Lung Cancer Prediction and Optimization Using Improved Vgg16 With Grey Wolf Optimizer (Gwo)

Abstract

This paper suggests a strong deep learning model for predicting lung cancer utilizing an Improved VGG16 structure and the Grey Wolf Optimizer (GWO). We think that this method will improve the accuracy of classifications and lower the number of wrong forecasts. The chest CT scan photographs taken by the system show a lot of different kinds of cancer, including adenocarcinoma, squamous cell carcinoma, large-cell carcinoma, and normal cells. Feature extraction is carried out with VGG16 convolutional layers and model parameter optimization is attained with grey wolf social behavior simulation in GWO. The model stabilizes and avoids overfitting with batch normalization and dropout layers. The model yields enhanced prediction performance with 92.31% sensitivity, 98.29% specificity, and 94.23% accuracy. Comparison with other models such as DNN, RBF SVM, Adaboost, and DCNN also indicates better performance in all the metrics. Experimental analysis performed using MATLAB verifies the efficacy of the hybrid system through performance curves and convergence patterns. The model also exhibits fast convergence during training and low error rates when utilizing a couple of iterations. Plots indicating model accuracy, loss patterns, and relative metrics further demonstrate its superiority. Grey Wolf Optimizer helps in avoiding the local optima and increasing the capacity for generalization. The contribution reflects the clinical importance of the proposed model for early detection, allowing timely treatment. Its capacity to distinguish between cancer and non-cancer images guarantees proper diagnosis. The system overall offers a scalable and understandable AI algorithm for real-time lung cancer detection. The proposed hybrid model sets a promising benchmark for future medical image analysis software.