A Hierarchical CNN Model for Brain Tumor Classification and Survival Rate

Automatic brain tumor segmentation and classification of MRIs are very important for diagnosing and treating brain tumors. Early and accurate detection greatly increases the chances of getting medical help on time, which in turn raises the chances of survival for patients. The dependability of these detection systems has a direct effect on how well doctors can make diagnoses and come up with effective treatment plans. Glioblastoma is among the most aggressive and lethal brain tumors; therefore, perfect accuracy in diagnosis and prognosis is essential to improving patient outcomes. This paper presents a Hierarchical Convolutional Neural Network (HCNN) model that utilizes relevant clinical data and tumor severity analysis to predict the overall survival of glioblastoma patients. The HCNN is a complex deep learning structure that combines structured rate of patient data with medical imaging to make tumor classification and survival prediction better. The model enhances MRI image feature extraction through transfer learning, thereby augmenting diagnostic accuracy and efficiency. The HCNN uses the image analysis capabilities of convolutional neural networks along with clinical records to quickly assess how a tumor is growing and how it will affect a patient's prognosis. The model works well, with a prediction accuracy of 99.67%. This shows that it could be a useful tool for making clinical decisions and planning personalized treatment. Fuzzy neural networks (FNNs) are often used to deal with unclear medical data, but they aren't very accurate, which makes them less useful in real-time clinical settings. So, the proposed HCNN architecture is a better and more reliable way to diagnose medical problems.