Evaluation and calibration of a binocular infrared pupillometer for measuring relative afferent pupillary defect

Binocular infrared pupillometry allows an estimate of the relative afferent pupillary defect (RAPD), designated the pupillometric RAPD (pRAPD). We calibrated the pRAPD of a commercially available pupillometer against neutral density filters (NDFs) of known attenuation. The performance of the pupillometer using its own proprietary algorithm is assessed and compared to that of alternative algorithms.

The pRAPDs of 50 healthy volunteers were measured with each of 4 filters of known attenuation: 0.0, 0.3, 0.6, and 0.9 log units, positioned unilaterally in the light stimulus pathway. The filter values were plotted against the pupillometer output, and the slope and intercept were used to determine a calibration factor. Corrected pRAPD results were used to assess physiological ranges of pRAPD. The sensitivity and specificity to 0.3 log unit differences between increasing filter densities using receiver operator characteristic (ROC) curves.

The calibrated physiological pRAPD ranged from 0 to 0.22 log units. The area under the ROC curve for detecting unilateral simulated pRAPD of 0.3 log units, the simulated disease progression from 0.3 to 0.6 log units, and a further progression from 0.6 to 0.9 log units by NDFs was 0.99 (95% confidence interval [CI], 0.95-1.00), 0.86 (95% CI, 0.78-0.92), and 0.79 (95% CI, 0.70-0.87), respectively. The optimum discrimination was for detecting a unilateral simulated pRAPD of 0.3 log units; sensitivity and specificity was 98% (95% CI, 88%-99%).

The commercially available pupillometer detects the RAPD induced by the NDFs with high sensitivity and specificity. The results suggest that it is best for detecting unilateral early disease but potentially useful for assessing progression of disease.