Semi-automated vertebral segmentation of human spine in MRI images

Image segmentation is an important task in medical image processing to assist physicians or radiologists in making faster and effective diagnosis and treatment. However, there is still lack of effective segmentation methods to extract human spine in Magnetic resonance imaging (MRI) images. In this study, we propose a segmentation method using 12-anatomical point representation (12-APR) method for human spine vertebra. The proposed method is a semi-automatic segmentation in which 12-points will be manually annotated on the region of interest (ROI) before the ROI can be extracted automatically. The performance of this segmentation is evaluated using six performance metrics and the results show that the proposed method gives the highest accuracy (99.87%), specificity (99.89%), Dice similarity coefficient (94.04%), Jaccard similarity coefficient (88.81%) and Cosine similarity coefficient (94.14%).