernel FCM based rough-set theory. Then, selected features are given to optimal neural network classifier to optimize the weight values of the neural network. In this work proposed classifier is Particle Genetic Swarm Neural Network classifier (GPSO-NN) which Integrates the characteristics of both genetic and particle swarm methods. The proposed system is implemented in the working platform of MATLAB. Compared to previous method our proposed technique is improved in terms of accuracy, sensitivity and specificity.
Keywords: Tuberculosis, GPSO-NN, Rough-set Theory, Multi-Kernel FCM, Feature Extraction, Feature Selection.

Scope of the Article: Classification