A Literature Survey on Early Detection of Malignant Tumor Through Computerized Technique

Abstract

Computer based initial stage classification and detection of breast cancer tumor must be taken as one of the most important parameter to detect cancer in order to assist radiologist. In this paper an overview of current development of computer based detection system has been reviewed. The abnormalities identification and classification based on texture feature and shape of tumor, statistically diagnostic parameters (mean, mode, standard deviation, median etc.), Entropy based, Elliptical Normalized Circumference (ENC), C-means, K-means, Watershed Transformation, Ratio of Perimeter:Area (P :A), Ratio of major axis:minor axis (L:S) and Content Based Image Retrieval (CBIR) have been presented. In this paper the accuracy of the result obtained by various techniques such as texture based classification (80%), statistical parameters based classification (70%), shape based classification (86%), CBIR based classification (60%), etc. are presented. The various imaging techniques and dedicated databases publicly available for the cancer research work have been listed, discussed and reported in this paper.