Artificial Bee Colony-based General Adversarial Network for Liver Cancer Detection using CT Images

Nowadays, liver cancer is the foremost cause of death and the Computed Tomography (CT) images are widely utilized for detecting liver cancer detection. However, the manual classification and detection of liver tumor based on the CT scan images is the challenging one. In this paper, the Artificial Bee Colony based General Adversarial Network (ABCbased GAN) is developed for detecting liver cancer using CT images. The developed model mainly consists of four stages, namely liver extraction, cancer lesion segmentation, feature extraction, and classification. Initially, the input liver CT images are fed to the liver extraction stage where the active contour method is employed for extracting the liver region from CT images. After that, the cancer lesion regions are segmented from the extracted liver images using the Markov Random Field (MRF). Once the cancer lesions are segmented, then, the feature extraction process is done based on statistical features, like mean, standard deviation, skewness, and kurtosis. Finally, the cancer classification is performed based on the extracted features using the GAN classifier where the ABC optimization algorithm is utilized to train the GAN classifier. Furthermore, the developed method achieved maximal accuracy of 95.16%, maximal sensitivity of 91.88%, and maximal specificity of 98.42%.