Detection of Cancer in Lung CT Image Using 3D CNN

Abstract

The use of image processing techniques to analyze CT scan pictures including lung cancer cells is gaining popularity these days. Lung illnesses are diseases that damage the lungs and weaken the respiratory system. Lung cancer is one of the top causes of death in individuals around the world. Humans have a better chance of surviving if they are detected early. The average survival rate for persons with lung cancer increases from 14 to 49 percent if the disease is detected early. While computed tomography (CT) is significantly more effective than X-ray, a full diagnosis requires the use of numerous imaging techniques to complement one another. A deep neural network is constructed and tested for detecting lung cancer from CT images. This strategy is more about diagnosing at ahead of schedule and critical stages with keen computational procedures with different noise is expected to be eliminated by detaching the CT images and calculating the survival rate which is the root idea of digital image processing. A highly linked convolution neural network was used to classify the lung image as normal or cancerous. A dataset of lung pictures is employed, with 85 percent of the photos being used for training and 15% being used for testing and classification.