A Graph-based Algorithm for Automatic Arterial Lumen Detection in Ultrasound Imaging

作者：Calderon Felix^*; Rogelio Tinoco Martinez Sergio; Carranza Madrigal Jaime

来源：Revista Iberoamericana de Automatica e Informatica Industrial, 2013, 10(4): 423-433.

DOI：10.1016/j.riai.2013.05.011

• Automatic detection
• ultrasonography
• carotid
• brachial
• arterial lumen
• graphs
• polynomial fitting

Cardiovascular diseases have become the first cause of dead in Mexico and the whole world. Intima-media thickness and brachial lumen diameter measurement in ultrasound images are the basis of two early diagnostic tests for this kind of illnesses. In this paper a methodology for automatic arterial lumen detection using ultrasound images, which is based on a graph and edge detection, is presented. The graph is created with middle points between edges and, its minimum spanning tree, is used together with decision criteria based on darkness and length, for the correct arterial lumen segmentation. In 294 images, a mean error in position detection of brachial lumen-intima interface on the near wall of 13.9 mu m, with a standard deviation of 12.3 mu m, was found; and, for same interface on the arterial far wall, mean error was of 15.4 mu m with a standard deviation of 15.0 mu m. Performance comparison of our algorithm was made against results presented in recent literature for automatic and semi-automatic techniques in this area, to whom it outperformed in accuracy.

• 出版日期2013-12