3D facial feature detection to aid clinical diagnosis

Abstract

Over the last decade, facial feature detection has been actively researched for face recognition. Nowadays, facial feature detection technology is widely used in the medical field to provide efficient support for medical research. Software application of facial feature detection is important in analyzing classifiers to aid clinical diagnosis of angle closure glaucoma. In this report, I present a 3D glaucoma facial feature detection software application - Glaucoma Detection System (GDS) which is able to measure and store large numbers of facial features. Experiments were conducted to investigate the relationship between different sets of facial features and Glaucoma. Certain sets of features are found to be greatly linked to Glaucoma, especially the width of the face and the intercanthal distance of the eyes. Using GDS, Glaucoma is detected in patients with an accuracy rate of 84.4% on average, with the help of 2 data mining algorithms – Local Weighted Learning (LWL) and Adaptive Boosting (AdaBoost).