Employing Novel Deep Learning Techniques to Identify Different Forms of Brain Tumors

Abstract

The ratio of specialist doctors such as hematologists and oncologists and the number of people per doctor is 1:20,366, which exists the developed countries like America. Then imagine how much it would be in the entire world. The medical industry has progressed tremendously over the past decades. Specifically, visual and image recognition is being used for many purposes in many fields very actively. The general Artificial Intelligence (AI) topics, such as Neural Networks (NN) and related concepts, have also gained more popularity in the last decades. This article proposes a Convolutional Neural Networks (CNN) architecture from scratch with data augmentation, image processing approaches, and neural network pattern recognition. Our proposed CNN model is compared with the existing architecture VGG-16. The comparison result states that the proposed CNN model is more suitable for detecting brain tumors. Although the data we have used to train the model is comparatively tiny compared to any industrial application database, the proposed model displayed better accuracy and results than the VGG-16 for the brain tumors detection problem.

Moreover, the proposed model uses less computational and memory power than the VGG-16 model. Secondly, the proposed CNN model reduces dataset usage from unknown sources. This paper shows how we can take the user consent and store the data to build an accurate data set for future educational and medical researchers and retrain the models for better results.