Brain Tumor Classification Using Convolutional Neural Network and Deep Transfer Learning Approach with MR Imaging

EasyChair Preprint 8995

Abstract

Brain tumors are one of the leading causes of death worldwide. It is a challenge to solve brain tumor segmentation and classification using only traditional medical image processing. a difficult and complex task In fact, medical research suggests that manual classification with human help may result in inaccurate prognosis and diagnosis. This is primarily owing to the similarities and contrasts between normal tissues and malignancies. Recently, deep learning approaches have demonstrated promising outcomes. The purpose of this study is to develop an effective strategy for identifying brain tumors using MRI that is based on transfer learning. This article employs the Convolutional Neural Network (CNN) and popular deep learning models to classify a brain tumor diagnosis system. Using pre-trained models such as Efficientb0, DenseNet121, and Densenet169. For the prediction of a brain tumor, the accuracy is achieved at 93% through CNN. Deep learning models have achieved accuracy efficientb0 of 48%, DenseNet121 96%, and DenseNet169 98% these models are Ensemble, and a 98 % accuracy level is attained Experimental findings suggest that an Ensemble of deep characteristics can significantly increase performance.

Keyphrases: Brain Tumor, DenseNet, Efficientb0, Transfer Learning, deep learning, ensemble learning

@booklet{EasyChair:8995,
  author    = {Insha-Ul Hassan and Tariq Ali and Ghulam Ali},
  title     = {Brain Tumor Classification Using Convolutional Neural Network and Deep Transfer Learning Approach with MR Imaging},
  howpublished = {EasyChair Preprint 8995},
  year      = {EasyChair, 2022}}