Segmentation and Prediction of Lung Cancer CT Scans through Nodules using Ensemble Deep Learning Approach

Lung cancer is considered as one of the most dangerous disease which world is facing today. Detecting the malignant nodules in the lungs is a difficult task and to perform the detection various methods has been implemented. Previous work done in this area is mostly based on convolutional neural network without masking. In this study, deep learning-based approach is implemented in which lungs CT scan images are segmented using U Net models and convolutional neural network is used for the prediction of cancer disease. In this study, the images are processed and the masks are generated. Convolutional neural network is used to reduce the false positives, so that lung nodules can be classified more accurately. The proposed model has achieved the sensitivity of 90.83%, specificity of 94.28%, precision of 93.1% and the accuracy of 91.23%. The work has shown better performance than the other existing algorithms for the prediction of lung cancer through the CT-SCAN images. The objective of the proposed model is to properly classify the CT scan images through lung nodules for the prediction of cancer.