Carotid Artery Plaque Image Recognition Using Gabor Wavelet and Principal Component Analysis

A pair of blood vessels inside of the human neck that serves to deliver blood to the brain is called carotid artery. Cholesterol in human body can form plaque, causes blockage to carotid artery that evoke atherosclerosis, stroke and heart disease which is a dangerous disease that can lead to death. If in certain long time it is not discovered, carotid artery will rupture. In clinical practice, the availability of ultrasound is wide also it is a low cost method to observe plaque in carotid artery. Unfortunately, ultrasound plaque images in carotid artery is diverse, noisy and not easy to be identified. It is also hard to develop computational techniques for recognizing plaque from ultrasound images. Therefore, it is a challenge to develop an optimal method that can be implemented in computer system to recognize plaque from ultrasound images. One method from many techniques available in pattern recognition is a feature extraction which can be obtained from various ways. In this work, A Gabor wavelet which is one of the powerful method in feature extraction is applied to recognize plaque characteristics. However a Gabor wavelet feature extraction will result a huge data, therefore to reduce the data dimension, the Principal Component Analysis (PCA) is applied to reduce such huge data. The result of this method for recognize plaque in carotid artery is satisfied with 100% recognition rate by using 8 orientations and 3 scales bank of Gabor with 100% eigenvectors configuration. In this research we used 24 carotid artery training images.