DETECTION OF RETINAL DISEASE BY LOCAL BINARY PATTERN

Transcription

1 Volume 119 No , ISSN: (on-line version) url: DETECTION OF RETINAL DISEASE BY LOCAL BINARY PATTERN N.P. Jeyashree [1], A Deepak [2] [1,2] Saveetha School Of Engineering, Saveetha Institute of Medical and Technical Sciences, Thandalam. 2577

2 Abstract: A new approach to diagnose retinal diseases named diabetic retinopathy and agerelated macular degeneration are proposed in this work. These two diseases are the frequent cause for the vision loss. For this purpose the performance of local binary pattern as texture descriptor is used. The main focus is to investigate the discrimination capabilities of the fundus image by classifying the retinal image and by analysing the texture of retina background and to identify whether the eye is pathological or healthy eye. Keywords: Diabetic retinopathy, retina, agerelated macular degeneration. 1 INTRODUCTION: Diabetic retinopathy (DR) and age-related macular degeneration (AMD) are nowadays two of the most frequent causes of blindness and vision loss. In addition, the rate of these diseases will be high in future due to increase in diabetes and ageing population in the society. Therefore the need for automated screening is raised [1]. The motive is to classify between diabetic retinopathy (DR), age related macular degeneration (AMD) and normal fundus images by examining the texture of the retina background. The disease classification is carried out with image processing as a tool in matlab software with the help of local binary pattern algorithm.[2] Retina is a thin layer present behind the eye ball [3][4]. It is connected to the brain and it is responsible for the image visualization. The two major retinal disease are Diabetic retinopathy and age-related macular degeneration [5] Diabetic retinopathy- The diabetic retinopathy is one of the main problems for diabetic patients [6]. This occurs when high blood sugar levels cause damage to blood vessels in retina. These blood vessels can swell and leak, or they can close. Sometimes abnormal new blood vessels grow on the retina [7]. All of these lead to vision loss. The patients are not aware of any symptoms until it is to late for the treatment. Only way to identify the diabetic retinopathy is by fundus image.[8] Age-related-macular degeneration- The AMD is a common eye problem caused to a person who s is 50 years and more[9]. The macular is a sensitive part of the retina, located at the back of an eye. It is build with more sensitive cells [10]. Due to age, family history and smoking these cells begins to die. This leads to blurred vision or vision loss. [11] II LOCAL BINARY PATTERN: S. Dhanushkodi and M.Vasuki et.al has done research on diagnosis of diabetic retinopathy to prevent vision loss. The method used is soft computing neural network. But as a result the methodology used is well suited for the early diagnosis of the diabetic retinopathy disease and the severity of DR cannot be easily identified to prevent vision loss. [1] The local binary pattern has become popular in recent years. It is considered to be a powerful algorithm for texture classification [12]. It works more efficiently when combined with histogram of oriented gradient descriptor, which rapidly increases the detection performance. [13] LBP divides an image into pixels. Each pixel present in a cell is compared to the neighbouring pixel, which follows circular pattern that is clockwise or anticlockwise direction [15]. If the centre pixel value is greater than the neighbour pixel, the value assigned is 0 or if not 1 is assigned. As a result a binary digit is obtained, which can also be named as binary string. Then histogram process is carried out [16], which provides combination of pixels which are smaller or greater than the centre. Finally normalisation of the histogram is done. The LBP methodology has led to significant progress in texture analysis. It is widely used all over the world both in research and applications [17]. Due to its discriminative power and computational simplicity, the method has been very successful in many such computer vision Problems [18] which were not earlier even regarded as texture problems, such as face analysis and motion analysis. Local Binary Pattern (LBP) is a simple yet very efficient texture operator. It is invariance to 2578

3 grayscale changes and provides good performance. III Input image IMAGE METHODOLOGY FINAL RESULTS Re-scaling Fig 1: Block diagram representing identification of retinal disease image segment texture descriptor Fig 1 represents the block diagram for identification of retinal disease, where the first step is to capture the background of an eye, the obtained image is called fundus image which is given as input to the system. The input image is in form of RGB image which is referred to as true colour image stored in Matlab. [19] The next step is rescaling process, which is an essential step for many of the image processing applications. This technique used here to sharpening of the CT images into same size. The size used is 256*256. It also removes noise present in the image and provides an image with high resolution. The noise removal is done by median filter which is digital filtering technique normally used to eradicate the noise from an image.[20]. Thirdly image segmentation take place to represent an image into something that have good clarity and less complexity to understand. Image segmentation is basically used to locate objects and boundaries present in an images. The output of image segmentation is in form of segments that collaborate the entire image. When this process is applied to an images, particularly in medical field, the resulting image can be used to create 3D reconstructions with the help of interpolation algorithms like Marching cubes. In this step color planes separation is carried out. Since the input image is a rgb image.[16]. The next process is texture descriptor which is done by algorithm called Local Binary Pattern. The LBP is an example distinguishing proof which extricates the veins from fundus picture. This procedure occur after picture division which extricates the veins from retinal picture, which would be useful to identify the issue. Local Binary Pattern (LBP) is a effective texture operator which, assign values to the pixels of an image by thresholding the neighbor pixel and obtaining the result in form of binary number.[14]. The final step involves the disease detection whether it is affected by diabetic retinopathy, age-macular degeneration or healthy eye using block matching method. The main use of block matching classifier is to classify the disease in the fundus image. Moreover block matching is a process of locating matching micro-blocks in a set of digital frames for the effect of motion estimation. Hence this act as a key factor to identify the right issue of a person. [15] IV RESULT AND DISCUSSION: Healthy eye: Fig 2: (b) Grey image 2579

4 Fig2: (c) Red plane Fig2: (g) output image Eye affected by age-related macular degeneration: Fig2: (d) Green plane Fig 3: (a) Input image Fig2: (e) Blue plane Fig 3: (b) Grey image Fig2: (f) veins extracted from retinal image using LBP algorithm 2580

5 Fig 3: (f) Veins extracted from retinal image Fig 3: (c) red plane Fig 3: (g) Output image Eye affected by diabetic retinopathy: Fig 3: (d) Green plane Fig 4: (a) input image Fig 3: (e) Blue plane Fig 4: (b) Grey image 2581

6 Fig 4: (c) Red plane Fig 4: (d) Green plane Fig 4: (e) Blue plane Fig 4: (g) Output image For this work three fundus images are taken. Fig2: (a-g) Represents the images obtained from healthy eye. Fig2: (a) represents an input image also called as fundus image which is in form of RGB. Fig2: (b) displays the gray scale image, which is obtained by converting RGB into gray. The next step is color plane separation which is one of the basic steps in image processing. Fig2: (c) shows red plane, Fig2: (d) depicts green plane and Fig2: (e) displays blue plane. From the above color plane separation images it is clear that the green plane image is clear when compared to other two planes which provides clear background of an eye. The next step is vessel extraction, Fig2: (f) show vessel extraction image by Local Binary pattern algorithm. Finally Fig2: (g) depicts the output image for a healthy eye. The same process is carried out for other two images. Fig3: (a-g) displays images obtained for age-related macular degeneration followed by Fig4: (a-g) represents the images for diabetic retinopathy. Generally diabetic retinopathy and age-related macular degeneration have different texture. From the images it is clear that in age-related macular degeneration many blood vessels are affected when compared to diabetic retinopathy. The local binary pattern provides the details on both the disease and helps to identify them. 5 CONCLUSION: Fig 4: (f) Veins extracted from retinal image From the obtained results it is clear that local binary pattern provides better result in texture identification process as it directly extracts the veins from retinal image and the time consumption is also 2582

7 VI less in image classification when compared to existing methods like probabilistic neural networks. This work can also be further applied for health-care monitoring. REFERENCES: 1..Dhanushkodi and M.Vasuki Diagnosis System for Diabetic Retinopathy to Prevent Vision Loss, applied medical informatics, vol. 33, no. 3, pp. 1-11, M. Mookiah, U. R. Acharya, R. J. Martis, C. K. Chua, C. Lim, E. Ng,and A. Laude Evolutionary algorithm based classifier parameter tuning for automatic diabetic retinopathy grading: A hybrid feature extraction approach, international journal of engineering and advanced technology, vol.6, pp B. Kochner, D. Schuhmann, M. Michaelis, G. Mann, and K.-H. Englmeier, Course tracking and contour extraction of retinal vessels from color fundus photographs: Most efficient use of steerable filters for model based image analysis, in Proc. SPIE Medical Imaging 1998, pp M. Lalonde, L. Gagnon, and M.-C. Boucher, Non-recursive paired tracking for vessel extraction from retinal images, in Proc. Conf. Vision Interface 2000, May 2000, pp M. Beaulieu, Algorithme de detection de la macula sur les images dela retine, Centre de recherche informatique de Montréal, Montréal, Canada, Tech. Rep. CRIM-00/07-05, July In French. 6. F. Mendels, C. Heneghan, P. D. Harper, R. B. Reilly, and J.-Ph. Thiran, Extraction of the optic disk boundary in digital fundus images, in Proc. 1st Joint BMES/EMBS Conf., Oct. 1999, p H. Li, O. Chutatape, Boundary detection of optic disk by a modified ASM method, Pattern Recognition, Vol. 36, No. 9, 2003, pp S. Tamura, Y. Okamoto and K. Yanashima, Zero-crossing interval correction in tracking eye-fundus blood vessels, Pattern Recognition, Vol. 21, No. 3, 1988, pp Z. Liu, O. Chutatape and S. M. Krishnan, Automatic image analysis of fundus photograph, Proceedings of the International Conference of the IEEE Engineering in Medicine and Biology Society, Vol. 2, 1997, pp H. Li, O. Chutatape, Automatic detection and boundary estimation of the optic disk in retinal images using a modelbased approach, Journal of Electronic Imaging, Vol. 12, No. 1, 2003, pp T. Ojala, M. Pietikainen, and T. Maenpaa, Multi-resolution gray-scale and rotation invariant texture classification with local binary patterns, Pattern Analysis and Machine Intelligence, IEEE Transactions on, vol. 24, no. 7, pp , M. Heikkil, M. Pietikinen, and C. Schmid, Description of interest regions with local binary patterns, Pattern Recognition, vol. 42, no. 3, pp , Z. Yang and H. Ai, Demographic classification with local binary patterns, in Advances in Biometrics, ser. Lecture Notes in Computer Science, S.-W. Lee and S. Li, Eds., 2007, vol. 4642, pp

8 14. K. Oppedal, K. Engan, D. Aarsland, M. Beyer, O. B. Tysnes, and T. Eftestol, Using local binary pattern to classify dementia in MRI, in Biomedical Imaging (ISBI), 9th IEEE International Symposium on, May 2012, pp Research, Journal of, vol. 16, pp , L. Nanni, A. Lumini, and S. Brahnam, Local binary patterns variants as texture descriptors for medical image analysis, Artificial Intelligence in Medicine, vol. 49, no. 2, pp , S. Zabihi, M. Delgir, and H.-R. Pourreza, Retinal vessel segmentation using color image morphology and local binary patterns, in Machine Vision and Image Processing (MVIP), 6th Iranian, 2010, pp [12] S. Dhanushkod 17. S. Morales, V. Naranjo, J. Angulo, J. J. Fuertes, and M. Alca niz, Segmentation and analysis of retinal vascular tree from fundus images processing, in International Conference on Bio-inspired Systems and Signal Processing (BIOSIGNALS 2012), 2012, pp X. Zhang, G. Thibault, E. Decenci`ere, G. Quellec, G. Cazuguel, A. Erginay, P. Massin, and A. Chabouis, Spatial normalization of eye fundus images, in ISBI 2012 : 9th IEEE International Symposium on Biomedical Imaging, S. Morales, V. Naranjo, J. Angulo, and M. Alca niz, Automatic detection of optic disc based on pca and mathematical morphology, Medical Imaging, IEEE Transactions on, vol. 32, no. 4, pp , April N. V. Chawla, K. W. Bowyer, L. O. Hall, and W. P. Kegelmeyer, SMOTE: Synthetic minority over-sampling technique, Artificial Intelligence 2584

9 2585

10 2586