Image Processing for Blood Vessel SegmentationKarianne Bergen

kbergen@stanford.eduJean-Baptiste Boinjbboin@stanford.edu

Sahinaz Sanjanisahinaz@stanford.edu

I. BACKGROUND AND MOTIVATION

Diabetic retinopathy is the leading cause of blindness amongadults aged 20-74 years in the United States [1] and isestimated to affect 28.5% of US adults with diabetes [2]. Ac-cording to the World Health Organization (WHO), screeningfor diabetic retinopathy is essential for diabetic patients andwill reduce the burden of disease [3]. Other diseases affectingthe retina can also be detected by a regular direct visualexamination.

The state of the retina is monitored by fundoscopic exam.These exams come in two forms: a hand-held fundoscopeused directly by the physician or fundus photography. Fundusphotography is usually performed by a fundus camera (Figure2) and has the advantage that the digital image of the retinacan be stored and analyzed later by a specialist.

Computer analysis assists the clinician in identifying clinicalmarkers for the presence, severity, and progression of retinaldiseases such as diabetic retinopathy. Vessel segmentationhighlights pathological features of blood vessels such as ab-normal branching, tortuosity, entropy and neovascularization[4]. Automatic blood vessel segmentation in the images canhelp speed diagnosis and improve the diagnostic performanceof less specialized physicians. Processing the screening imagecould highlight the locations of anomalies and comparingcurrent images to the ones of previous tests could even pointout the evolutions of the retina automatically.

The segmentation of retinal blood vessels has another po-

Fig. 1: A fundus camera (source: Wikipedia)

Fig. 2: Examples of retinal images and segmentation resultsfrom [9] (source: SpringerImages).

tential application outside of medical diagnostics. Similar tohuman fingerprints, the microvascular system of our retinais unique for each individual and usually does not evolveover the course of an individual’s lifetime. Using the properdevices, this retinal “fingerprint” is also easily accessible.Because of these properties, this could be used as a methodof identification.

II. PROJECT TASKS

For these reasons, algorithms extracting the vessels from thebackground are a necessary first step for many applications.The problem of retinal vessel segmentation has been widelystudied in the literature and we would like to understandand extend existing work on this subject. Our first goal willbe to review the existing literature, and to implement someof the state-of-the-art methods for vessel segmentation. Weintend to compare the performance of different methods anddevelop a scheme for merging different algorithms to improveperformance. As a second step, we hope to look more into aparticular application of these algorithms; for example, usingsegmentation results to diagnose a particular condition, or toidentify anomalies in a database of retinal photographies.

III. RETINAL IMAGE DATA SET

The data available to us for analysis comes from the DRIVEdatabase of retinal images [5], [6]. This database contains 40

images, 20 for training and 20 for testing. These images weremanually segmented by trained researchers. These imagescan be used for training in supervised algorithms, and formeasuring algorithm performance. Additional retinal imagesmay also be made available to use from Stanford School ofMedicine.

We will not be using an ANDROID device for thisproject.

REFERENCES

[1] Klein R, Klein B. Vision disorders in diabetes. In: National DiabetesData Group, ed. Diabetes in America. 2nd ed. Bethesda, MD: NationalInstitutes of Health, National Institute of Diabetes and Digestive andKidney Diseases; 1995, pp. 293337.

[2] Zhang X et al. ”Prevalence of Diabetic Retinopathy inthe United States, 2005-2008 JAMA. 2010; 304(6):649-656.http://jama.jamanetwork.com/article.aspx?articleid=186384#ref-joc05096-1

[3] World Health Organization. Prevention of Blindness and VisualImpairment. http://www.who.int/blindness/causes/priority/en/index6.html

[4] G. S. Ramlugun, V. K. Nagaraian, C. Chakraborty, Small retinal vesselsextraction towards proliferative diabetic retinopathy screening, ExpertSystems With Applications, 2012, vol. 39, pp. 1141-1146.

[5] J.J. Staal, M.D. Abramoff, M. Niemeijer, M.A. Viergever, B. vanGinneken, ”Ridge based vessel segmentation in color images of theretina”, IEEE Transactions on Medical Imaging, 2004, vol. 23, pp.501-509.

[6] M. Niemeijer, J.J. Staal, B. van Ginneken, M. Loog, M.D. Abramoff,“Comparative study of retinal vessel segmentation methods on a newpublicly available database”, in: SPIE Medical Imaging, Editor(s): J.Michael Fitzpatrick, M. Sonka, SPIE, 2004, vol. 5370, pp. 648-656.

[7] A.D. Hoover, V. Kouznetsova, M. Goldbaum, Locating blood vesselsin retinal images by piecewise threshold probing of a matched filterresponse, IEEE Transactions on Medical Imaging 19 (2000) 203210.

[8] M.M. Fraza, P. Remagninoa, A. Hoppea, B. Uyyanonvarab, A.R.Rudnickac, C.G. Owenc, S.A. Barmana, “Blood vessel segmentationmethodologies in retinal images A survey”, Computer Methods andPrograms in Biomedicine Volume 108, Issue 1, October 2012, Pages407433.

[9] U.M. Akram and A.S. Khan. “Automated Detection of Dark and BrightLesions in Retinal Images for Early Detection of Diabetic Retinopathy.”Journal of Medical Systems, Volume 36, Issue 5, November 2012.