NON-INVASIVE IMAGE ENHANCEMENT OF COLOUR RETINAL FUNDUS IMAGES FOR A COMPUTERISED DIABETIC RETINOPATHY MONITORING AND GRADING SYSTEM

Diabetic Retinopathy (DR) is a sight threatening complication due to diabetes
mellitus affecting the retina. The pathologies of DR can be monitored by analysing
colour fundus images. However, the low and varied contrast between retinal vessels
and the background in colour fundus images remains an impediment to visual analysis
in particular in analysing tiny retinal vessels and capillary networks. To circumvent
this problem, fundus fluorescein angiography (FF A) that improves the image contrast
is used. Unfortunately, it is an invasive procedure (injection of contrast dyes) that
leads to other physiological problems and in the worst case may cause death.
The objective of this research is to develop a non-invasive digital Image
enhancement scheme that can overcome the problem of the varied and low contrast
colour fundus images in order that the contrast produced is comparable to the invasive
fluorescein method, and without introducing noise or artefacts. The developed image
enhancement algorithm (called RETICA) is incorporated into a newly developed
computerised DR system (called RETINO) that is capable to monitor and grade DR
severity using colour fundus images. RETINO grades DR severity into five stages,
namely No DR, Mild Non Proliferative DR (NPDR), Moderate NPDR, Severe NPDR
and Proliferative DR (PDR) by enhancing the quality of digital colour fundus image
using RETICA in the macular region and analysing the enlargement of the foveal
avascular zone (F AZ), a region devoid of retinal vessels in the macular region. The
importance of this research is to improve image quality in order to increase the
accuracy, sensitivity and specificity of DR diagnosis, and to enable DR grading
through either direct observation or computer assisted diagnosis system.