Machine Learning-Based Radiomics in Neuro-Oncology.

Area of Science:

• Neuro-oncology
• Medical Imaging Analysis
• Artificial Intelligence in Medicine

Background:

• Modern medicine generates vast, high-dimensional data, necessitating advanced computational methods.
• Machine learning (ML) and artificial intelligence (AI) are crucial for processing and analyzing complex medical data.
• Radiomics, a field combining radiology and informatics, extracts quantitative biomarkers from medical images.

Purpose of the Study:

• To review novel applications of ML- and deep learning-based radiomics in primary and secondary brain tumors.
• To discuss the implications of these advanced techniques for future neuro-oncology research.
• To highlight the potential of radiomics in predicting patient outcomes and differentiating tumor characteristics.

Main Methods:

• Review of current literature on radiomics, machine learning, and deep learning in neuro-oncology.
• Analysis of radiomics applications in predicting survival, tumor discrimination, and progression assessment.
• Exploration of radiogenomics for molecular phenotyping in brain tumors.

Main Results:

• Radiomics enables non-invasive assessment of quantitative radiological biomarkers.
• Applications include predicting survival, discriminating tumor types, and distinguishing progression from pseudo-progression.
• Radiogenomics shows promise for molecular phenotyping in both primary and secondary brain tumors.

Conclusions:

• ML and deep learning-based radiomics offer significant potential in neuro-oncology.
• Standardization of workflows and availability of multicenter data are key challenges for widespread adoption.
• Future research should focus on addressing these challenges to fully realize the benefits of radiomics in brain tumor management.