Despeckling of Intravascular Ultrasound images using curvelet transform

Intravascular Ultrasound (IVUS) images are usually corrupted by different types of noises in the process of receiving, coding and transmission. Speckle noise is an inherent property of such images; it causes considerable degradation in the image quality, and thereby reducing the diagnostic value of this wide medical imaging procedure. Accordingly, the speckle denoising is considered to be an important and essential preprocessing step to be used for analysis and segmentation, whenever IVUS images are used for atherosclerotic lesions/plaques assessment. Curvelets based approach is the new multiscale geometric transform that has been widely used for noise reduction. Therefore, the curvelet transform (CT) is more efficiently than wavelet in the representation of image edges; a better enhancement scheme can be achieved. This paper presents an attempt for IVUS image denoising and structure enhancement using curvelet transform. The comparative evaluation of the speckle reduction performance is shown using statistical parameters for different noise types. The image quality parameters that have been used here are peak signal-to-noise ratio and normalized mean square error. The performance of the CT has also been compared with the wavelet transform method. The results included a series of in vivo IVUS images from ten patients.