Convolutional neural network based spine image classification and detection of abnormalities in mri spine image

Abstract

Spine tumor is a rare disease, a fast-growing abnormal tissue in or newlinesurrounding the spinal column which affects many people. Thousands of newlineresearchers have focused on this disease for increased awareness of tumor then newlineclassification to provide more effectual treatment to the patients. The treatment newlineof spine tumor depends on tumor size, tumor s growth rate, stage of tumor and newlineother characteristics. Various treatments are available, which include drugs, newlinesurgery, chemotherapy, and immunotherapy. An effort has been achieved in newlinethis study to see the correlation among clinical, radiological also pathological newlinediagnosis of spine tumor. 95% of clinical diagnosis correlated with the newlineradiological findings for all kinds of tumors. newlineAt present, one of the most effective ways to detect tumors or masses in newlinethe spine through Magnetic Resonance Imaging (MRI). MRI is a powerful newlineimaging technique for producing high-resolution images of the various newlinebiological tissues with good contrast. MRI images can detect early spine newlinetumors, MRI s sensitivity is inversely proportional to tumor density. The newlinechallenge lies in accurate detection to overcome the development of spine newlinetumor, which will spread from other regions of the body to the spine easily. newlineThe dataset contains various MRI spine images of different patients with newlinedifferent ages and groups, both male and female, at various stages of image newlinefrom Bharath Scan Research Centre, Chennai, and Spineweb database. The first newlinedataset obtained from contains 40 set of patients with and without tumor newline(Normal, Astrocytomas, Meningiomas) which have T1-weighted (T1-W), T2- newlineweighted (T2-W) with axial, sagittal and coronal plane image. newline