• Application of Machine Learning-based Radiomics Feature in classification of High Grade from Low Grade Brain Tumors
  • Amirreza Sadeghinasab,1,* Mahmoud Mohammadi-Sadr,2 Fatemeh Mazaheri,3
    1. Department of Radiologic Technology, School of Allied Medical Sciences, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran. and Students Research Committee, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran.
    2. Department of Medical Physics, School of Medicine, Isfahan University of Medical Sciences, Isfahan, Iran
    3. Department of Radiologic Technology, School of Allied Medical Sciences, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran. and Students Research Committee, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran.


  • Introduction: Brain tumors are complex diseases with significant variability in behavior and prognosis. High-grade and low-grade tumors, in particular, present distinct challenges in diagnosis and treatment planning. While imaging plays a crucial role in tumor characterization, accurate differentiation between these tumor types remains a clinical challenge. This study sought to investigate the potential of radiomics, a quantitative image analysis technique, to enhance the differentiation between high-grade and low-grade brain tumors using features extracted from magnetic resonance images.
  • Methods: PubMed, Science Direct, Web of Science, and Google Scholar databases were explored up to August 2024, using different combinations of the keywords: "Brain tumors ", "Radiomics", "Machine Learning", "Magnetic resonance imaging" and "Classification". Finally, five more recent and relevant records were included in the study.
  • Results: Findings have demonstrated that XGBoost, SVM, and Random Forest classifiers exhibit robust and reliable performance in classifying brain tumors into low-grade and high-grade categories. The RF classifier achieved an accuracy of approximately 0.83 and an AUC of 0.81, which are considered excellent. Additionally, in another study, the XGBoost classifier reported an accuracy of 0.88. These results indicating a promising ability to accurately classify tumors based on their imaging characteristics.
  • Conclusion: The high performance of ML models in different studies demonstrated the value of radiomics as a complementary tool for brain tumor characterization. By providing quantitative insights into tumor heterogeneity, radiomics may aid in improving diagnostic accuracy and treatment planning. Further research is essential to validate these results in larger, independent cohorts and to explore the clinical utility of radiomic models in routine practice.
  • Keywords: Brain tumors, Radiomics, Machine Learning, Magnetic resonance imaging, Classification