An Effective Approach to Detect and Identify Brain Tumors Using Transfer Learning

Brain tumors are considered one of the most serious, prominent and life-threatening diseases globally. cause thousands deaths every year around globe because rapid growth tumor cells. Therefore, timely analysis automatic detection brain required to save lives people globe. Recently, deep transfer learning (TL) approaches widely used detect classify three types tumors, i.e., glioma, meningioma pituitary. For this purpose, we employ state-of-the-art pre-trained TL techniques identify pituitary tumors. The aim is performance nine classifiers, Inceptionresnetv2, Inceptionv3, Xception, Resnet18, Resnet50, Resnet101, Shufflenet, Densenet201 Mobilenetv2, by automatically identifying detecting using a fine-grained classification approach. this, algorithms evaluated on baseline (MRI) dataset, which freely available Kaggle. Additionally, all (DL) models fine-tuned with their default values. experiment demonstrates that inceptionresnetv2 algorithm performs better achieves highest accuracy in classifying hence it can be classified as best algorithm. We achieve 98.91% accuracy, 98.28% precision, 99.75% recall 99% F-measure values algorithm, out-performs other DL algorithms. ensure validate compare efficacy hybrid approaches, use convolutional neural networks (CNN) for feature extraction Support Vector Machine (SVM) classification. Similarly, experiment’s results show algorithms, particular, out-perform MRI images into meningioma, experiments Mobilnetv2, Densenet201, Squeeznet, Alexnet, Googlenet, Inceptionresnetv3, VGG19 Shufflenet.