Dual-grader Specular Microscopy Analysis by Manual Cell Count. DIRC has validated a multi-grader process for manual analysis of non-contact (small-field) specular microscopy images, with very high intra- and inter-grader reproducibility. Accuracy and reproducibility of different manual grading methods (Center and Flex Center) have been validated separately. DIRC has been actively comparing these methods to each other and to automated and semi-automated alternatives under different conditions (seeResearch/Publications) in an effort to identify the most suitable method for minimizing the analysis time while ensuring accurate results for both normal and pathologic corneas. Parameters evaluated during the analysis (for either method) include cell density (CD), percent hexagonal cells (HEX), coefficient of cell size variation (CV), cell number used to derive the CD, presence and extent of guttae, image quality rating, and the relative standard deviation (RSD) between CD values. Our multi-step grading process, which relies entirely on masked certified graders, ensures minimal variation in most reported cases (RSD < 2.5%). The DIRC platform is currently optimized only for Konan imaging systems (all software versions), but efforts are already under way to convert images from other systems (e.g. Topcon, Tomey, HAI) for analysis (inquire here, if interested).
The Doheny Image Reading Center has developed two custom-built, CFR Title 21, part 11-compliant and validated software tools, GRADOR and OCTOR. Using these tools, DIRC has been able to go beyond the data that can be extracted from currently existing commercial algorithms.
GRADOR is a validated, part 11-compliant, reading center planar image grading software that allows the grader to manually mark and label the boundaries of all structures of interest. The GRADOR program then quantifies the number of pixels for each structure and converts this to an area or linear measurement based on the instrument/image-specific pixel-mm conversion factor. GRADOR-derived measurements are used for all comparative analyses between image types.
OCTOR is a validated, part 11-compliant, image-grading software tool. Graders use a computer mouse to draw the boundaries of all structures of interest. With the OCTOR grading tool, we can access a considerable amount of potentially valuable quantitative information that had previously been unavailable. Using this computer-assisted grading software, our highly trained graders produce highly reproducible, quantitative sub-analyses, even in eyes with complex diseases such as neovascular age-related macular degeneration.
3D-OCTOR platform supports multiple capture system devices, such as the Cirrus HD-OCT Macular Cube protocol, yielding the same results, regardless of the OCT capture system available and/or accessible at a clinical study site location.