ABSTRACT

In the present scenario lung cancer is considered as a common disease which improves the death ratio. So detection of cause of the cancer in early stage may improve the survival chances of patients by identify the level and position (or) status of the tumor. Computed technology (CT) scan is considered as a new tool for classification of lung cancer implementing deep convolutional neural network (DCNN) with support vector machine (SVM). To indentify lung nodules which are cancerous and non cancerous in nature, extraction of deep features plays a vital role similarly reduction of dimensionality is important. Here we have used linear discriminate analysis (LDR). CT images are considered as an input to DCNN then it is classify by support vector machine classifier to achieve better accuracy. Comparative result shows the DCNN-SVM approach provides a sensitivity, specificity and accuracy of 97.27%, 93.66% and 98.67% respectively.