| Abstract: |
In this review, we have explored many AI and profound learning techniques for finding of cerebrum cancer identification in X-ray filters. We tried Convolutional Neural Networks (CNNs), Multilayer Perceptrons (MLPs) with Principal Component Analysis(PCA), and Transfer Learning using InceptionV3 to find out what could be the best possible methods for automated tumor classification. The highest performance the CNN model; 86.27% accuracy in this case, due to its ability of learning and classifying features from MRI images directly possible as showed by our results InceptionV3 with Transfer Learning was also efficient in catching performance of 82.78% thus proved to be most useful way leveraging pre-trained model, even so couldn't beat up the results of CNNs. On the contrary, The MLP with PCA model reached an accuracy of 76.47%, proving that perhaps continuously half way projecting to reduce dimensionality may not catch some important features for classification. Charges such as Logistic Regression, Random Forests and AdaBoost preceded by Naive Bayes (NB) secmenu followed behind in terms of performance but were generally weak relative to the CNN and Transfer Learning-based methods. These outcomes demonstrate that future improvements in feature extraction and model training could achieve better brain tumor detection |
