Abstract
Purpose: :
To compare automated versus manual analysis of Spectral Domain Optical Coherence Tomography (SDOCT) scans of patients with non-neovascular age related macular degeneration (AMD).
Methods: :
15 eyes from 15 patients with non-neovascular AMD with both drusen and geographic atrophy were prospectively imaged on the Topcon 3D-OCT 2000 (Topcon Medical Systems, Oakland, NJ) and the Cirrus HD-OCT (Carl Zeiss Meditec, Inc., Dublin, CA) using the 3D cube scan protocols on each device. Color fundus photos were taken of each eye at the same time as the scan was captured on the Topcon 3D-OCT 2000. The OCT data was then analyzed using the automated analysis software on the Topcon 3D-OCT 2000 to determine drusen count and area and on the Cirrus Advanced Retinal Pigment Epithelium (RPE) Analysis Tool to determine drusen count, drusen area and area of geographic atrophy in a 6mm diameter circle centered on the fovea. The values obtained with automated analysis were compared to manual determination of the drusen count and geographic atrophy area on the color fundus photographs. The images were then compared to the automatic segmentation data to determine if errors were made.
Results: :
On manual analysis the average total geographic atrophy area was 4.1 square mm and the average drusen count was 34. The Cirrus Advanced RPE tool reported a total geographic atrophy area of 4.8 square mm, average drusen count of 46, average drusen area of 3.8 square mm. The Topcon Drusen Analysis tool reported an average drusen count of 24, average drusen area of 6.4 square mm and did not calculate geographic atrophy area. Comparing the fundus photos to the Cirrus showed that there was a high correlation between manual measurement of the above parameters to the automated results.
Conclusions: :
The Cirrus Advanced RPE Analysis tool was much more accurate in automatically identifying drusen compared to the Topcon Drusen Analysis software; however, there was still a significant error rate in identifying drusen especially when pigment epithelium detachments (PED) were present. All low lying PEDs, also known as drusenoid PEDs, were incorrectly identified as drusen by the software on both devices. This resulted in an inaccurate drusen count and drusen area calculation. The Cirrus Advanced RPE analysis tool was very accurate in determining area of geographic atrophy but tended to result in a larger area compared to manual measurement on fundus photos. Further work is necessary prior to being able to use this tool to accurately determine drusen count and area on either device.
Keywords: age-related macular degeneration • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • imaging/image analysis: clinical