A Hybrid Deep Learning Approach for Brain Tumor Classification

A brain tumor is an abnormal cell that grows in a certain region of the brain. The classification of tumors is usually conducted by experts in the medical field and manually performed by analyzing Brain MRI scans. This is usually a time-consuming task and prone to human error which can lead to improper diagnosis. In this study, a proposed hybrid CNN-Stacked LSTM model was developed and implemented to classify whether brain MRI scans contain tumors. The proposed approach achieved a classification accuracy of 94% in classifying brain MRI scans compared to the performance of the traditional CNN model which achieved a 92% classification accuracy. This shows that the proposed approach can be used to perform analysis on brain MRI scans and determine whether it contains a tumor. In addition, it was shown that combining deep learning models can be used to improve the performance of the traditional models in performing tasks such as classification. Furthermore, future work can be conducted to implement the proposed approach in performing multiclass classification and fine-tune hyperparameters to achieve optimal performance. Combining other existing deep learning models in performing image classification and performing parallelization can also be explored.