Deep Learning Technique-Based 3d Lung Image-Based Tumor Detection Using segmentation and Classification

Abstract

In the whole world, one of the most dangerous diseases that leads to death is Lung cancer. Tumor position is find by a Computed Tomography (CT) scan and tumor level in the body is also identified by this scan. An innovative automated diagnosis classification approach for CT lung images are presented in this study. This research proposed novel technique in segmenting and classifying the lung tumor using deep learning architectures. The input has been taken as CT images of lung and processed for noise removal and image resize. Lung image has been segmented using feed forward neural network (Fe_FNeuNet) where the image is trained and predict the presence of tumor. Then this segmented and trained image has been classified and their features are extracted using 3D Pre-trained deep convolutional neural network (3D-PrDConvNN). The results of comparison proved that the proposed classifier achieves the accuracy 98%, precision of 94.9% and recall of 96% and F-1 score of 95%.