Segmentation and Labelling of Human Spine MR Images Using Fuzzy Clustering

摘要

Computerized medical image segmentation is a challenging area because of poor resolutionand weak contrast. The predominantly used conventional clustering techniques and thethresholding methods suffer from limitations owing to their heavy dependence on userinteractions. Uncertainties prevalent in an image cannot be captured by these techniques. Theperformance further deteriorates when the images are corrupted by noise, outliers and otherartifacts. The objective of this paper is to develop an effective robust fuzzy C- means clusteringfor segmenting vertebral body from magnetic resonance images. The motivation for this work isthat spine appearance, shape and geometry measurements are necessary for abnormalitydetection and thus proper localisation and labelling will enhance the diagnostic output of aphysician. The method is compared with Otsu thresholding and K-means clustering to illustratethe robustness. The reference standard for validation was the annotated images from theradiologist, and the Dice coefficient and Hausdorff distance measures were used to evaluate thesegmentation.