Exploring Machine Learning Algorithms for Neurofibromatosis Disease Classification

Abstract

Neurofibromatosis is a genetic condition which origins tumors growth on the nervous system. It leads to severe complications such as vision impairment, skeletal abnormalities, and neurological dysfunctions. Early and correct diagnosis is vital for effective disease management. When predicting the types of neurofibromatosis, MRIscans are crucial. It takes a lot of time and is prone to human mistake to analyze MRI scans by hand. In this study, we propose eight different hybrid models with four machine Learningclassifierviz, SupportVectorMachine, K-Nearest Neighbors, Decision Tree, and Random Forest and twodeeplearningbasedfeature extractorviz, DenseNet121and ResNet50toclassifytheMRIimageinto Neurofibromatosis Type 1 (NF1), Plexiform Neurofibroma and brain tumor. Additionally, Dimensionality reduction has been applied through Principle Component Analysis to enhance the computational performance. Finally we performs the comparison analysis among the eight models and identifies that SVM classifier with ResNet50 achieve the highest accuracy of 96.88% which is higher than the remaining models. The proposed approach enhances early diagnosis, assisting radiologists in making accurate clinical decisions.

Ähnliche Arbeiten