Brain Tumor Automated Detection System Based on Hybrid Deep Learning Networks Using MRI Images

Abstract

Technology undeniably has a significant impact on human life, particularly in advancing various aspects of healthcare. Artificial intelligence (AI) has greatly contributed to the early and precise detection of tumors, especially in the realm of medical image processing, with a specific focus on brain tumor detection using magnetic resonance imaging (MRI). However, manually analyzing numerous MRI images is a challenging task in clinical settings. Hence, the utilization of a computer-aided diagnostic (CAD) system is imperative for the timely identification of normal and abnormal brain tissues. This study proposes a hybrid detection architecture to determine the presence of brain tumors. The system incorporates five different deep convolutional neural networks (CNNs) for feature extraction, each comprising five models that identify the optimal five layers for feature extraction before employing Support Vector Machine (SVM) for classification. The system undergoes training and evaluation to ascertain the most effective detection outcomes. The study utilizes two datasets to assess the system's performance and functionality