Imagine Segmentation for Brain Tumor Prognosis

A brain tumour is a lump or development of abnormal cells in your brain. Brain tumours come in many different forms. There is evidence of both cancerous and non-cancerous brain tumours. The goal of this work is to develop an automated brain tumour detection system using classification & segmentation of MRI images. The size & shape of brain tumors are completely variable. This motivates us to employ large-scale deep neural networks. An overview of what we performed and the approach we employed is given in this article. We utilized the Kaggle dataset and tested the trained model. 5000 images are included in this dataset for the segmentation job and 8000 images are included for the classification task. Patch segmentation is a technique used to separate MRI scans. For the test data set, this methodology displays an accuracy of roughly $95.6 \%$. To evaluate the depth of our neural network, we ran a number of experiments and identified the ideal architecture. A convolutional neural network(CNN) was used to pinpoint the exact site of the tumor. To locate the spectral localization of gliomas in lilac cells, we combined deep and convolutional neural networks. Long-term state predictions can also be made using this design.