3d Mri Based Brain Tumor Grade Classification Using Multi Model Algorithms And Prediction By Modal Value

  • B.Jefferson, R.S.Shanmugasundaram
Keywords: Brain Tumor Grade; Magnetic Resonance Image; 3D MRI; Segmentation; Watershed; Classification; Mode; GLCM; RF;KNN;3D CNN.

Abstract

Although Magnetic Resonance Imaging(MRI) is an essential component, accurate brain tumor classification in the medical field can’t be achieved due to the limitations of Machine Learning algorithms. It is important to distinguish between benign and malignant brain tumors before pursuing further treatment. Tumor of grade-I and grade-II belongs to benign type and grade-III and grade-IV belongs to malignant type.  In this study, three classification algorithms are used to improve the prediction accuracy. 3D MRI images are used because they have more features than 2D images. Watershed algorithms are used for segmentation. For extraction of features from the segmented region Gray Level Co-occurrence Matrix (GLCM) is employed. The modal values of three different classification algorithms are used. In a set of values, a value that appears maximum number of  times is referred to as the mode value. The final prediction is made using the Modal Value Prediction(MVP) based on a combination of three classification algorithms(KNN,RF,3D-CNN).

Published
2021-09-03
How to Cite
R.S.Shanmugasundaram, B. (2021). 3d Mri Based Brain Tumor Grade Classification Using Multi Model Algorithms And Prediction By Modal Value. Design Engineering, 5329-5338. Retrieved from http://www.thedesignengineering.com/index.php/DE/article/view/3964
Section
Articles