Diagnosis of Brain Tumor Using Deep Learning Stacked Classifier -Ml Algorithms

Abstract

Brain tumours are very lethal, and radiologists have a difficult time classifying them due to the varied characteristics of tumour cells. Recently, computer-aided diagnostic (CAD) systems have shown promise as an enabling tool for detecting brain tumours using MRI. The bottom layers of natural photos and medical images are distinct, although both are used in modern applications of pre-trained models. This research presents a strategy for early identification of brain tumours that involves the extraction and concatenation of many characteristics at different levels. These results are credible since VGG-16 is a pretrained deep learning model. These two models were used to compare two scenarios for identifying and categorising brain tumours. To begin classifying brain tumours, characteristics were first retrieved from various VGG modules using the pretrained VGG_16 model. Second, we used PCA, LDA, and ICA to derive features. These merged characteristics were then fed into a hybrid classifier to determine the kind of brain tumour. Finally, the presence of brain tumour is be identified.The results obtained using matlab tool are compared in terms of various parameters evaluated.