Osteoporosis Recognition Based on Similarity Metric with SVM

Abstract

The purpose: Applying different techniques of classification to osteoporotic bone tissue texture analysis, exploring the recognition rate of the different classification methods. Methods: Using gray-level co-occurrence matrix (GLCM) and running a length matrix texture analysis to extract bone tissue slice image characteristic parameters, and to classify respectively 4x and 10x microscope images of the two groups: the sham (SHAM) and the ovariectomized (OVX) group image. Results: The metric support vector machine (SVM) classification algorithm, based on SVM learning or recognition rate, was higher than the stand-alone measure, and the classification results were stable. Conclusion: Measurement of the SVM classification algorithm for osteoporotic bone slices texture analysis revealed a high recognition rate.

Keywords

• Osteoporosis
• Texture analysis
• Information-theoretic metric learning (ITML)
• Support vector machine (SVM)