Bayesian Network Based Classification of Optical Coherence Tomography Images for Diagnosis of Glaucoma using Discrete Wavelet Transform Compression

Abstract: In worldwide, Glaucoma is a second major retinal disease which results permanent blindness. Loss of Retinal Nerve Fiber Layer (RNFL) is the result of glaucoma disease. RNFL thickness is evaluated from Optical Coherence Tomography (OCT) images is an important diagnostics indicator for glaucoma disease. At the same time in medical field they were maintaining large volume of medical image data with low quality of image contrast, speckle noise, exact compression of OCT is difficult. To solve above issues, Discrete Wavelet Transform (DWT) based OCT and image compression is proposed in this work. In this work speckle noise are removed by using radar improved frost filter, secondly the RNFL features are extracted by using Improved Linear Discriminant Analysis. Then the OCT image is segmented by using K-mean clustering algorithm. Hence the severity of Glaucoma is classified by using Bayesian network. Finally Discrete Wavelet Transform (DWT) is used to compress the image without any significant loss in the diagonsability of the real image. Experimental result shows that the proposed Bayesian network is efficient for detecting the severity of the Glaucoma. Keywords: Optical Coherence Tomography, RNFL, Radar improved frost filter, Discrete Wavelet Transform, K-mean clustering algorithm, Bayesian Network.