Bio-Inspired Optimization for Medical Data Mining

6Classification of Brain Tumor Using Machine Learning Techniques: A Comparative Study

Gandla Shivakanth¹*, Bhaskar Marapelli², A. Shivakumar Reddy², Dasari Manasa² and Samtha Konda³

¹Department of CSE, Koneru Lakshmaiah Education Foundation, Hyderabad, Telangana, India

²Department of CSE, Koneru Lakshmaiah Education Foundation, Vijayawada, India

³Department of IT, Muffakam Jah Engineering College, (Sultan-UL-Uloom Education Society) Hyderabad, Telangana, India

Abstract

Due to their complexity and heterogeneity, brain tumors are a significant medical illness that can be difficult to precisely identify. Based on medical imaging data, machine learning approaches have shown potential in helping in the diagnosis and categorization of brain cancers. Support vector machines (SVM), random forests (RF), k-nearest neighbours (KNN), and deep neural networks are some of the machine learning methods we compare in this paper for the classification of brain tumors (DNN). Using a dataset of brain tumor MRI images, we assessed the performance of these algorithms using a variety of evaluation criteria, including accuracy, precision, recall, and F1 score. Our findings show that, with an accuracy of 90.5%, DNNs outperformed the competition, followed by SVM (85.3%), RF (81.9%), and KNN (78.5%). In order to determine which aspects were most crucial for each algorithm, we also did feature selection. We discovered that various algorithms favoured various feature sets. Our research sheds light on ...