A Comparative Analysis of PGGAN with Other Data Augmentation Technique for Brain Tumor Classification

Nowadays, the number of brain tumor cases among people is increasing globally across the world due to several reasons such as obesity, overweight, excess levels of stress in life, exposure to ionizing radiation, and many more. In previous years, many investigators have provided a range of solutions and effective tools for the identification and categorization of brain tumors. Nevertheless, the existing developed models for brain tumor identification and categorization have diverse limitations such as minimal accuracy and precision values. In this paper, the authors developed a novel model for the comparative analysis of the Progressive Growing-Generative Adversarial Network (PGGAN) with other data augmentation techniques for brain tumor classification. Because of the availability of finite datasets, the brain tumor classification algorithm along with the convolutional neural networks (CNNs) must be enhanced to be more competent for brain tumor classification and identification in real-time diagnosis. The outcome of the proposed model demonstrates that PGGAN delivers higher accuracy, as well as precision, and the Recall with the F1 score is 99.22%, 98.11%, 98.66%, and 97.45%, respectively. In the future, the developed model performance could be measured with other data augmentation techniques for larger datasets for performance constraints computations for further study and implementation of the model for real-time diagnosis of the patients.