Abstract:
Brain tumors are a significant pathological condition within the field of medicine that poses
challenges in terms of accurate diagnosis. In recent times, the utilization of deep learning
techniques has been employed to create automated systems aimed at classifying brain tumors
based on MRI images. The present study aimed to assess the efficacy of eight distinct deeplearning
models in the classification of brain tumors. The models underwent training using a
dataset consisting of 4,483 MRI images. Subsequently, their accuracy was assessed by
employing a separate set of 1,250 test images. The findings indicated that the MobileNet model
exhibited superior performance, achieving an accuracy rate of 99.38%. The remaining models
also exhibited strong performance, achieving accuracies within the range of 97.68% to 98.39%.
The findings of this study indicate that employing deep learning techniques to automate the
classification of brain tumors using MRI images is a promising and advantageous strategy.
Keywords: MRI images; Brain tumor; MobileNet; Deep learning
