Comparative study of four novel approaches developed for early detection of breast cancer and its stages

This paper aims at describing and comparing four different approaches used for developing Computer Aided Diagnosis systems that aims to detect, classify and identify the stages of malignant tumors in MRM images. Approach I uses a hybrid combination of techniques for detection of Region of Interest (ROI) and size of the tumor is used for detecting the stage of classified malignant tumor. Approach II uses modified approach for segmentation and area of the tumor is used for detection of stages. Approach III uses an advanced levelset approach for detection of ROI followed by energy and entropy features being used for identification of stages of malignant tumor. Finally the main approach, namely approach IV has been developed using a very novel algorithm for detection of ROI and a very novel approach for detection of stages of malignant tumor. The approaches I and II are applicable only for Mammographic Image Analysis Society (MIAS) database images. Approaches III and IV are applicable for both MIAS and Real time hospital images. This paper also provides a complete analysis of performance of all the four approaches using various parameters.