Optimal segmentation of pupillometric images for estimating pupil shape parameters

Area of Science:

• Ophthalmology
• Medical Imaging
• Computational Neuroscience

Background:

• Pupil morphological parameters are vital for non-invasive diagnosis of central nervous system (CNS) disorders.
• Conditions like diabetes, Alzheimer's, schizophrenia, and addiction affect CNS responses, detectable via pupillometry.
• Accurate estimation of pupil features (diameter, area, centroid) is essential for early disease detection.

Purpose of the Study:

• To develop an accurate and efficient method for determining pupil morphological parameters from pupillometric data.
• To apply this method for early, non-invasive diagnosis of CNS disorders.
• To overcome limitations of existing numerical approximation schemes in pupil segmentation.

Main Methods:

• Utilizes an image segmentation algorithm based on the level set formulation of a variational problem.
• Proposes a discrete setup of the segmentation problem yielding a unique optimal solution.
• Employs a difference equation to evolve an initial curve to the optimal segmentation boundary, avoiding numerical approximations.

Main Results:

• Successfully estimates pupil geometrical features like diameter, area, and centroid coordinates.
• The proposed discrete method provides a unique optimal solution for segmentation.
• Eliminates the need for numerical approximation schemes common in continuum formulations.

Conclusions:

• The developed method offers an efficient and accurate approach for pupil parameter estimation.
• This technique has significant potential for the early, non-invasive diagnosis of various neurological and psychiatric conditions.
• The discrete formulation presents an advancement over traditional continuum-based segmentation methods in pupillometry.