Analysis and Classification of the Lung Cancer with CNN Implementation

A Lung malignancy is considered to be one among the prominent cause of disease and mortality in many countries, and radiologists face a difficult burden in diagnosing the disease early on. Observing, analyzing and medication of lung cancer has been probably a great trouble for the physicians over decades. Thus early detection of a tumor would encourage in saving an immense count of lives over the world reliably. Also early detection of lung nodules prevents the patient from meta-staging nodules. The existing image processing and machine learning techniques consume more execution time and are expensive. In our proposed system; the human lung CT scans image is given as input to the preprocessing stage. Binarization is applied to the pre-processed image to transform the complete binary image and equate it with the threshold value for detecting lung cancer. The lung CT scan image is then segmented, and each component of the segmented photos is familiarized with a solid element extraction approach. This methodology uses a Convolution Neural Network (CNN) to arrange the tumor cells identified in the human lung as threatening (malignant) or generous (benign). Thus the proposed method includes the exactness acquired by using CNN is 95%, which is highly effective when contrasted with precision obtained by the traditional neural system frameworks.