Purchase this article with an account.
M. Moscaritolo, R. Zeimer; Development of an Algorithm to Assess Fixation Performance During Fundus Photography. Invest. Ophthalmol. Vis. Sci. 2009;50(13):326.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Proper target fixation is important in subjects undergoing multi-field tests such as fundus photography. A simple method to alert the operator of poor tracking of fixation targets could be useful during data acquisition to ensure quality of the results.
The eye was illuminated by two infra-red diodes and imaged live (12 frames per second) by a Webcam type camera. An automated algorithm was developed on a Matlab (Mathworks, USA) platform to analyze the digital images and detect the location of the diode reflections on the cornea. The pair of reflections was detected by convolution with a dedicated filter and the coordinates were used to generate a region of interest containing the pupil. The pupil was delineated by applying a dedicated 2-dimensional filter and performing a pattern recognition process based on blob analysis. The coordinates of the blob center were taken as the center of the pupil. A pilot test was carried out to assess the precision. Fields acquired at 11 locations on the fundus with the DigiScope (Eyetel Imaging, Inc) were examined and the corresponding pupil images were included if the location on the fundus was correct. The first 20 left eyes with adequate fixation at all fields were analyzed.
The standard deviation of the pupil center was 1.40 mm horizontally and 0.94 mm vertically. When translated to location on the fundus the precision was on average 3.8° vertically and 2.5° horizontally. When the subject is presented with a new fixation target the center of the pupil is expected to shift accordingly. The probability of a random shift by 2° and 4° in both the vertical and horizontal directions is 6% and less than 1%, respectively.
This pilot study demonstrates that a simple method can be implemented to assess fixation performance. The precision of determining the gaze is too low for most purposes but adequate to detect small changes in fixation reliably thereby providing an indication that the subject is following a target correctly.
This PDF is available to Subscribers Only