Spatial anisotropic diffusion and local time correlation applied to segmentation of vessels in ultrasound image sequences

The aims of this work are the reduction of speckle noise, the detection of the residual lumen shape, and the segmentation of arteriosclerosis plaque, in intravascular ultrasound (IVUS) image sequences, when blood flow and plaque present a similar echogenicity. This can be achieved with a 2D+T method using spatial and time information combining a speckle reducing edge-enhancing anisotropic diffusion method with a temporal local correlation evaluating the blood displacement area. This process allows obtaining accurate vessel segmentation and an efficient de-noising effect on the IVUS images.