Abstract:

Because of  shortcomings of the existing contour detection of the intravascular ultrasound (IVUS) image, an improved method used for automatically detecting  the lumen and mediaadventitia contours of  sequential IVUS images was presented. First, the speckle noise of the image was effectively removed  in the contourlet transform domain. Then, the initial lumen and mediaadventitia contours of the image were estimated according to physical properties and structural information of sequential images. Finally, using the active contour model and the grey level gradient of the denoised image, and according to the dynamic programming technique, the lumen and mediaadventitia contours of the image were automatically detected. Experimental results showed that the proposed method is algorithmically simple, statistically accurate, and has clinical value.

Key words:  intravascular ultrasound, speckle denoising, active contour model, dynamic programming, contour detection