Date of Award
2006
Degree Type
Thesis
Degree Name
Master of Engineering Science
Program
Biomedical Engineering
Supervisor
Dr. Aaron Fenster
Second Advisor
Dr. Hanif Ladak
Third Advisor
Dr. Deidre Batchelor
Abstract
Segmenting the prostate boundary is essential in determining the dose plan needed for a successful brachytherapy procedure - an effective and a commonly used treatment for prostate cancer. However, manual segmentation is time consuming and can introduce inter and intra-operator variability. In this thesis, we describe an algorithm for segmenting the prostate from two dimensional ultrasound (2D US) images, which can be either semi-automatic, requiring only one user input, or fully-automatic, with some assumptions of image acquisition. Segmentation begins with the user inputting the approximate centre of the prostate for the semi-automatic version of the algorithm, or with assuming the centre of the prostate to be at the centre of the image for the fully-automatic version. The image is then filtered with a Laplacian of Gaussian (LoG) filter that identifies prostate edge candidates. The next step removes most of the false edges (not on the prostate boundary), and keeps as many true edges (on the boundary) as possible. Then, domain knowledge is used to remove any prostate boundary candidates that are probably false edge pixels. The image is then scanned along radiai lines and only the first-detected boundary candidates are kept. The final step involves the removal of some remaining false edge pixels by fitting a polynomial to the image points, removing the point with the maximum distance from the fit, and repeating the process until this maximum distance is less than 4mm. The resulting candidate edges form an initial model that is then deformed using the Discrete Dynamic Contour (DDC) model to obtain a closed contour of the prostate boundary. The accuracy of the prostate boundary that was produced by both versions of the algorithm was evaluated by comparing it to a contour that was manually iii outlined by an expert radiologist. We segmented 51 2D Transrectal ultrasound (TRUS) prostate images using both versions of the algorithm and found that the mean distance between the contours produced by our algorithm and the manual outlines was 0.7 ± 0.3 mm for the semi-automatic version and 0.8 ± 0.4 mm for the fully- automatic version. The accuracy and the sensitivity of the algorithm to area measurements were (94.3 ± 4.2)% and (92.1 ± 3.6)% for the semi-automatic version, respectively and (92.9 ± 6.9)% and (91.2 ± 5.1)% for the fully-automatic version, respectively.
Recommended Citation
Saikaly, Manale, "AUTOMATIC PROSTATE BOUNDARY SEGMENTATION FOR 2D ULTRASOUND IMAGES" (2006). Digitized Theses. 4628.
https://ir.lib.uwo.ca/digitizedtheses/4628