Detection of Kidney Condition Using Hidden Markov Models Based on Singular Value Decomposition

Detection of Kidney Condition Using Hidden Markov Models Based on Singular Value Decomposition

Detection of Kidney Condition Using Hidden Markov Models Based on Singular Value Decomposition

Vol.15 No.2 2015/Telkomnika (SCOPUS INDEX)

Siska Anraeni, Ingrid Nurtanio, Indrabayu

Abstract

The frequencies of chronic kidney disease are likely to continue to increase worldwide. So people need to take a precaution, which is by maintaining kidney health and early detection of renal impairment by analyzing the composition of the iris is known as iridology. This paper presents a novel approach usinga one-dimensional discrete Hidden Markov Model (HMM) classifier and coefficients Singular Value Decomposition (SVD) as a feature for image recognition iris to indicate normal or abnormal kidney. To accelerate algorithms and reduce computational complexity and memory consumption in hardware implementations, we used in a number of SVD small coefficients and 7-state HMM for the image of the model configuration.The system has been examined on 200 iris images.The total images of the abnormal kidney condition were 100 images and those for the normal kidney condition were 100 images. The system showed a classification rate up to 100% using total of image for training and testing the system unspecified, resize iris image 56×46 pixels, coefficients of singular values consists of orthogonal matrix is (1,1) and diagonal matrices are (1,1) and (2,2), quantized values [18 10 7], and classify by 7-state HMM with .pgm format database.

Keywords: Iridology, SVD, HMM