Preliminary evaluation of a SmartCam-based system for real-time pupil light reflex analysis

Abstract

The early detection of diabetes-related complications can support the clinicians decisions on appropriate interventional therapy and disease management for improved outcomes. The clinical investigation of the human pupil responsiveness to light stimulation, a technique known as dynamic pupillometry, has shown potential for a simple, inexpensive and non-invasive screening tool in supporting earlystage complications detection with. To facilitate the continuity of this research and to evaluate the applicability of this technique in clinical settings, a low-cost real-time stand-alone pupillometer is desired. In this paper, we present the preliminary results of a developed prototype capable of delivering such features. The proposed real-time pupillometer is based on a high-performance SmartCam system with applicationspecific software. This prototype was developed for automatic determination of the pupil radius enabling the determination of important pupillometry parameters such as the maximum and the minimum pupil diameters, the latency time to start the pupil constriction as a response to light flash stimulus, the pupil constriction velocity, among others. This SmartCambased prototype system using a digital signal processor can support real-time dynamic pupillometry at 30 frames per second using only 31.8% of the computational resources. Therefore, it could be used in our clinical studies with the capability to support additional features such as the real-time pupil light reflex evaluation in both eyes simultaneously. © Springer-Verlag Berlin Heidelberg 2007.