Shallow Convolution Neural Network for Brain Tumor Classification using T1-weighted MRI scans

Recent advancements in deep learning has opened spaces for new research which is also well complimented with advancements in computational power. But it is still out of reach of many poor countries because of its cost. This work is in search of that solution. It is the view of the author that a lot of our problems can be solved through simple models. So in order to make it affordable shallow networks are designed. In this work, three shallow Convolutional Neural Network (CNN) i.e. Model 1, Model 2 and Model 3 and additionally three pre-trained model such as Resnet50, InceptionV3 and MobileNetV2 were used for classification of Brain tumor. Three classes of tumor are present namely Meningioma, Glioma and Pituitary. Model 1 showcased the best accuracy with 98.47%. Model 2 had an accuracy of 95.41%. Among the pre-trained model InceptionV3 had the highest accuracy of 92.36%.