An Approach based on Convolutional Neural Network and ACO-PSO for Colon Cancer Disease Diagnosis

Abstract

The diagnosis of colon cancer has evolved into a global preoccupation, reflecting its profound impact on public health and healthcare systems worldwide. In this study, the diagnosis of colon cancer is performed using convolutional neural networks (CNN) and metaheuristic methods. Various CNN architectures, including GoogLeNet and ResNet-50, were employed to extract features related to colon disease. However, inaccuracies were introduced in both feature extraction and data classification due to the abundance of features. To address this issue, feature reduction techniques were implemented using combined Ant Colony Optimization (ACO) and particle swarm optimization (PSO). Superior convergence speed in optimizing the fitness function was observed in the case of ACO-PSO. With ResNet-50 producing 2048 features and GoogLeNet generating 1024 features, the reduction of feature dimensions proved to be crucial in identifying the most informative elements. Encouraging results were obtained in the evaluation of metrics, including sensitivity, specificity, accuracy, and F1 score, which were found to be 99.50%, 99.93%, 99.97%, and 99.97%, respectively.