Coronary Artery Disease Diagnosis Using IVUS Images Based on Empirical Curvelet Transform and Multiclass SVM

Abstract

: Intra-Vascular UltraSound (IVUS) is a medical imaging technique designed by catheter which has the miniaturized ultrasound probe which is attached to the end of the catheter. The computerized ultrasound equipment is attached to the proximal end. The amount of atheromatous plaque built up in the coronary artery is mainly identified by the IVUS image. In this study, a method for Coronary Artery Disease Diagnosis (CADD) system using IVUS is presented. The CADD system uses different IVUS images (normal, fibro fatty, calcium and thrombus) for performance evaluation. The CADD system uses pre-processing, feature extraction and classification techniques. Firstly, the speckle noise in the different input IVUS images is removed by frost filter. Then Empirical Curvelet Transform (ECT) is used for decomposition. The energy features is used to extract the features from decomposed images. Finally, prediction is made by multiclass Support Vector Machine (mSVM). The CADD system produces the accuracy of 95 % using ECT based energy features and mSVM.