Multi Distance And Angle Models Of The Gray Level Co-Occurrence Matrix(Glcm) To Extract The Acute Lymphoblastic Leukemia (All) Images

Abstract

Acute Lymphoblastic Leukemia (ALL) is one of the most dangerous cancer diseases. Therefore, it is necessary to classify this disease accurately. We proposed a method to classify ALL using multi-distance models of the Gray Level Co-occurrence Matrix (GLCM). We employed sixteen distance models to obtain several features of the main object. We applied three channels of the image's enhancement results, where each channel has been extracted using second-order statistic models, which are homogeneity, entropy, energy, and contrast. We have obtained one hundred and ninety-two features for each object. Moreover, the feature extraction performances have been classified using Canberra and Chebyshev techniques. We have evaluated our proposed method using the ALL Image database and produced accuracy at 96.97%, where 2.27% false positive and 0.75% false negative. Our developed method has been compared to other approaches, such as Support Vector Machine (SVM)-Linear Binary Pattern (LBP), SVM Shape,Naïve Bayes Deterministic, Fuzzy based Leukemia detection, Color Correlation, Hausdorff SVM-based Leukemia detection, Automated Differential Learning Vector Quantization (LVQ), and Multi-Level Perceptron (MLP). The results show that our proposed method outperformed the others.