Noninvasive Glioma Grading with Deep Learning

Authors: G. Danilov, V. Korolev, M. Shifrin, E. Ilyushin, N. Maloyan, D. Saada et al.
Published: MEDINFO 2021: One World, One Health, 2022
Medical Imaging Deep Learning Brain Tumors

Abstract

This pilot study explores the use of deep learning for non-invasive grading of brain gliomas from MRI scans. We develop convolutional neural network models to classify tumor grade without requiring surgical biopsy, potentially improving treatment planning and patient outcomes.

Clinical Impact

Non-invasive diagnosis: Reduce need for surgical biopsy in tumor grading
Treatment planning: Earlier and more accurate grade assessment
Patient outcomes: Faster diagnosis enables timely intervention
Clinical integration: Designed for real-world neurosurgery workflows

Technical Approach

We trained deep convolutional neural networks on MRI sequences from glioma patients, developing models that can distinguish between low-grade and high-grade tumors based on imaging features alone.

Abstract

Clinical Impact

Technical Approach

Related Topics