Automatic segmentation of carotid B-mode images using fuzzy classification

Abstract

This paper presents a new method for the automatic segmentation of the common carotid artery in B-mode images. This method uses the instantaneous coefficient of variation edge detector, fuzzy classification of edges and dynamic programming. Several discriminating features of the intima and adventitia boundaries are considered, like the edge strength, the intensity gradient orientation, the valley shaped intensity profile and contextual information of the region delimited by those boundaries. The adopted fuzzy classification of edges helps avoiding low-pass filtering. The method is suited to real-time processing and user interaction is not required. Both the near and far wall boundaries can be detected in arteries with plaques of different types and sizes. Both expert manual and automatic tracings are significantly better for the far wall, due to the better visibility of the intima and adventitia boundaries. The automatic detection of the far wall shows an accuracy similar to the manual detections. For this wall, the error coefficient of variation for the mean intima-media thickness is in the range [5.6, 6.6 %] for automatic detections and in [6.7, 7.1 %] for manual ones. In the case of the near wall, the same coefficient of variation is in [11.2, 13.0 %] for automatic detections and in [5.9, 9.0 %] for manual detections. The mean intima-media thickness measurement errors observed for the far wall [Formula: see text] are among the best values reported for other fully automatic approaches. The application of this approach in clinical practice is encouraged by the results for the far wall and the short processing time (mean of 2.1 s per image).

Robust common carotid artery lumen detection in B-mode ultrasound images using local phase symmetry

“…The first strategy consists on performing a column-wise vertical analysis of the image to locate candidate points at the CCA lumen or walls, which are then classified and connected into line segments [6,7,8,9]. The second strategy uses a dynamic programming (DP) algorithm to find a path, from one image border to the other, by optimizing a global evaluation function that points the interesting image locations [2,5]. The proposed method in this paper follows this second strategy, as it provides an appropriate framework to integrate fuzzy low level image features and criteria that do not only consider the vertical direction.…”

“…The proposed method in this paper follows this second strategy, as it provides an appropriate framework to integrate fuzzy low level image features and criteria that do not only consider the vertical direction. Regardless of the strategy, the CCA detection process may be based on vessel wall shape detection [6,7,8,9], or based on lumen region features as the average echogenicity [7,5], gray level variability [6,7] or lumen diameter [6,5].…”

“…Several accurate and automatic IMT measurement algorithms have been proposed in the last few years [2,3,4,5]. Most of the existing methods require some kind of user interaction to locate a region of interest (ROI) in the image containing the carotid artery walls.…”

“…Some other algorithms, usually referred to as fully-automatic, incorporate a CCA detection algorithm as a previous step for ROI location [2,5,6,7,8,9]. These algorithms can either provide information about the lumen [2,6,5], which is the area of the vessel where the blood flows, or a more detailed ROI that include approximations of the artery walls [7]. In either case, the objective is to locate the vessel structure whithin the image so that the posterior IMT segmentation and measurement process does not get confused with other similar structures in the image.…”

“…These guidelines recommend to adjust the overall gain to avoid most of lumen noise, and the time gain compensation to achieve similar brigthness of both artery walls. That is why some intensity-based methods have been succesfully used for CCA detection [7,5], regardless of not being specifically well suited for ultrasound image analysis. Nevertheless, these ultrasound settings are ultimately stabilished according to the subjective judgement of the operator, and it may not be possible to achive homogeneous imaging along the whole CCA.…”