We recruited 10 test subjects, age 28 to 60, none of them were on any medication affecting pupil size, and all properly consented. After a period of 10 min of room-light adaptation, the subjects were asked to fixate on a white-light accommodative target (a red dash with a white border, , optically 33 cm from the eye. The target was front-on illuminated constantly by a faint electric bulb, providing background luminance in the area of the target of about , just enough to enable the test subject to fixate. The ambient illumination was turned down immediately after initiation of the recording, from 27 to . Pupil diameter was measured under monocular conditions (with one eye occluded) by means of an eye tracking apparatus14 using video oculography and comprising an infrared-sensitive USB video camera ( resolutions; Web Digital Camera, Hong Kong) equipped with a 12-mm fixed-focal-length lens (Fig. 1). Near-infrared illumination of the pupil was provided by an infrared light emitting diode (OD-50L, 880 nm; Opto Diode Corp., Inc., Newbury Park, California). The camera was connected to a desktop computer that displayed the incoming stream of frames and controlled the video frame capture using custom acquisition software written in MATLAB® (MathWorks, Inc., Natick, Massachusetts). We used image acquisition with a frame rate of 5 fps, with continuous recording. The recorded eye’s image sequences were analyzed off line. Pupils were approximated with circles, and their diameters were calculated with commercial eye tracking software (IRIS; Chronos Vision, Berlin, Germany). Pupil detection uses edge detection and the Hough transform to identify a circle in a parameterized space.15,16 Blinks were detected as abrupt drops of more than 30% in pupil diameter, lasting for 200 to 400 ms, and were replaced by the preceding value. Pupil area was calculated based on the diameter measured from each frame. In order to compare pupil behavior across test subjects and possibly derive a general trend, the pupil area traces were normalized as Display Formula
(1)where is the area measured in time, is the baseline value at the initial moment when the light was turned off, and is the normalized area. Of the 10 subjects recorded, only five managed to maintain both arousal and fixation during the first 6 min of measurement, as determined by the video monitoring. The data from the uncooperative subjects were excluded from analysis.