PURPOSE: To find the most appropriate color-coded scales for the anterior and posterior elevation maps of scanning slit topography in the screening of abnormal corneas such as keratoconus. DESIGN: Retrospective case-control study. PARTICIPANTS: Eighty eyes of 40 normal subjects and 175 eyes of 95 patients with keratoconus. INTERVENTION: Anterior and posterior corneal elevations were assessed using Orbscan 2. Best-fit sphere maps were drawn with several color-coded scales: 2-, 5-, 10-, and 20-microm height per each color interval. MAIN OUTCOME MEASURES: The maps were judged to be abnormal when more than three colors (discriminant number) were found within the central 3-mm area. For each color-coded scale, sensitivity, specificity, positive predictive value, negative predictive value, and sensitivity + specificity were calculated. After determining the most appropriate color-coded scales for the anterior and posterior elevation maps, validity of the discriminant number was assessed. By varying the discriminant number from two to eight, receiver operator characteristic (ROC) curves were created using the sensitivity and specificity for each threshold number. RESULTS: The highest sensitivity + specificity values and highly balanced predictive values were obtained with the 10- and 20-microm scales for the anterior and posterior elevation maps, respectively. The ROC curve analyses showed that the best discriminant color number is three, indicating that maps with four or more colors within the central 3-mm area are judged abnormal in screening. CONCLUSIONS: The 10- and 20-microm interval color scales are most appropriate for the anterior and posterior elevation maps of the scanning slit topography, respectively.
PURPOSE: To find the most appropriate color-coded scales for the anterior and posterior elevation maps of scanning slit topography in the screening of abnormal corneas such as keratoconus. DESIGN: Retrospective case-control study. PARTICIPANTS: Eighty eyes of 40 normal subjects and 175 eyes of 95 patients with keratoconus. INTERVENTION: Anterior and posterior corneal elevations were assessed using Orbscan 2. Best-fit sphere maps were drawn with several color-coded scales: 2-, 5-, 10-, and 20-microm height per each color interval. MAIN OUTCOME MEASURES: The maps were judged to be abnormal when more than three colors (discriminant number) were found within the central 3-mm area. For each color-coded scale, sensitivity, specificity, positive predictive value, negative predictive value, and sensitivity + specificity were calculated. After determining the most appropriate color-coded scales for the anterior and posterior elevation maps, validity of the discriminant number was assessed. By varying the discriminant number from two to eight, receiver operator characteristic (ROC) curves were created using the sensitivity and specificity for each threshold number. RESULTS: The highest sensitivity + specificity values and highly balanced predictive values were obtained with the 10- and 20-microm scales for the anterior and posterior elevation maps, respectively. The ROC curve analyses showed that the best discriminant color number is three, indicating that maps with four or more colors within the central 3-mm area are judged abnormal in screening. CONCLUSIONS: The 10- and 20-microm interval color scales are most appropriate for the anterior and posterior elevation maps of the scanning slit topography, respectively.