Hyperband-Optimized Convolutional Neural Network Model for Efficient Brain Tumor Classification and Prediction

Human life depends heavily on health. Since the brain is the vital, distinctive and invaluable organ, its health is very crucial. A group of abnormal cells in the brain that might spread to other tissues and endanger life that is called a brain tumor. For treatment planning to be effective, a precise diagnosis is necessary. Like other organs, the health of the brain can also be analyzed with magnetic resonance images (MRI) for the purpose of diagnosis. Artificial intelligence can handle huge amounts of data. There are so many researchers who have developed various deep learning based models for addressing the brain tumor prediction issues. The model may take more no. of parameters and may lead more time to training the model. For addressing this issue, we have proposed the HB-optimized CNN model which automatically selects optimal hyperparameters, fine-tuning the CNN to achieve high prediction accuracy with minimal computational overhead. For evaluating the proposed approach, we have collected 3,000 MRI images from a publicly available dataset and the experiments are carried out using different optimization parameters. Lastly, experimental findings showed that our suggested model outperformed the conventional CNN, AlexNet, and VGG-16 Net models, achieving 88.23%, 91.15%, and 93.40%, respectively.