Segmentation of Cortical and Cancellous Bone with Osteogenesis Imperfecta using Thresholding-based Method

Abstract

Osteogenesis Imperfecta (OI) is a genetic bone disorders that mainly affect the bones which commonly leads to the multiple fractures. The purpose of this study is to segment the cortical and cancellous bone of tibia affected with the OI from the CT images. This project consists of two sections, segmentation of the cortical and cancellous bone, and the evaluation of the image performance. Contrast adjustment was implemented to enhance the contrast of images. K-Means and multi-threshold were implemented to segment the cortical bone at the proximal, diaphysis, and distal tibia. Post-processing was applied to further refine the segmented images. All segmented images were then evaluated by using the ground truth images. From the results, the process with contrast adjustment and multi-threshold obtained the highest accuracy and Dice-coefficient of 99.91% and 81.42% respectively at the proximal region while 99.93% and 73.88% at the distal region. However, pathway that make use of K-Means performed best in diaphysis region. This method has obtained the highest accuracy and Dice-coefficient of 99.98% and 95.59% respectively.