The Lung Cancer Predictive Accuracy for Non-Smokers Using Classification and HFPS Algorithm

Abstract

Lung cancer identification and projection at the earliest point of comparison may be really helpful to boost the patient's survival rate. But cancer detection and prediction remains one of radiologists' toughest obstacles. An intelligent, computer-assisted diagnostic device may be very helpful for radiologists in identifying, forecasting and diagnosing lung cancer. This paper proposed a Hybrid algorithm as Firefly–Particle Swarm Algorithm (HFPS) for feature selection process. In this study, we conduct the simulation by taking the LIDC-IBRI dataset. Initially the raw data is pre-processed by using Laplacian filters. The segmentation of images by using the modified KFCM techniques. Then extraction of feature by using transformation techniques as DWT. After completion of feature selection and extraction then we classified the data as cancer or normal. By this classification process is successfully done with the help of deep learning techniques as recurrent neural networks, Auto Encoder and LSTM algorithms are used. In this study we compare the analysis by using three different classifier with input of feature selection method and without that. In our model, we achieved the successful classification rate of HFPS Feature Selection with RNN classifier achieved 97.43% of accuracy. In, these methods used to detect the lung cancer in early stage to reduce the mortality rate.