Classifying Brain Tumor from MRI Images Using Parallel CNN Model

Abstract

Listen

Translate article

Abstract Brain tumor, commonly known as intracranial tumor, is the most general and deadly disease which leads to a very short lifespan. It occurs due to the uncontrollable growth of cells which is unchecked by the process that is engaged in monitoring the normal cells. The survival rate due to this disease is the lowest and consequently the detection and classification of brain tumor has become crucial in early stages. In manual approach, brain tumors are diagnosed using (MRI). After the MRI displays the tumor in brain, the type of the tumor is identified by examining the result of biopsy of sample tissue. But having some limitations such as accurate measurement is achieved for finite number of image and also being time consuming matter, the automated computer aided diagnosis play a crucial rule in the detection of brain tumor. Several supervised and unsupervised machine learning algorithms have been established for the classification of brain tumor for years. In this paper, we have utilized both image processing and deep learning for successful classification of brain tumor from the MRI images. At first in the image preprocessing step, the MRI images are normalized and through image augmentation the number of images is enriched. Further the preprocessed images are passed through a parallel CNN network where the features of the images are extracted and classified. Our experimental result shows an accuracy of 89% that is promising.KeywordsBrain tumorData augmentationConvolution neural networkDeep learning