Hybrid RNN-FFBPNN Optimized with Glowworm Swarm Algorithm for Lung Cancer Prediction
K. Priyadarshini, Manjunathan Alagarsamy, K Sangeetha, Dineshkumar Thangaraju · IETE Journal of Research · 2023
Computer-aided diagnosis in health centres detects lung cancer at an early period. The feature set dimensions and over fitting of lung cancer features make it extremely difficult to achieve accuracy in cancer detection. To overcome these issues, in this article, a hybrid recurrent neural network and feed-forward back propagation neural network (Hyb-RNN-FFBPNN) optimized with the glow worm swarm algorithm (GWSA) is proposed for lung cancer prediction (LCP). Initially, input pictures are taken from the dataset of the National Institute of Health (NIH) chest X-ray image using the Kaggle repository for predicting lung cancer. The images are pre-processed utilizing contrast-limited adaptive histogram equalization filtering (CLAHEF) approach for removing sounds and enhancing the image quality. These pre-processed images are given to feature extraction. The empirical wavelet transform (EWT) is applied during feature extraction. Then, the hybrid FFBPNN is constructed for classifying the extracted features into benign, malignant, and normal. The proposed Hyb-RNN-FFBPNN approach does not exhibit any optimization methods adoption to determine optimal parameters and assure proper lung cancer prediction. Therefore, the hyperparameter of Hyb-RNN-FFBPNN is tuned along glowworm swarm optimization approach (GSOA), which classifies lung cancer exactly. The proposed technique is activated in MATLAB; its performance is analysed to the other existing techniques.