~ 308 ~ ISSN Print: 2394-7500 ISSN Online: 2394-5869 Impact Factor: 8.4 IJAR 2021; 7(5): 308-313 www.allresearchjournal.com Received: 10-03-2021 Accepted: 12-04-2021 Pranav Shetty Department of Computer Science, All India Shri Shivaji Memorial Society’s College of Engineering Savitribai Phule Pune University, Pune, Maharashtra, India Suraj Singh Department of Computer Science, All India Shri Shivaji Memorial Society’s College of Engineering Savitribai Phule Pune University, Pune, Maharashtra, India Rasvi Jambhulkar Department of Computer Science, All India Shri Shivaji Memorial Society’s College of Engineering Savitribai Phule Pune University, Pune, Maharashtra, India Kajal Sheth Department of Computer Science, All India Shri Shivaji Memorial Society’s College of Engineering Savitribai Phule Pune University, Pune, Maharashtra, India Deepali Ujlambkar Professor Department of Computer Science, All India Shri Shivaji Memorial Society’s College of Engineering Savitribai Phule Pune University, Pune, Maharashtra, India Corresponding Author: Pranav Shetty Department of Computer Science, All India Shri Shivaji Memorial Society’s College of Engineering Savitribai Phule Pune University, Pune, Maharashtra, India Detection of brain tumor using CNN and ML Pranav Shetty, Suraj Singh, Rasvi Jambhulkar, Kajal Sheth and Deepali Ujlambkar DOI: https://doi.org/10.22271/allresearch.2021.v7.i5e.8585 Abstract An automated neurological disorder identification system that uses computer vision on magnetic resonance imaging to locate brain tumors (MRI). The most common and dangerous form of brain cancer is gliomas. Gliomas are tumors that, at their most advanced stage, result in a much shorter life span. Preparing for therapy is an important step in maintaining a better quality of life for oncology patients. Magnetic resonance imaging (MRI) is a technique for examining the structures and components of the human body as well as for medical diagnosis, determining the stage of disease, and monitoring without the use of ionizing radiation. The significant spatial and structural changeability of brain tumors complicates segmentation. As a result, an automatic and consistent segmentation technique is used, based on Convolutional Neural Networks (CNN). Because of the relatively low number of network weights, the use of small kernels enables the development of a deeper architecture, with such a positive effect on over-fitting. It also explores the use of intensity normalization as a pre- processing phase, which is not widely used in segmentation techniques based on the Convolution Neural Network, but has been shown to be efficient in segmenting brain tumors using Magnetic Resonance Imaging (MRI) in conjunction with data augmentation. Keywords: deep learning, brain tumors, convolutional neural networks (CNN), magnetic resonance imaging (MRI) 1. Introduction The human brain is a highly developed and diverse organ made up of extremely spongy and soft tissues. It is widely regarded as the human body's central processing device. Our brain helps us to transmit our expressions, execute our actions, and share beliefs, thoughts, and feelings. In such altered conditions, brain tissue development is unregulated. This abnormal rise in tissue density is referred to as a lump, and if it occurs within the brain, it is referred to as a brain tumor. Tumors appear to form new arteries in the blood. Malignant tumor diagnosis becomes more complex in large tumors. CSF (Cerebral Spinal Fluid) is normally affected by a brain tumor. CSF leakage can cause life-threatening conditions such as meningitis, brain infections, and stroke. As a result, early detection and proper diagnosis are needed. The intended machine identifies the tumor and its shape using computer technology such as image recognition; in general, it is a framework for interpreting and processing recorded images in digital format for detailed data such as color and resolution. The concepts of image processing and MRI were used to develop a scan-based imaging technique for the detection and screening of brain tumors (Magnetic Resonance Imaging). This procedure is not only limited to the detection of tumors in the brain but it may also be used to scan the whole internal anatomy of the human body to detect any tumor. 2. Existing System Magnetic resonance imaging (MRI) is the imaging modality of choice for brain tumor research. A single MRI examination can provide a plethora of data and details. When a radiologist has to make a diagnosis, he or she is faced with a larger number of sources of evidence but less testing instruments [1] . We suggest an automated data-driven tumor recognition approach in which both localization (segmentation) and characterization are used in the identification (signature) process. International Journal of Applied Research 2021; 7(5): 308-313