A Comparative Study of Machine Learning Algorithms for Brain Tumor Detection

This study examines brain imaging to identify areas with tumors and categorizes these regions into three distinct types: meningioma, glioma, and pituitary tumors. This paper also compares different machine learning algorithms for the identification of brain tumors. The term "brain tumor" describes the excessive growth of cells in the brain, that can be either benign or malignant.
In this study, machine learning algorithms have been utilized to both identify what kind of brain tumor in patients who have been diagnosed with one using multi-class classification and to detect whether brain tumors are present or absent by binary classification. There are two types of brain MRI pictures in the dataset used for the binary classification challenge: those containing tumors and those without. Several machine learning algorithms, including Support Vector Machine (SVM), Logistic Regression, K-Nearest Neighbor (KNN), Naïve Bayes (NB), Decision Tree (DT) classifier, and Random Forest classifier, have been applied to classify the MRI images.
Accuracy, recall, precision, and F1-score were among the performance indicators used in a comparative analysis of machine learning algorithms. With an accuracy of 90.8%, recall of 95.2%, precision of 81.1%, and F1-score of 87.6% across 1,475 brain tumor images, the Random Forest classifier outperformed the other techniques.