Analyzing Sub-Classifications of Glaucoma via SOM Based Clustering of Optic Nerve Images

We present a data mining framework to cluster optic nerve images obtained by Confocal Scanning Laser Tomography (CSLT) ill normal subjects and patients with glaucoma. We use self-organizing maps and expectation maximization methods to partition the data into clusters that provide insights into potential sub-classification of glaucoma based oil morphological features. We conclude that our approach provides a first step towards a better understanding of morphological features ill optic nerve images obtained from glaucoma patients and healthy controls.