International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
Brain Tumor Detection Using Machine Learning Algorithm Ramnesh Kumar1, Sankalp Rajpoot2, Prateek Kumar Verma3, Mr. Suresh Kumar4 1B.Tech student, Information Technology, Galgotias College Of Engineering & Technology, Uttar Pradesh, India
2B.Tech student, Information Technology, Galgotias College Of Engineering & Technology, Uttar Pradesh, India 3B.Tech student, Information Technology, Galgotias College Of Engineering & Technology, Uttar Pradesh, India
4B.Tech faculty, Information Technology, Galgotias College Of Engineering & Technology, Uttar Pradesh, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Detecting brain tumors via Magnetic Resonance
originate from primary tumors in organs such as the kidneys, lungs, breasts, or from melanomas on the skin. A brain scan offers a detailed visualization of the brain's internal structure. Among the most frequently utilized methods for brain imaging is MRI (Magnetic Resonance Imaging), renowned for its ability to provide exceptional insights into the human body. To categorize MR Images, two primary methodologies are employed: supervised techniques like support vector machines, k-nearest neighbors, and artificial neural networks, and unsupervised techniques such as fuzzy c-means and self-organizing maps (SOM). Many studies have utilized a combination of both supervised and unsupervised techniques to distinguish MR Images as either normal or abnormal. This study employs supervised machine learning techniques to categorize five distinct types of abnormal brain MR Images, including Ependymoma, Lymphoma, Cystic Oligodendroglioma, Meningioma, and Anaplastic Astrocytoma, alongside normal images
Imaging (MRI) is crucial but challenging due to the intricate nature of these abnormalities. A proposed method involves several steps, including sigma filtering, adaptive thresholding, and region detection, to analyze MR images. Shape features such as Major Axis Length, Euler Number, Minor Axis Length, Solidity, Area, and Circularity are extracted to characterize the tumors. This method employs two supervised classifiers: a C4.5 decision tree algorithm and a Multi-Layer Perceptron (MLP) algorithm. These classifiers distinguish between normal and abnormal brain cases, with abnormalities further classified into benign or malignant tumors. With a dataset of 250 brain MR images, the MLP algorithm achieves a notable precision of approximately 80%. Key Words: Magnetic Resonance Imaging (MRI), Image Acquisition, Detection Region, Image preprocessing, Image Segmentation, Feature Extraction
1.INTRODUCTION Brain tumors are solid neoplasms found within the skull, arising from uncontrolled and abnormal cell division. They typically develop in the brain itself, but can also manifest in other locations such as lymphatic tissue, blood vessels, cranial nerves, and brain envelopes. Additionally, brain tumors can result from the metastasis of cancers originating elsewhere in the body. The classification of brain tumors hinges on factors like their location, the tissue type from which they originate, their malignant or benign nature, and other considerations. Primary brain tumors originate within the brain and are named based on the cell types from which they originate. They may be benign, such as Meningioma, which cannot metastasize. Conversely, they can be malignant and invasive, exemplified by Lymphoma (characterized by a ring-like appearance), cystic oligodendroglioma (displaying rounded cells with distinct borders and a central nucleus resembling a "fried egg"), Ependymoma (arising from ependymal cells and exhibiting malignant behavior despite benign histology), and Anaplastic astrocytoma (a common high-grade astrocytoma).
Fig 1. Types of MR images
2. LITERATURE WORK 1. Suraj Grover and fellow writers unveiled an unfamiliar method for segmenting brain tumors in 3D MR pictures. Initially, segmentation of brain MR pictures was carried out utilizing an inventive technique for tumor detection. Afterward, tumor detection leaned on selecting uneven regions. This technique considers the brain's asymmetrical plane and utilizes blurry classification. The results act as the
Secondary brain tumors, also known as metastatic brain tumors, develop from cancer cells that have migrated to the brain from other parts of the body. Typically, these cancers
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 2441