Enhanced Model for Brain Tumor Detection Accuracy Using Inceptionresnetv2 compared to VGG19 and MobileNet Models

Abstract

Brain tumors, arising from unregulated and accelerated cellular proliferation, provide considerable health hazards if not promptly addressed. Notwithstanding significant progress, precise segmentation and classification continue to pose difficulties. Artificial intelligence (AI) and medical imaging technologies have advanced recently, greatly improving illness analysis and prediction, particularly with regard to brain tumor (BT) identification. Precise classification and segmentation of tumors in the brain are essential for creating individualized treatment plans, and early detection of brain tumors is essential for improving patient prognosis and available treatments. Even with the increasing usage of Magnetic Resonance Imaging (MRI) for brain imaging and the development of AI-based detection techniques, developing a reliable and effective model to identify and classify cancers from MRI images is still a difficult task. Brain tumor analysis is essential for timely diagnosis and effective patient management. Tumor morphology, encompassing features such as dimension, position, appearance, and heteromorphic shape in medical images, poses challenges for analysis. Advancements in DL approaches for the automatic identification of tumors present in the brain using MRI data have led to the utilization of numerous algorithms for this purpose. The InceptionResNetV2 model is employed due to its efficiency and efficacy; it is a simple deep learning architecture, rendering it a suitable candidate for medical image processing applications. The proposed model attains an accuracy rate of 100% at an epoch rate of 30 and surpasses the performance of both MobileNet and VGG19 models.