Deep Learning Techniques for Detecting of Lun...

Deep Learning Techniques for Detecting of Lung Cancer and Prediction of Life Expectancy Post Thoracic Surgery

Author Name : Prof. Salma Itagi, Keerthitha, Mamatha N, Priyanka D R

ABSTRACT

The early detection of Lung Cancer is important for the diagnosis process and it gives the higher chances for successful treatment. Therefore, the objective of this project is to develop a system that detects lung cancer and further predicts the life expectancy post thoracic surgery for the lung cancer detected patient. The input given to the system is the Computed Tomography (CT) scanned lung images on which the analysis is done to predict the presence of nodules in the lungs. Image processing techniques and machine learning techniques are used in this project to accomplish the objective. The system performs three major tasks which include Image Processing, Feature Extraction and Classification. The image processing involves the extraction and segmentation of the input CT image using image processing techniques. To select and extract the useful features of the segmented image, Genetic Algorithm is used. The Classification task uses a classifier which is modelled using the Convolution Neural Network (CNN) that will classify the input image as cancerous depending upon the features obtained from the segmented image. Once the cancer is detected, the thoracic surgery is one of the options for the treatment of Lung Cancer. The project also involves the analysis of the patient’s dataset who underwent thoracic surgery and an attempt is made to model a classifier that will predict the survival of the patient post-surgery. The model is implemented using MLP algorithm. The project can be considered as a promising tool to support the medical specialists to make a more precise diagnosis and prognosis concerning the Lung nodules.

Keywords— Computed Tomography (CT), Image Processing, Feature Extraction, Classification, Genetic Algorithm, Convolution Neural Network, MLP algorithm.