A Hybrid CNN-RNN Model for Accurate Detection of Malignant Brain Tumors

for the treatment planning and outcome of patients, properly identifying malignant brain tumors is very important. Nonetheless, accurate and timely detection of brain tumors can take time to achieve owing to the complexity and variability across tumor types. There has also been an immense establishment of deep learning techniques such as Convolutional Neural Networks CNNs and Recurrent Neural Networks RNNs for various tasks in medical image analysis. Here, we endorse a hybrid convolutional and recurrent neural network CNN-RNN model to detect malignant brain tumors accurately. The model comprises two main facets: a CNN-based feature extractor and an RNN-based sequential decoder. MRI images train a CNN feature extractor, two per tumor type, as input for the encoder and discriminator pipelines. The RNN-based decoder is fed with these features and learns how the temporal dependencies between sequential images can help accurately classify malignant brain tumors. We performed the experiments on a publicly available brain tumor dataset to evaluate our proposed model. Thus, we achieved high accuracy, sensitivity, specificity, and area under the curve AUC. In relatively fewer epochs during the training procedure, this approach surpassed all traditional machine learning methods and even deep learning architectures with several more parameters than it is for the precise prediction of malignant brain tumors.