Abstract
Purpose:
To develop an algorithm for the quantitative assessment of the retinal capillary microcirculation in diabetes mellitus type I (DM I) patients.
Methods:
Thirty-two images of 32 eyes (16 eyes of 9 DM I patients and 16 eyes of 10 non DM patients) were chosen from the University Hospital of Heraklion digital fluorescein angiography database. The age was 18±5 years for the patient group (range: 12-26 years) and 17±7 years (range: 6-26 years) for the control group. For each eye a high resolution image was chosen and underwent a processing procedure using a commercial software (MatLab R2011a; The MathWorks Inc.). The so far developed algorithm traces both manually (by choosing with the cursor) and automatically the perifoveal capillary network in a subimage of field 20o*20o of the original one and provides measurements of the foveal avascular zone (FAZ) surface, the capillary density and the branch point density in the mentioned area.
Results:
The capillary mapping revealed a FAZ area of 0.23±0.06 degrees2 in the DM I group versus 0.22±0.05 degrees2 in the control group. The capillary density (capillary length in degrees/total area in square degrees) was 2.48±0.55 degrees-1 and 2.76±0.21 degrees-1 in each group, respectively. The last metric estimated, the branch point density, was 2.86±0.73 branch points/degrees2 in the diabetic group and 3.14±0.7 branch points/degrees2 in the control group. It seems that there is a slight reduction of these indexes in DM I patients when comparing with controls but none of these differences were statistically significant.
Conclusions:
This approach constitutes our first attempt in order to develop an algorithm for the quantification of retinal microvessel alterations in young DM I patients. Further improvement of the algorithm will help us optimize the detection module and develop automated metrics for the quantification of diabetic retinal microangiopathy.