2021/03/10 by Kyungmoo Lee, Alexis Warren, Lee, Kyungmoo +19
Medicine · #FOS: Biological sciences #FOS: Electrical engineering #Glaucoma and retinal disorders #Image and Video Processing (eess.IV) #Quantitative Methods (q-bio.QM) #Retinal Diseases and Treatments #Retinal Imaging and Analysis #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2103.06425
openalex publication_date 2021/03/10 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Background: Changes in choroidal thickness are associated with various ocular\ndiseases and the choroid can be imaged using spectral-domain optical coherence\ntomography (SDOCT) and enhanced depth imaging OCT (EDIOCT). New Method: Eighty\nmacular SDOCT volumes from 80 patients were obtained using the Zeiss Cirrus\nmachine. Eleven additional control subjects had two Cirrus scans done in one\nvisit along with EDIOCT using the Heidelberg Spectralis machine. To\nautomatically segment choroidal layers from the OCT volumes, our\ngraph-theoretic approach was utilized. The segmentation results were compared\nwith reference standards from two graders, and the accuracy of automated\nsegmentation was calculated using unsigned to signed border positioning\nthickness errors and Dice similarity coefficient (DSC). The repeatability and\nreproducibility of our choroidal thicknesses were determined by intraclass\ncorrelation coefficient (ICC), coefficient of variation (CV), and repeatability\ncoefficient (RC). Results: The mean unsigned to signed border positioning\nerrors for the choroidal inner and outer surfaces are 3.39plusminus1.26microns\n(mean plusminus SD) to minus1.52 plusminus 1.63microns and 16.09 plusminus\n6.21microns to 4.73 plusminus 9.53microns, respectively. The mean unsigned to\nsigned choroidal thickness errors are 16.54 plusminus 6.47microns to 6.25\nplusminus 9.91microns, and the mean DSC is 0.949 plusminus 0.025. The ICC (95%\nCI), CV, RC values are 0.991 (0.977 to 0.997), 2.48%, 3.15microns for the\nrepeatability and 0.991 (0.977 to 0.997), 2.49%, 0.53microns for the\nreproducibility studies, respectively. Comparison with Existing Method(s): The\nproposed method outperformed our previous method using choroidal vessel\nsegmentation and inter-grader variability. Conclusions: This automated\nsegmentation method can reliably measure choroidal thickness using different\nOCT platforms.\n