Standard Deviation Based Otsu for Breast Cancer Classification
ASM Science Journal · 2020
Segmentation is one of the essential components in Computer Aided Diagnosis (CADx).This paper proposed a modified Otsu method for segmentation phase in classifying breast cancer.In Otsu's method, the process of computation within-class variance for two classes produce lower accuracy for the classification, hence the standard deviation will be utilized to overcome this limitation.After that, the features of the segmented images will be extracted by using Speed Up Robust Features (SURF).Finally, the corresponding features will be classified using several classifiers.It is found that the Support Vector Machine (SVM) classifier shows the highest accuracy rate among other classifiers.Moreover, experimental results show that the proposed method is superior to the original Otsu method.