Abstract
Brain tumor classification is a very important and the most prominent step for assessing life-threatening abnormal tissues and providing an efficient treatment in patient recovery. To identify pathological conditions in the brain, there exist various medical imaging technologies. Magnetic Resonance Imaging (MRI) is extensively used in medical imaging due to its excellent image quality and independence from ionizing radiations. The significance of deep learning, a subset of artificial intelligence in the area of medical diagnosis applications, has macadamized the path in rapid developments for brain tumor detection from MRI to higher prediction rate. For brain tumor analysis and classification, the convolution neural network (CNN) is the most extensive and widely used deep learning algorithm. In this work, we present a comparative performance analysis of transfer learning-based CNN-pretrained VGG-16, ResNet-50, and Inception-v3 models for automatic prediction of tumor cells in the brain. Pretrained models are demonstrated on the MRI brain tumor images dataset consisting of 233 images. Our paper aims to locate brain tumors with the utilization of the VGG-16 pretrained CNN model. The performance of our model will be evaluated on accuracy. As an outcome, we can estimate that the pretrained model VGG-16 determines highly adequate results with an increase in the accuracy rate of training and validation.
Highlights
Tumor is termed as neoplasm produced by uncontrolled growth of anomalous cells [1]
Dataset Collection and Preprocessing. e dataset acquired in this study is a collection of images of brain Magnetic Resonance Imaging (MRI) scans. ere exist around 256 raw MRI images of different dimensions, usually measured in terms of pixel values. e sample MRI brain images are gathered from the Kaggle dataset. e collected images are in Joint Photographic Experts Group (JPEG) format [23]. e image database is categorized into two segments, Yes and No, based on the existence of the tumor in an MRI brain image
As there are a lot of differences between natural images and MRI images, we need to fine-tune our model with transfer learning methods with convolution neural network (CNN)-based pretrained hyperparameter models [5]
Summary
Tumor is termed as neoplasm produced by uncontrolled growth of anomalous cells [1]. MRI images are one of them that are used by medical experts It is a noninvasive technique used for accurate image data analysis of human brain tumors in determining the patient’s condition [4]. Meningioma tumors arise from cerebral membranes that encapsulate the brain and spinal cord within the inner portion of the skull These passive growth tumors occur on the three membrane layers called meninges. Image classification is the most significant and crucial task, especially in medical healthcare domains, including biomedical imaging, detecting, and diagnosing the disease accurately, which helps radiologists to improve diagnostic efficiency and provide a better path for surgical treatment. E brain tumour is a life-threatening neurological disorder that occurs due to the uncontrollable growth of abnormal cells in the human nervous system. Medical treatment options depend on whether or not the tumor is spreading and affecting the other organs of the body or within the part of the central nervous system (CNS)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
