J Korean Soc Med Ultrasound.
2004 Jun;23(2):105-111.
Effective Computer-Aided Diagnosis Analysis for the Plaque Measurement on the Ultrasound image of the Carotid Artery
- Affiliations
-
- 1Center for Computer Graphics and Virtual Reality, Ewha Womans University, Korea. mhkim@mm.ewha.ac.kr
- 2Department of Radiology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Korea.
Abstract
- PURPOSE
To evaluate the accuracy and utility of using Computer Vision techniques such as image filtering and image segmentation instead of using manual method for measuring plaque of the carotid artery on ultrasound examination. MATERIALS AND METHODS: The image filtering called Speckle Reducing Anisotropic Diffusion (SRAD) was used for ultrasound images of three patients with plaques on their carotid arteries. After filtering, segmentation was executed using the expanding plaque region, which based on the computer operator inputting the seed point. Finally, the post-processing was performed to complement the measured result. These experiments were processed using the Visual C++ program on a personal computer (Pentium 2.4, 512 ram, NVIDIA GeForce4MX440). The numbers of SRAD processing, the threshold values for the expanding plaque region, and the numbers of interaction for segmentation were examined. The computer-aided diagonosis measures real shape of the plaque more correctly than the manual method. RESULTS: In this research, SRAD was repeatedly done over 100 times for efficient image filtering, and the appropriate results came out in the case of 0.25 or 0.3 being inputted as the initial value, on the average, although the initial values were different according to the images. The threshold value for region segmentation was 70. Although the area of extracted plaque was different according to the character of the image, the appropriate results were derived when the user did 4 interactions. CONCLUSION: This research has shown that the more objective and accurate measurement of plaque was possible using image filtering and segmentation on a computer than by the manual method by clinicians. The current plaque measuring method allows the clinician to get the slick shape of ellipsoid but this new method enables the physicial to characterize the plaque according to shape. Therefore it could be efficiently used in the test and diagnosis.