Classification of Breast Cancer with Improved Self-Organizing Maps
Hicham Omara, Mohamed Lazaar, Youness Tabii · 2017
Over the past few decades, extensive death of women due to breast cancer has precipitated the need for the best classification of breast cancer. Classification of breast cancer can help in choose the appropriate and the convenient treatment. Many researchers have studied the breast cancer using artificial neural network model (ANN) for his ability to visualize high-dimensional data. The use of learning machine and artificial intelligence techniques has revolutionized the process of diagnosis and prognosis of the breast cancer. However, there are still some problems applying ANN algorithm such as longer training time and lower classification accuracy. To overcome these problems, the SOM model based on Distance travelled by neurons (DSOM) has been proposed, using the Wisconsin diagnostic breast Cancer datasets (WDBC) to distinguish between different types of breast cancer. The data set consists of nine attributes that represent the input layer to the neural network. The neural network will classify the input vectors into two classes of cancer type (benign and malignant). The proposed approach tested on the database, resulted in 97 % succession rate of classification. We can concluded that our approach seems an efficient method to classify in medical applications and especially for the breast cancer classification.