Texture Analysis and Characteristic Identification about Plaque Tissues of IVUS

Intravascular ultrasound can provide clear real-time cross-sectional images, including lumen and plaque. In practice, to identify the plaques tissues in different pathological changes is very important. However, the grayscale differences of them are not so apparent. In this paper a new textural characteristic space vector was formed by the combination of Co-occurrence Matrix and fraction methods. The vector was projected to the new characteristic space after multiplied by a projective matrix which can best classify those plaques according to the Fisher linear discriminant. Then the classification was completed in the new vector space. Experimental results found that the veracity of this classification could reach up to 88%, and it would be an accessorial tool for doctors to identify each plaque.