Investigation of breast cancer detection by using artificial neural networks

Abstract

The foremost objective of this research is to develop a new strategy for detecting newlinethe breast cancer at the early stage. This is a non-invasive method using Artificial newlineNeural Network algorithms (ANN) to diagnose the cancerous lesions in the newlinemammary glands. The presence of nodules which forms a part of the mammary newlineglands may confuse the physicians to detect the breast cancer during diagnosis. An newlineinvestigation of breast cancer by using artificial neural network along with the newlineclinical inputs from medical practioners is proposed. Outcomes of the discussions newlinefrom various medical practioners and various algorithms were reviewed. Almost newlineall physicians rely on mammograms to diagnose the lesions. Due to incorrect newlineperceptions, erroneous results may be produced. Hence machine learning newlinealgorithms combined with Image Processing (IP) concepts are very much helpful newlinein this aspect. The clinical images from medical database are acquired and used for newlinetesting and training phase of the proposed algorithm. K-Means and wavelet newlinetransform algorithms are used for noise removal, edge detection, training and newlinetesting of features.The various features are Area, Mean, Standard Deviation, Mode newlineand Median combined with Mean Absolute Deviation from wavelet transform. newlineThe ANN based classification which uses all the above features proves to be an newlineefficient method of breast cancer diagnosis.