Related Experiment Video
Updated: Mar 2, 2026
05:45
Quantitative Immunohistochemistry of the Cellular Microenvironment in Patient Glioblastoma Resections
Published on: July 31, 2017
10.2K
SU-E-J-191: A Multivariate Framework for N-Tissue Classification in Treatment Assessment of Glioblastomas
E Schreibmann1, I Crocker1, H Shu1
1Emory University School of Medicine, Atlanta, GA.
Medical Physics
|May 19, 2017
Summary
Glioblastoma pseudoprogression, often mistaken for tumor growth, can now be identified using a new MRI-based algorithm. This tool helps differentiate true progression from treatment effects, preventing unnecessary interventions for brain tumor patients.
Area of Science:
- Neuro-oncology
- Radiology
- Medical Imaging
Background:
- Glioblastoma is a fatal brain tumor with treatment monitoring challenges.
- New contrast-enhancing lesions can mimic tumor progression but may be pseudoprogression due to treatment effects.
- Distinguishing pseudoprogression from true progression is crucial for patient management.
Purpose of the Study:
- To develop a predictive model to differentiate between true glioblastoma progression and pseudoprogression.
- To improve treatment monitoring and patient outcomes in glioblastoma.
Main Methods:
- A classification algorithm combining perfusion and diffusion MRI (T1, rCBV, ADC imaging) was developed.
- An expectation-maximization (EM) algorithm was used, trained on cases with known clinical outcomes.
- The algorithm classifies voxels into tissue types based on imaging features and proximity.
Main Results:
- The EM algorithm modeled typical imaging values for pseudoprogression, tumor, edema, necrosis, vessels, and brain anatomy.
- The algorithm automatically classifies voxels in new cases using pre-sampled values and Mahalanobis distance.
- A training set of 20 cases with known outcomes was utilized.
Conclusions:
- Advanced classification techniques enable automated voxel labeling into normal, pseudoprogression, or tumoral tissue types.
- Early detection of pseudoprogression can spare patients unnecessary surgery or toxic chemotherapy.
- This technique offers a non-invasive method for improved glioblastoma treatment monitoring.
