Brain Tumor classification using BrainNet: A Deep Learning Approach

Cancers of any type are life-threatening diseases and totally hamper normal body functions. One of the leading causes of death in both adults and children is malignant brain tumors. Asa result, detecting brain tumors in their early stages is critical for accurate diagnosis. Magnetic resonance imaging (MRI) is commonly used to diagnose brain tumors. Brain tumors are complex due to extremely erratic shapes and locations and are challenging to fully understand the nature of the tumor. For MRI analysis, a skilled neurosurgeon is also required. Because there are frequently insufficient highly trained medical experts and a general lack of knowledge about tumors in developing countries, obtaining results from MRIs can be extremely difficult and time-consuming. To overcome this drawback of this, proposed a novel Convolution Neural Network(CNN)with BrainNet for the multiclass classification of tumors. The proposed BrianNet performs better than transfer learning models VGG13, VGG16, VGG19, Squeeznet, and InceptionResV2 which are pre-trained on the Imagenet dataset. The proposed CNN (BrainNet) architecture achieved a training accuracy of 99.96% and test accuracy of 97.71% and an average precision of 94.75%. Medical professionals can classify tumours with the aid of the proposed work.