Morphological and Textural Data Fusion for Breast Cancer Classification Based on Inter and Intra group Variances
International journal of intelligent engineering and systems · 2024
Nowadays, the most predominant cancer disease is Breast Cancer that has a higher death rate and women gender is the most affected by this disease.But detecting Breast Cancer in early stage is challenging as the malignance growth at this stage occurs in the duct that are undetected as symptoms are less.This paper addresses the challenge of early detection of Breast Cancer cells by proposing the fusion scheme of morphological and texture features of the cells for analysis.Morphological features such as the shape and marginal characteristics of the mass are considered as per the Breast Imaging Reporting and Data System (BI-RADS) standard.Texture features of the mass were also extracted to understand the characteristics of pixel variation in the masses.These features are combined and its dimension is normalized using Exhaustive Feature Selection (EFS).The accuracy of the proposed feature on the INbreast dataset is 94.75% on an average.The accuracy for the Curated Breast Imaging Subset of Digital Database for Screening Mammography (CBIS-DDSM) Calc RoI dataset is 95% and for CBIS-DDSM Mass RoI dataset it is 94.5%.The result is further compared with contemporary methods and found that the fused feature is performing well.