Texture classification of the ultrasonic images of rotator cuff diseases based on radial basis function network
ABSTRACT This article studies the usages of texture analysis methods to classify ultrasonic rotator cuff images into the different disease groups that are normal, tendon inflammation, calcific tendonitis and tendon tear. The adopted texture analysis methods include the texture feature coding method, gray-level co-occurrence matrix, fractal dimension and texture spectrum. The texture features of the four methods are used to analyze the tissue characteristic of supraspinatus tendon. The mutual information feature selection and F-scoring feature ranking method are independently used to select powerful features from the four texture analysis methods. Furthermore, the trained radial basis function network is used to classify the test images into the ones of four disease group. Experimental results tested on 85 images reveal that the classification accuracy of proposed system can achieves 84%.
- SourceAvailable from: Nojun Kwak[Show abstract] [Hide abstract]
ABSTRACT: Mutual information is a good indicator of relevance between variables, and have been used as a measure in several feature selection algorithms. However, calculating the mutual information is difficult, and the performance of a feature selection algorithm depends on the accuracy of the mutual information. In this paper, we propose a new method of calculating mutual information between input and class variables based on the Parzen window, and we apply this to a feature selection algorithm for classification problems.IEEE Transactions on Pattern Analysis and Machine Intelligence 01/2003; 24(12):1667- 1671. · 4.80 Impact Factor