Diagnosing Brain Tumors Using a Super Resolution Generative Adversarial Network Model

2. Literature Review

Anton-Rodriguez et al. (2021) presented a fully automatic brain tumour classification technique using Convolutional Neural Network, and their approach was multiscaled. The images supplied in input had three spatial scales, and their processing also included different pathways. Their method was able to classify tumours with an accuracy of 0.973, which was higher than their compatriots. Badza et al. (2020), exclaimed that machine learning could help the radiologist in tumours detection, so they presented a CNN architecture. They evaluated their performance using four different approaches consisting of possible combinations of 10-fold cross-validation and used two different databases. In doing so, they also achieved an accuracy of around 0.965. According to Anaraki et al. (2019), Gliomas are the most common type of tumour in the brain. So their early detection is always important, so they deployed a combination of genetic algorithms and convolutional neural networks. Hence, the architecture of CNN is evolved using GA, and their accuracy came around 0.942.