A Hybrid Neural Network Model for Brain Tumor Detection in Brain MRI Images

Brain tumors are a specific sort of malignant growth that may manifest in the brain. In most cases, clinicians must manually interpret magnetic resonance pictures in order to arrive at a diagnosis, which significantly limits the imaging technique's usefulness. Strategies to improve diagnostic accuracy and turnaround time are needed to use MRI scans of the brain to detect cancers. It is challenging to isolate tumor areas from images because of the visual variety and similarities between the brain tumor and normal tissues. To deal with this problem, a combination (CNN-LSTM) model along with a traditional machine learning algorithms employing MR images was proposed in our research to identify a brain tumor. The dataset used in this study includes 5,712 brain MR pictures, of which 4,128 are reported to contain tumors. In the standard classifier section, we created three classic classifiers using scikit-learn: Multilayer Perceptron (MLP), Support Vector Machine (SVM), and Logistic Regression, coupled with a deep learning model CNN. Then, we went on to the suggested CNN-LSTM, which is accomplished through (tensorFlow and keras), since its excellence is superior than the conventional algorithms. The suggested CNN-LSTM model can identify brain cancers with a maximum accuracy of 99.30% in a short amount of time and we comprehend the conclusions drawn from the models offered by several researchers must demonstrate that our paradigm is more trustworthy than earlier frameworks. Therefore, the suggested model may assist clinical professionals in determining if an individual has had a brain tumor and, subsequently, speed up the course of therapy.