The accuracy of the detection procedure was assessed by determining its specificity from control data and its sensitivity both from control data and from subjects with albinism. The specificity of the detection of the misrouting of the optic nerves is indicated in
Figure 2 , where the mean correlation coefficients ± 2 SD (97.5% confidence interval) of the 16 control subjects are depicted. It is evident that a correlation coefficient threshold of 0 allows one to detect the misrouting with a
specificity of 100% (i.e., without false alarms).
2 To assess the
sensitivity of the detection procedure, one has to determine the proportion of negative correlation coefficients obtained for a pair of cortical responses that are
known to be localized on opposite hemispheres. Such a measure can be derived from the control data. In controls, responses to stimuli in opposing visual hemifields are lateralized on opposing hemispheres. Thus, sensitivity can be estimated for each stimulated eccentricity by correlating the difference VEPs obtained for stimulation in the left and the right hemifields at the same absolute eccentricity. We determined the proportion of negative correlation coefficients as a measure for the sensitivity from the control data for the respective eccentricities (0°-2.5°, 2.5°-6.0°, 6.0°-11.5°, 11.5°-17.0°, and 17.0°-27.0°): 96.5%, 100%, 100%, 91%, and 88%. The decrease of sensitivity toward the periphery is likely to be associated with a decrease of the signal-to-noise ratio of the signals in the periphery. Further, interhemispheric asymmetries of cortical morphology are likely to contribute to a reduction of the sensitivities. While the above sensitivity estimate is derived from control data, a similar approach can be taken directly with the data from the subjects with albinism. In albinism, the nasal retinas of both eyes follow the normal projection pattern and are represented on opposing hemispheres. A correlation of the interhemispheric activation differences for stimulation at the same eccentricities of the nasal retina of both eyes is, therefore, expected to yield negative correlation coefficients. Indeed, we obtained, in all subjects with albinism and for the 5 visual field positions tested, negative correlation coefficients (i.e., 100% sensitivity). These results underscore the high sensitivity of our approach and indicate that interocular differences within a subject do not corrupt the correlation analysis. Taking the results for controls and subjects with albinism together, the data indicate the following sensitivities for the respective eccentricities (0°-2.5°, 2.5°-6.0°, 6.0°-11.5°, 11.5°-17.0°, and 17.0°-27.0°): 98%, 100%, 100%, 94%, and 92% (average sensitivity, 97%).