Hybrid Concatenation of Emphysema Features for Improving Classification in Computed Tomography Images
Authors: Ibrahim MA, Ojo OA
Previous studies have recently demonstrated the effectiveness of multifractal based methods for the classification of histo-pathological cases by calculating the local singularity coefficients of an image using different intensity measures. In this paper, we propose to improve on the existing results on multifractal techniques by investigating the features from the combination of alpha-histograms and multifractal descriptors in the classification of Emphysema in computed tomography (CT) images. The performances of the classifiers were measured by using classification accuracy (error matrix) and area under curve (AUC). The experimental results compared favourably with the local binary patterns (LBP) approach, a state-of-the-art measure for pulmonary Emphysema. The results also showed that the proposed cascaded approach significantly improved the overall classification accuracy.
Affiliations: Department of ICT, Faculty of Basic and Applied Sciences, Osun State University, Osogbo, Nigeria.
Keywords: Emphysema Classification, Multifractal Analysis, Multifractal Spectrum, Histogram Comparison, Statistical Self-similarity
Published date: 2020/12/30