Retinal Blood Vessels and Optical Disc Segmentation in Branch Retinal Vein Occluded Fundus Images Using Digital Image Processing Techniques

 

Ganesan P^1*, B.S. Sathish², L.M.I. Leo Joseph³, K.M. Subramanian⁴, V. Kalist⁵, K. Vasanth⁶

¹Department of Electronics and Communication Engineering, Vidya Jyothi Institute of Technology, Hyderabad

²Department of  Electronics and Communication Engineering, Ramachandra College of Engineering, Eluru

³Department of  Electronics and Communication Engineering, S.R.Engineering College, Warangal, India.

⁴Department of Computer Science and Engineering, Shadan College of Engineering, Hyderabad, India

⁵School of Electrical and Electronics, Sathyabama University, Chennai,  India

⁶ Department of Electronics and Communication Engineering, Vidya  Jyothi Institute of Technology, Hyderabad

 

ABSTRACT:

The segmentation of retinal blood vessels and optical disc is the most vital and challenging task to investigate the rigorousness of the various retinal diseases such as branch retinal vein occlusion.  There are lot of methods and algorithms are developed to address this issue i.e., for the precise segmentation of optical disc and blood vessels. However, every method has its own pros and cons. Retinal vein occlusion (RVO) happens due to the obstruction (blockage) of veins transporting blood with required nutrients and oxygen to the nerve cells in the eye’s retina. An obstruction in any one of the four smaller branch veins is referred to as a branch retinal vein occlusion (BRVO). It is one of the main retinal illnesses next only to diabetic retinopathy. Our proposed approach is a simple image processing based detection of optical disc and retinal blood vessels of branch retinal vein occluded fundus images.

 

 

 

1. INTRODUCTION:

The retina is a very thin film of light sensitive nerve tissue that positions the third inner layer of the eye¹. When light fall on the eye, it penetrates through the iris to the retina. In retina, image is focused and there is a conversion of chemical substances into electrical impulses that eventually activate the nerve impulses^2-3. These impulses are transmitted to the brain through the optic nerve to the brain resulting in sight. The function of the retina is same as the photographic film in a camera^5-8. So a scene of the visual is created in retina through the lens and cornea. Retinal vein occlusion (RVO) happens due to the obstruction (blockage) of veins transporting blood with required nutrients and oxygen to the nerve cells in the eye’s retina⁴.

 

If the obstruction is due to the main vein of the retina, it is termed as central retinal vein occlusion (CRVO), whereas an obstruction in any one of the four smaller branch veins is referred to as a branch retinal vein occlusion (BRVO) ^19-22.

 

Branch retinal vein occlusion (BRVO) is one of the main retinal illnesses next only to diabetic retinopathy²³. Approximately one percentage of population suffering from BRVO. BRVO could cause macular edema, intra retinal haemorrhage and vitreous haemorrhage etc., which would finally lead to vision impairment or even blindness²⁴. There is a lot of possibility for elderly people with cardiovascular disease and /or hypertension to endure from BRVO.  BRVO is not properly diagnosed and well treated; it could severely impair the patient’s vision. It could cause a potential damages as retinal edema, blurred vision, or blindness. 

 

 

 

 

 

 

 

Fig.1 (a) CRVO Vs BRVO                                                                                                          (b) BRVO fundus image

Figure 1(a) (Image Courtesy: South Bay Ophthalmology) illustrates the fundamental difference between CRVO and BRVO. The appearance of the BRVO fundus image is depicted in figure 1(b) (Image Courtesy: DRIVE,  https://www.isi.uu.nl/Research/Databases/DRIVE/)

 

 

Branch retinal vein occlusion (BRVO) is a widespread retinal illness leading to vision loss and is related to number of threat factors. The main reason for BRVO includes both local anatomic and systemic factors. Cardiovascular disease, diabetes mellitus, hyperlipidaemia, high body mass index, haematological disorders and hypertension are systemic risk factors whereas hyperopia and glaucoma are example for anatomic factors. It is important to note those high blood pressures can seriously disturb the vision by destructing retinal veins in the eye. High blood pressure is the most general complaint linked with BRVO. It causes a painless reduction in vision, resulting in distorted vision. The main symptoms of BRVO are listed as Blood vessels leaking into the retina (retinal hemorrhage), Swelling with fluid (Retinal edema) and twisted and thickened blood vessels

 

The segmentation of retinal blood vessels and optical disc is the most vital and challenging task to investigate the rigorousness of the various retinal diseases such as branch retinal vein occlusion²⁵.  There are lot of methods and algorithms are developed to address this issue i.e., for the precise segmentation of optical disc and blood vessels²⁴. Our proposed approach is a simple image processing based detection of optical disc and retinal blood vessels of branch retinal vein occluded fundus images.

 

2. METHODOLOGY:

Our suggested method for the segmentation of the retinal blood vessels and optical disc of branch retinal vein occluded retinal fundus image is depicted in figure 2. The proposed method is based on the basic digital image processing techniques and mathematical morphology. The outcome of the discrete wavelet transform for the segmentation of the blood vessels also discussed.  The proposed approach is elucidated as follows.

 

The input test image is acquired from the DRIVE database. The Digital Retinal Images for Vessel Extraction (DRIVE) database is freely accessible on the internet to facilitate relative investigations on the detection and segmentation of retina’s blood vessels in retinal fundus images for our research and experiments. (https://www.isi.uu.nl/Research/Databases/DRIVE/)^{4 ,17}. The pre-processing is the low level preliminary process in image processing to improve the quality of the image by suppressing the unnecessary disturbances (distortion) and / or enhance the necessary image features significant for further analysis and processing. The example for the pre-processing includes filtering, intensity adjustment, rotation, scaling, and translation. In our work image is resized to [400 700] for rapid processing. Contra harmonic mean filter is a non-linear mean filter which is better at removing gaussian type noise and preserving edge features than the mean filter. In our work contra harmonic mean filter of mask size 3 is utilized to remove the unnecessary distortion.

 

RGB color input image is comprised of three different channels such as red, green and blue^9-12. The reason for selecting only green channel for image processing applications is simple. Our human eye is more sensitive to green color as compared to others¹³. With natural light source (sunlight), the green channel generally has a large amount of light in it. In digital camera, there are twice as many green pixels as red or blue pixels. In addition, the green channel has not as much of noise than the red or blue channel. This is the reason why green channel of RGB color image is opting for many image processing applications. The CLAHE algorithm partitions the images into contextual regions and applies the histogram equalization to each one. CLAHE combines neighboring tiles using bilinear interpolation to eliminate artificially induced boundaries. To avoid amplifying any noise that might be present in the image, CLAHE to limit the contrast, especially in homogeneous areas.

Morphological opening is the composite operation of erosion followed by dilation. For dilation operation on binary images, the output pixel is set to 1 if any of the pixels is set to the value 1. Similarly, for erosion operation the output pixel is set to 0 if any of the pixels is set to 0. Opening smoothest the inside of the object contour, brakes narrow strips and eliminates thin portion of the image¹⁶. The median filter is a simple edge-preserving smoothing filter. It may be applied prior to segmentation in order to reduce the amount of noise in the images¹⁸. The filter works by sorting pixels covered by a NxN mask according to their grey value. The center pixel is then replaced by the median of these pixels, i.e., the middle value. In the next step, the median filtered out image is transformed into binary image using a global threshold method (otsu). The major advantage of the otsu method is the selection of the threshold level to reduce the intra-class difference of the 0 and 1 i.e., black and white pixels. It is necessary to remove all connected objects (components) the unnecessary small pixels to improve the appearance of the segmented output image. For this, ‘area opening’ operation is performed to eliminate all the associated objects in the threshold image that have lesser than 50 pixels.

 

 

Fig.2 Proposed system for the detection and segmentation of retinal blood vessels and optical disc

 

3. EXPERIMENTAL RESULTS:

Figure 4 illustrates the test image for the evaluation of the proposed method of retinal blood vessels and optical disc segmentation in branch retinal vein occluded fundus images using digital image processing techniques.

 

Fig. 4 Test Image

The test image is in RGB color space. So it is necessary to split the channels to gather necessary information^14-15. Figure 5 illustrates the distribution of red, green and blue pixels of test image.

 

With natural light source (sunlight), the green channel generally has a large amount of light in it. In digital camera, there are twice as many green pixels as red or blue pixels. In addition, the green channel has not as much of noise than the red or blue channel. This is the reason why green channel of RGB color image is opting for many image processing applications. The red, green and blue channels of test image is shown in fig 6.

 

 

Fig. 5 The distribution of red, green and blue pixels of test image

 

                                        (i) Red Channel                                     (ii) Green Channel                               (iii) Blue Channel

Fig. 6 the red, green and blue channels of test image

 

 

The green channel is complemented for more visualization which is shown in fig 7. The CLAHE algorithm partitions the images into contextual regions and applies the histogram equalization to each one. After performing the equalization, CLAHE combines neighboring tiles using bilinear interpolation to eliminate artificially induced  boundaries. To avoid amplifying any noise that might be present in the image, CLAHE to limit the contrast, especially in homogeneous areas. Image after CLAHE process is illustrated in fig 8.    

Fig. 7 Complement of green Channel

 

Fig 8. Image after CLAHE

 

Morphological opening smoothest the inside of the object contour, brakes narrow strips and eliminates thin portion of the image. The outcome of this operation is shown in fig 9. The optical disc detection and segmentation is depicted in fig 10.

 

Fig. 9 Morphological Open

 

 

 

Fig. 10 Segmentation of optic disk

 

The median filter is a simple edge-preserving smoothing filter. It may be applied prior to segmentation in order to reduce the amount of noise in the images^{22, 25}. The output of the median filter is shown in fig 11.

Fig. 11 Median Filtering

 

In the next step, the median filtered out image is transformed into binary image using a global threshold method (otsu). This is shown in figure 12(a). The major advantage of the otsu method is the selection of the threshold level to reduce the intra-class difference of the 0 and 1 i.e., black and white pixels. It is necessary to remove all connected objects (components) the unnecessary small pixels to improve the appearance of the segmented output image. For this, ‘area opening’ operation is performed to eliminate all the associated objects in the threshold image that have lesser than 50 pixels. This process’s outcome is illustrated in fig 12(b). Figure 13 is the segmentation of the retinal blood vessels.

 

Fig 12 (a) Thresholding

 

Fig 12 (b) Removal of small objects

 

Fig. 13 The outcome of the proposed work

4. CONCLUSION:

A simple image processing based detection of optical disc and retinal blood vessels of branch retinal vein occluded fundus images is explained and analyzed. The proposed method is based on the simple image processing operations such as preprocessing, filtering, morphological operations and thresholding. The proposed method is tested with the images acquired from the DRIVE database. The experimental analysis demonstrated the competence of the proposed method for detection of optical disc and retinal blood vessels of branch retinal vein occluded fundus images.

 

5. ACKOWLEDGMENT:

The authors would like to thank the Image Sciences Institute for the utilization of the image (DRIVE database) for our research work

 

6. ETHICS AND CONSENT:

This article does not contain any studies with human participants or animals performed by any of the authors. No direct participation of human entertains in this article.

 

7. CONFLICT OF INTEREST:

We are declaring that, there is no conflict of interest regarding the publication of this paper.

 

8. REFERENCES:

1.          Rossant F, Badellino M, Chavillon A, Bloch I, and   Paques M. A morphological approach for vessel segmentation in eye fundus images, with quantitative evaluation. J. Med. Imaging. Health. Inf.  1(2); 2011; 42–49.

2.          Soares J and Cree M. Retinal vessel segmentation using the 2D Gabor wavelet and supervised classification.  IEEE Trans. Med. Imag.  25; 2006; 1214–1222.

3.          Zhang B, Zhang L, Zhang L, and Karray F. Retinal vessel extraction by matched filter with first-order derivative of Gaussian. Comput. Biol. Med.  40(1); 2010; 438–445.

4.          J.J. Staal, M.D. Abramoff, M. Niemeijer, M.A. Viergever, B. van Ginneken, "Ridge based vessel segmentation in color images of the retina", IEEE Transactions on Medical Imaging, 2004, vol. 23, pp. 501-509.

5.          Li H, Chutatape, O. Automated feature extraction in color retinal  images by a model based approach.. IEEE Trans. Biomed. Eng. 51; 2004; 246–254.

6.          Jelinek HF, Cree MJ, Leandro JJG, Soares JVB, Cesar RM, and Luckie A, "Automated segmentation of retinal blood vessels and identification of proliferative diabetic retinopathy", JOSA A 24(5): 1448-1456, 2007.

7.          Luo, C. Opas, and Shankar M. Detection and measurement of retinal vessels in fundus images using amplitude modified second-order Gaussian filter. IEEE Trans. Biomed. Eng. 49(1); 2008; 168–172.

8.          Perfetti R, Ricci E, Casali D, et al., "Cellular neural networks with virtual template expansion for retinal vessel segmentation", IEEE Transactions on Circuits and Systems II 54(2): 141-145, 2007.

9.          Kalist V, Ganesan P,  Sathish BS, and Jenitha JMM. Possiblistic-Fuzzy C-Means Clustering Approach for the Segmentation of Satellite Images in HSL Color Space.  Procedia Computer Science. 57; 2015; 49-56.

10.        Shaik KB, Ganesan P, Kalist V, and Sathish BS. Comparative Study of Skin Color Detection and Segmentation in HSV and YCbCr Color Space.  Procedia Computer Science. 57; 2015; 41-48.

11.        Ganesan P and Shaik KB. HSV color space based segmentation of region of interest in satellite images.  2014 International Conference on Control, Instrumentation, Communication and Computational Technologies (ICCICCT). 2014; 101-105.doi: 10.1109/ICCICCT.2014.6992938

12.        Sajiv G and Ganesan P.  Comparative Study of Possiblistic Fuzzy C-Means Clustering based Image Segmentation in RGB and CIELuv Color Space.  International Journal of Pharmacy & Technology. 8(1); 2016; 10899-10909.

13.        Malay Bhushan, Niraj Bishwash, and kalist V. Wireless Power Transfer Platform for Smart Home Appliances. International Journal of Pharmacy & Technology. 8(3); 2016; .15669-15674.

14.        Sajiv G. Unsupervised Clustering of Satellite Images in CIELab Color Space using Spatial Information Incorporated FCM Clustering Method.  International Journal of Applied Engineering Research. 10(20); 2015.

15.        Sathish BS, Ganesan P and Khamar Basha.Shaik. Color Image Segmentation based on Genetic Algorithm and Histogram Threshold. International Journal of Applied Engineering Research. 10(6); 2015; 123-127.

16.        Thakur M, Raj I and Ganesan P. The cooperative approach of genetic algorithm and neural network for the identification of vehicle License Plate number. International Conference on Innovations in Information, Embedded and Communication Systems (ICIIECS). 2015; 1-6.

17.        https://www.isi.uu.nl/Research/Databases/DRIVE/

18.        Ganesan P and B. S. Sathish. Automatic Detection of Optic Disc and Blood Vessel in Retinal Images using Morphological Operations and Ipachi Model. Research J. Pharm. and Tech. 10(8): August 2017; 2602-2607.

19.        Adam Hoover and Michael Goldbaum, “ Locating the optic nerve in a retinal image using the fuzzy convergence of the blood vessels”, IEEE Transactions on Medical Imaging, Vol. 22, No. 8, August 2003, pp.951-957.

20.        Guillermo Ayala, Teresa León, and Victoria Zapater, “ Different Averages of a Fuzzy Set with an application to Vessel Segmentation”, IEEE Transactions on Fuzzy Systems, Vol. 13, No. 3, June 2005, pp.384-393.

21.        Ganesan P, M.Ganesh , L.M. I. Leo Joseph and V. Kalist, “ Central Retinal Vein Occlusion: An Approach for the Detection and Extraction of Retinal Blood Vessels”,  J. Pharm. Sci. & Res. Vol. 10(1), 2018, 192-195.

22.        Meindert Niemeijer, Bram van Ginneken,  Maria S. A. Suttorp-Schulten, and Michael D. Abràmoff, ” Automatic Detection of Red Lesions in  Digital Color Fundus Photographs, IEEE Transactions on Medical Imaging, Vol. 24, No. 5, May 2005, pp.584-592.

23.        Tatijana Stoˇsic´ and Borko D. Stoˇsic´, “Multifractal Analysis of Human Retinal Vessels”, IEEE Transactions on Medical Imaging, Vol. 25, No. 8, August 2006. pp.1101-1108.

24.        Ganesan P,, “Detection and Segmentation of Retinal Blood Vessel in Digital RGB and CIELUV color space Fundus Images”,   Research J. Pharm. and Tech. 11(6): 2018, 2326-2330.

25.        Ana Maria Mendonça,  and Aurélio Campilho, ” Segmentation of Retinal Blood Vessels by Combining the Detection of Centerlines and Morphological Reconstruction”, IEEE Transactions on Medical Imaging, Vol. 25, No. 9, September 2006, pp.1200-1213.

 

 

 

 

 

Received on 07.12.2018           Modified on 19.01.2019

Accepted on 08.02.2019         © RJPT All right reserved