Unveiling preclinical idiopathic macular hole formation using support vector machines

Macular holes are ruptures in the central part of the retina that if left untreated may lead to serious vision loss. Although there is a lot yet to know about this pathology, it is established that if one suffers from unilateral idiopathic macular hole (IMH), then there is an increased risk of developing the same condition in the fellow eye. The goal of this work is to use optical coherence tomography (OCT) scans and, resorting to automatic pattern recognition algorithms, develop a classifier that distinguishes eyes at risk of developing IMH from healthy controls. From the collected data we were able to estimate a set of parameters that allow for the classification of eyes into the group of eyes at risk or healthy controls with an accuracy of 95.1%, sensitivity of 96.9% and specificity of 93.1%.