Abstract
Abstract: :
Purpose: To present a software algorithm which, in conjunction with scanning laser tomography, will provide an absolute measure of retinal thickness and a high resolution thickness map over a 15° extent of retina. This will assist physicians in detection of subtle or diffuse increases in retinal thickness, and will enable early diagnosis and monitoring of Diabetic Macular Edema (DME). Methods: Scanning laser tomography images were acquired with the Heidelberg Retina Tomograph II (HRT II). The acquired image stack was aligned automatically to compensate for eye movement during acquisition. For each pixel (x, y) in the 15° image planes, the reflectance measured along the optical axis (z) was modeled by a forward model based on Poisson noise, and optics in the SLT imaging system and the human eye, developed from our previous work[1]. A maximum likelihood estimation algorithm was used to extract the retinal thickness parameter. Results: 48 data sets from 8 normal subjects and 8 diabetic subjects with clinically significant DME were analyzed. The test results demonstrate consistent accuracy when compared to benchmark measurements obtained with optical coherence tomography. The results of different data sets from the same subject at the same visit were highly reproducible. Conclusions: Our retinal thickness analysis algorithm with scanning laser tomography is a reliable, robust and accurate algorithm. Compared to existing retinal thickness algorithms, this new approach is advantageous in both providing an absolute measure of retinal thickness and a high resolution thickness map. [1] N. Hutchings et al. IOVS (2001) 42: Program #4253
Keywords: imaging/image analysis: clinical • diabetic retinopathy • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound)