PURPOSE: We evaluated the ConfoScan3 confocal microscope and its associated software that allows automated analysis of the corneal endothelial morphology. METHODS: Images were taken on 30 normal subjects and 29 contact lens wearers with the Konan SP-9000 specular microscope and the ConfoScan3. The Konan images were analyzed with the KSS-300 software (center method). The Confoscan3 images were first analyzed with the automated method and then edited manually (semiautomated method). The agreement between methods was evaluated by calculating the difference between pairs of measurements, determining the mean and standard deviation of these differences, and the 95% limits of agreement. RESULTS: For normal subjects, all methods gave similar density values. The agreement with the Konan system was slightly better for the semiautomated method than the automated method. The automated method overestimated the degree of polymegethism (p < 0.001) and pleomorphism (p < 0.001). The semiautomated method showed substantial improvement. For contact lens wearers, agreement with the Konan system was poor for both automated and semiautomated methods. The automated method overestimated the degree of pleomorphism (p < 0.001) and the semiautomated method showed only modest improvement (p < 0.1). Both methods overestimated pleomorphism (automated p < 0.001; semiautomated p < 0.025). CONCLUSIONS: When using ConfoScan3 to evaluate the corneal endothelium of normal subjects, investigators should manually edit the processed images to achieve results comparable with the Konan system. When evaluating contact lens wearers, values from the ConfoScan3 are not interchangeable with those from the Konan system.
PURPOSE: We evaluated the ConfoScan3 confocal microscope and its associated software that allows automated analysis of the corneal endothelial morphology. METHODS: Images were taken on 30 normal subjects and 29 contact lens wearers with the Konan SP-9000 specular microscope and the ConfoScan3. The Konan images were analyzed with the KSS-300 software (center method). The Confoscan3 images were first analyzed with the automated method and then edited manually (semiautomated method). The agreement between methods was evaluated by calculating the difference between pairs of measurements, determining the mean and standard deviation of these differences, and the 95% limits of agreement. RESULTS: For normal subjects, all methods gave similar density values. The agreement with the Konan system was slightly better for the semiautomated method than the automated method. The automated method overestimated the degree of polymegethism (p < 0.001) and pleomorphism (p < 0.001). The semiautomated method showed substantial improvement. For contact lens wearers, agreement with the Konan system was poor for both automated and semiautomated methods. The automated method overestimated the degree of pleomorphism (p < 0.001) and the semiautomated method showed only modest improvement (p < 0.1). Both methods overestimated pleomorphism (automated p < 0.001; semiautomated p < 0.025). CONCLUSIONS: When using ConfoScan3 to evaluate the corneal endothelium of normal subjects, investigators should manually edit the processed images to achieve results comparable with the Konan system. When evaluating contact lens wearers, values from the ConfoScan3 are not interchangeable with those from the Konan system.
Authors: Ronald A Schachar; Susan Raber; Kristina V Thomas; Beth Ann M Benetz; Loretta B Szczotka-Flynn; Min Zhang; Scott J Howell; Jonathan H Lass Journal: Cornea Date: 2013-03 Impact factor: 2.651