Automated Vertebrae Extraction Using Watershed Segmentation and Tree-based Modelling Approach

Abstract

Vertebra segmentation and extraction are essential steps for automated vertebrae fracture assessment system. This study focuses on vertebra segmentation, which is a challenging task due to high noise and poor contrast of X-Ray images. The objective of this study is to develop and test an automated cervical vertebrae extraction to facilitate the medical researchers and physicians. The novelty of the proposed method is the combination of watershed segmentation and tree-based modelling algorithm. The proposed method has been successfully applied to a set of 50 cervical X-Ray images. Five performance metrics are calculated, which are accuracy, sensitivity, specificity, correlation and error where the best obtained results are 0.9886, 0.0655, 0.99, 0.7246 and 0.01135, respectively.