A new computerized measurement approach of carotid artery stenosis on tomographic image sequence

The stenosis degree of carotid artery (CA) can be a critical factor for treatment of cerebrovascular disease and for determining candidate of carotid endarterectomy. Currently, three different measuring methods are applied only on projectional cervical images. These measurement methods introduce several demerits such as a thromboembolic event, and three reference positions provide the different measurement results even on same subject. In addition, projection image could not provide the most severe stenosis position by nature; and the manual measurements also provide the inter-observer and intra-observer variability. Therefore, a computerized measuring scheme is necessary to overcome these drawbacks.

By applying local adaptive thresholding technique on cervical magnetic resonance angiogram image sequence, CA objects are initially identified. These are used to determine the three-dimensional central axis of CA by using circumscribed quadrangle. The oblique slices are reformatted into two-dimensional image planes, which are perpendicular to the central axis of CA, to provide the circular shape of blood vessel provided that the artery runs horizontally across the scanning axis. After that, region growing technique is applied on obliquely reformatted image sequence followed by geometrically restoration of segmented CA objects.

The percentage of stenosis can be defined by the area ratio of segmented CA to restored CA object. The stenosis grading of is [(A-B)/A]×100%, where A represents area measure of restored object, B represents area measure of segmented CA object. Experiments have been conducted on both phantom that simulated the mild (30%), moderate (50%), and severe (70%) stenosis degree for validation of proposed measurement approach and 86 carotid arteries from 43 clinical data sets (including 5 occlusion cases).

The automated approach is recommended to measure the carotid stenosis by using axial image sequence. This technique is not only accurate as possible but also robust, simple to handle, and less time consuming as compared to manual measurements. In addition, a computerized carotid stenosis measuring method is necessary to overcome the drawbacks introduced by using the projectional image and measurement variability of inter-observer, intra-observer.