Abstract:
Lung Cancer is chronic disease causing death for human beings in almost all over the world. The identification of disease nature and response to treatment differs widely among patients. Thus it is very important to detect diseases like lungs cancer in its early stage. Cancer treatment can be successful if it is not too expanded and detected in its very initial stages. Cancer detection is tough when it spread over lungs and other organs. Most of the lung cancers are Non-Small Cell Lung Cancer (NSCLC) which is 84 % of all types of lung cancer. More than 81% patients of lungs cancer can survive for a year or more if diagnosed at early stage as compared to other 15% diagnosed with advanced stage of the lungs cancer. In this study we have proposed an intelligent system for lung cancer detection. The proposed system is composed of various stages which include Preprocessing of the subject images and segmentation. U-Net Model is being used for segmentation also. Pretrained model are used for classification and detection purpose. The proposed system uses 70% of the images including healthy and unhealthy lung images for training while30% of the images are used for the testing. The proposed system accurately detect lung cancer in 91.63% of the subject images and Results are encouraging and gives promising direction for future work.
