作者: Isabel M. Adame , Rob J. van der Geest , David A. Bluemke , João A.C. Lima , Johan H.C. Reiber
DOI: 10.1002/JMRI.20662
关键词:
摘要: Purpose To develop an automated technique to trace the contours of lumen and outer boundary aortic wall, measure wall thickness in axial MR images. Materials Methods The algorithm uses prior knowledge vessel morphology. A geometrical model (ellipse) is deformed, translated rotated obtain a rough approximation contours. Model-matching based on image gradient measurements. To enhance edges, images were preprocessed using gray-level stretching. Refinement performed by means dynamic programming. Wall computed measuring distance between inner contour wall. Results The has been tested high-resolution from 28 human subjects descending thoracic aorta. The results demonstrate: High correspondence automatic manual area measurements: (r = 0.99), 0.96), 0.85). Conclusion Though further optimization required, our powerful tool automatically draw boundaries devoid major lesions. J. Magn. Reson. Imaging 2006. © 2006 Wiley-Liss, Inc.