Advertisement
Published in:
2018 | OriginalPaper | Chapter
Knee Cartilage Ultrasound Image Segmentation Using Locally Statistical Level Set Method
Authors : Amir Faisal, Siew-Cheok Ng, Siew-Li Goh, Khin Wee Lai
Published in: 2nd International Conference for Innovation in Biomedical Engineering and Life Sciences
Publisher: Springer Singapore
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Abstract
Boundary delineation of a hypoechoic layers between the surrounding tissues and the bone structure is a necessary step in order to compute the knee cartilage thickness from ultrasound images. Speckle noise and intensity bias often complicates the segmentation task in the ultrasound images. This paper presents knee cartilage boundary segmentation using locally statistical level set method (LSLSM). Comparing to other methods in segmenting the cartilage, LSLSM produces a more satisfying outcome. Application of LSLSM on a set of 80 images illustrates a significant agreement with Cohen’s \( \kappa \) coefficient equal to \( 0.73 \) for the segmentation quality of the cartilage region rated by two raters. The quantitative evaluation measures of Dice coefficient and Hausdorff distance indicate the overall average values of \( 0.91 \pm 0.01 \) and \( 6.21 \pm 0.59 \) pixels, respectively. These good and consistent segmentation performances indicate that the segmented images can be applied for making the thickness computation in the ultrasound images.