An empirical mode decomposition approach for automatic diagnosis of retina digital images

In this study six statistical textural features are extracted from retina digital images processed with the empirical mode decomposition (EMD). They are the mean, standard deviation, smoothness, third moment, uniformity, and entropy. The purpose is to classify normal and abnormal images. Five different pathologies are considered. They are artery sheath, blot hemorrhage, circinates, age-related macular drusens, and microaneurysms. Support vector machines are employed as classifier. Ten random folds are generated to perform cross-validation tests. The average and standard deviation of the correct recognition rate, sensitivity and specificity are computed for each simulation to assess the performance of the classifier. The obtained results generally outperform those given by using the discrete wavelet transform (DWT) instead of the EMD.