Feature Space-Based Mammogram Image Analyse Using Decomposition Method

This paper has focused on detecting breast tumors from mammograms using the Fourier decomposition approach with enhancing feature space. Detecting breast tumors from mammograms is a challenging task without regional feature space. Thus, the texture features are considered for analysis of regional features for breast tumors from mammogram images. The Fourier decomposition method-based framework is considered for texture features in mammogram images. These texture features collect imperfections, roughness, and smoothness from the above framework. Further, the linear regression method enhances the feature space compared to old feature space towards effective classification. The impacts on classification performance are examined after producing an enhanced feature space based on linear regression and ensemble it with the original feature space. The suggested framework achieves sensitivity (98.48%), specificity (99.74%), accuracy (99.06%), and area under the curve (0.99). Investigating the tiny image dataset also provides the suggested framework's resilience.