Ultrasound estimation of prostate volume using the Frustum model

Prostate cancer threatens human life around the world, which leads researchers to focus on improving prostate image quality for diagnosis, depending on medical image segmentation to calculate the accurate prostatic volume. This study aims to highlight an easy, simple, and fully automated segmentation algorithm for segmenting the prostate gland, these algorithms were Level set and Active contour, then we applied a hybrid technique to improve the segmentation quality, and performance. This research additionally aims to find an alternative approach for prostate volume calculation besides the classic method the prolate elliptical method; we used a novel model frustum model which measures prostatic volume using only one image instead of two views. Findings showed that the segmentation process had been improved by using a hybrid technique. The highest performance value of the 0.8 thresholds and DSC 0.9-1 was (48.71 %) for the homogenous images from the GHHS database, while for the 0.9 thresholds and DSC 0.8-1, the highest value was (90%). In contrast, the maximum performance value of the 0.8 thresholds and DSC 0.9-1 for the Erbil database (was 36%), meanwhile for the 0.9 thresholds and DSC 0.8-1 was (92%). On the other hand, the statistics for the proposed volume estimation model show a small difference between the mean values of both techniques (0.662) with P – value equal to (0.854), with R2 equal to (0.982) respectively which indicates that the result of both methods is non – significant statistically. We conclude that prostate US segmentation can be improved by selective optimization of the US image quality which can be fit to clinical diagnosis and segmentation, also frustum model is an acceptable way to calculate prostate volume.

Text FPSK (p) 2024 18 - Declaration Form.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (541kB) Text FPSK (p) 2024 18 - Full Text.pdf Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (4MB) Text FPSK (p) 2024 18.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (4MB)

Actions (login required)

View Item