Brain tumour classification using saliency driven nonlinear diffusion and deep learning with convolutional neural networks (CNN)

摘要

Experts notice and classify various Regions of Interest (ROI) manually for identification, analysis, and development of a treatment. To overcome errors and discrepancies of the data in this state, automated analysis is utilized. A novel method for the classification of MRI brain tumor is proposed in this paper using the saliency driven image representation and CNN based classification with optimization. Initially, the preprocessing on MRI images is carried out using canny edge detection algorithm followed by saliency driven image representation using modified minimum barrier distance and nonlinear diffusion at multiple level. Finally, feature extraction and image classification is carried out by CNN and the optimization by ADAM optimizer. The implementation is carried out and the results are evaluated which outperforms the earlier methods.