Spatial and Hierarchical Feature Extraction Based on Sift for Medical Images

Image segmentation plays a major role in the analysis of medical image disease diagnosis. Image features extracted is the basis for precise image segmentation. Variable nature of image features, such as size, shape, intensity, color, texture etc., cause complexity in the image segmentation and analysis of the image nature. Existing work estimate the effectiveness of the level-set shape beside with fractal texture and intensity features to discriminate PF (Posteriorfossa) tumor from other tissues in the brain. In addition explore the effectiveness of the skin texture with feature ranking and selection strategies such as entropy and KLD. Segmentation based on graph cut and expectation maximization (EM) for PF brain tumor segmentation is also done with the chosen features. Further, presented features may not be adequate to differentiate amongst the medical images. To overcome the issue, we present a Feature Rich Image Segmentation and Multi-Classifier Framework for Medical Image analysis and disease diagnosis. In this work, we first employ, unsupervised learning model to extract features from the medical images using spatial and hierarchical structures based on scale invariant feature transformation. This would enrich the features extracted from the medical image for segmentation compared to the existing method features of intensity, FD and shape model. The experimental performance is evaluated with benchmark data sets extracted from research repositories of both real and synthetic data sets. The performance parameter used for the analysis of the proposed spatial and hierarchical feature extraction based on sift for medical images are Feature size, Dominant Feature Threshold, Spatial feature size.