BRAIN TUMOR DETECTION IN MRI USING CONVOLUTIONAL NEURAL NETWORK

Abstract

Today, the medical world is living with an enormous amount of DATA information, which has arisen from the womb of laboratory tests and clinical and physiological observations. As clinicians began to shift in clinical practice from accidental analysis and reliance on the accuracy of their observation to the analysis of various data and structured algorithms, relying on continuously updated sets of data to improve the ability to diagnose disease, or predict patient outcomes. Therefore, machine learning cannot replace the doctor, but doctors who learn and use Artificial Intelligence will replace the traditional doctors who do not keep up with the new digital revolution.

The machine learning algorithm was able to obtain a final result in the classification of medical images by convolutional neural networks. In this research, magnetic resonance images of human brain were collected to search for human tumors. we applied appropriate stages to process this images as a first step, and then used deep convolutional neural networks to extract best features from these processed images and classify them. we designed two different models by structure. Then, we trained two models and studied operands of each model loss, accuracy, f-score by Python language.