Electronic Thesis/Dissertation

 

Image Segmentation Of Vascular B-Mode Ultrasound Images For Determining Carotid Plaque Area Open Access

Downloadable Content

Download PDF

The principle objectives of this research were to perform image segmentation on vascular B-mode ultrasound images, to segment and partition the plaque deposits within the image, to accurately measure the plaque dimensions, and to determine if this method was a viable method for determining and monitoring plaque size. This research focused on determining if image segmentation could be performed on vascular B-mode ultrasound images of the carotid arteries to partition plaque deposits from the surrounding blood and tissue, and determining the plaque dimensions.Current methods for grading carotid artery stenosis are based on blood flow velocity and not on actual measurements of plaque size. This research may provide physicians with a method to accurately quantify the amount of plaque in the peripheral arteries, while also providing a more accurate method for monitoring the effects that exercise, diet, and medication have on plaque size within the carotid arteries. The clinical B-mode ultrasound images used in this research were acquired with the help of Dr. Lisa Martin, and access to patient data was granted through the IRB approval process. The data included 40 data sets of patients with diseased carotid arteries and 20 data sets of patients with normal carotid arteries. The image with the best visual representation of plaque edges was chosen from each data set, and 60 images were analyzed in total. Out of those 40 images represented diseased carotid arteries and 20 represented normal carotid arteries. The average area of plaque determined for diseased carotid arteries was 0.126 cm squared with a standard deviation of ±0.083. The average area of plaque determined for normal carotid arteries was 0.057 cm squared with a standard deviation of ±0.025.

Author Language Keyword Date created Type of Work Rights statement GW Unit Degree Advisor Committee Member(s) Persistent URL
License

Relationships

Items