Extraction of the Longitudinal Movement of the Carotid Artery Wall using Consecutive Ultrasonic Images: a Block Matching Algorithm

Document Type : Original Paper

Authors

1 Ph.D. in Medical Physics, Medical Physics Dept., Tarbiat Modares University, Tehran, Iran

2 Prof., Medical Physics Dept., Tarbiat Modares University, Tehran, Iran

3 Associate Professor, Radiology Dept., Tehran University of Medical Sciences, Imaging Center of Imam Khomaini Hospital, Tehran, Iran

4 Ph.D. Student in Medical Physics, Medical Physics Dept., Tarbiat Modares University, Tehran, Iran

Abstract

Introduction: In this study, a computer analysis method based on a block matching algorithm is presented to extract the longitudinal movement of the carotid artery wall using consecutive ultrasonic images. A window (block) is selected as the reference block in the first frame and the most similar block to the reference one is found in the subsequent frames.
Material and Methods: The program was applied to the common carotid artery of ten healthy subjects and the longitudinal movement of the arterial wall was extracted in three cardiac cycles. Manual measurements were carried out to validate the automatic method. The maximum, minimum and average longitudinal displacements of the arterial wall were extracted from the automatic method and compared with the manual measurements, with 10 mm displacement of two blocks in first frame.
Results: Paired t-test analysis showed no significant differences between the automatic and manual methods (p>0.05). There was significant correlation between the changes in the instantaneous longitudinal movement of the common carotid artery measured with the manual and automatic methods (correlation coefficient 0.97, p<0.05). Bland-Altman analysis with the agreement region of 0.03 to 0.17 mm and difference average of 0.10±0.03 mm indicated good agreement of the two methods.
Discussion and Conclusion: This study shows that the block matching algorithm can automatically extract the longitudinal movement of the arterial wall from consecutive ultrasonic images.

Keywords

Main Subjects