Abstract
Purpose: :
To determine the ability of various imaging modalities to detect clinically significant diabetic macular oedema (CSMO).
Methods: :
A total of 20 eyes with CSMO (defined using Binocular Indirect Ophthalmoscopy– BIO) and 40 diabetic eyes without oedema were used. Simultaneous stereoscopic images of all eyes were obtained using a Nidek 3–Dx camera; images were digitised and viewed stereoscopically using a Z screen (Stereographics Corp., San Raphael, CA). Three expert observers recorded the location of any CSMO using the Z screen. Images of all eyes were also obtained using a Retinal Thickness Analyser (RTA, Talia technology Ltd., Israel) with a retinal thickness scan and a Heidelberg Retina Tomograph II (HRT II, Heidelberg Engineering, Dossenheim–Germany) using the retinal module. To obtain sensitivity and specificity values for the RTA and the HRT II, two observers subjectively assessed the thickness scan for the RTA and the oedema maps on the HRT II for CSMO.
Results: :
The HRT II oedema index (using a 600 micron circle radius) for non CSMO and CSMO diabetic patients was 1.3 and 1.58, respectively. The RTA foveal thickness values (using a 600 micron circle radius) for non CSMO and CSMO diabetic patients was 170 microns and 192.1 microns, respectively. Both the RTA and HRT II values between non CSMO and CSMO patients were significantly different (unrelated t–test, p<0.05). Sensitivity and specificity values for detecting CSMO, as defined by BIO, for stereoscopic images viewed with a Z screen for a majority of observers were 15% and 97.5% respectively. Sensitivity and specificity values for the RTA were 60% and 70% respectively, and 75% and 77.5%, respectively for the HRT II.
Conclusions: :
The HRT II showed the highest sensitivity compared to the clinical assessment and the Z screen had the highest specificity (but lowest sensitivity). Instruments such as the RTA and HRT II can objectively identify CSMO and, therefore, have potential value in the management of diabetic patients.
Keywords: imaging/image analysis: clinical • diabetic retinopathy