A Novel Hybrid Optimization Tuned Deep-Long Short-Term Memory Model in COVID-19 Chest CT Image Segmentation

Abstract

Corona viral disease (COVID-19) has been declared a worldwide pandemic due to its fast spread. In order to prevent future infection and to protect lives, early detection of COVID-19 cases is crucial because failure to do so might result in death. The segmentation of patient COVID-19 chest computed tomography (CT) image data is proposed in this study using the tuned Deep Long Short-Term Memory (Deep-LSTM) classifier by Grey Wolf Grasshopper Optimisation (GWGHO). The proposed model segmentation accuracy is enhanced by applying the recommended approach to tune the Deep-LSTM weights optimally. The suggested method is created by combining the characteristics of grasshoppers and grey wolves in such a manner as to inherit both animals' benefits in solving optimisation challenges. By applying the GWGHO algorithm's tuning procedure, the convergence is also improved. The suggested GWGHO-Deep LSTM model's efficacy is confirmed by an investigation of the model's performance indices, including segmentation accuracy, recall, precision, and Area under curve (AUC). The suggested GWGHO-Deep LSTM model's segmentation accuracy was achieved at 97.5226%, demonstrating the effectiveness of the model in segmenting data.