The intima–media thickness (IMT) of the common carotid artery (CCA) is being used as a reliable and early detector of atherosclerosis. Atherosclerosis may be unnoticed fo years before triggering severe illnesses such as stroke, embolisms or ischemia. Hence, the use of IMT leads to an early atherosclerosis diagnosis that can prevent more serious cardiovascular diseases. Usually, IMT is manually extracted from ultrasound images, which is a non-invasive technique, but unfortunately its measurement is prone to error. This paper addresses a fully automatic method to segment the artery layers of the CCA over ultrasound images. Unlike other methods, the segmentation is not restricted to IMT, the artery diameter can be extracted too, which can help to determine cardiovascular risk together with IMT. The proposed technique is based on a frequency-domain implementation of active contours, which are computationally faster than the original space-formulation, while providing soft final contours. Working with three different probes over a range of spatial resolutions from 0.029mm/pixel to 0.081mm/pixel, the method presents an IMT error of only 13.8±31.9μm (in mean±standard deviation) when tested on an database containing 46 images. The automatic results were compared to the average of 2 manual observations performed by 2 observers (4 observations) over each image in our database.
