Related Experiment Video
Updated: Jun 25, 2025
09:33
An Acute Retinal Model for Evaluating Blood Retinal Barrier Breach and Potential Drugs for Treatment
Published on: September 13, 2016
7.2K
A Novel Predictive Model Utilizing Retinal Microstructural Features for Estimating Survival Outcome in Patients with
Rebekah Smith1, Ranjit Sapkota2, Bhavna Antony2
1School of Medicine, University of California, Davis.
Research Square
|May 27, 2024
Summary
Glioblastoma patients often have visual field defects. Retinal thinning and microvascular changes in these patients may predict survival, with machine learning showing 78% accuracy in predicting outcomes.
Area of Science:
- Neuro-oncology
- Ophthalmology
- Medical Imaging
Background:
- Glioblastoma is an aggressive brain tumor with a poor prognosis.
- Visual complications of glioblastoma are not well understood.
- Early detection of visual changes is crucial for patient management.
Purpose of the Study:
- To characterize visual outcomes in glioblastoma patients.
- To investigate the relationship between tumor location and visual field defects.
- To identify retinal biomarkers for predicting glioblastoma patient survival.
Main Methods:
- Neuro-ophthalmic examinations were performed on 19 glioblastoma patients.
- Retinal microstructures and microvasculature were imaged.
- Perimetry was used to assess visual fields.
- A machine learning model was developed to predict survival.
Main Results:
- Occipital tumors were associated with significantly worse visual fields compared to frontal tumors.
- Patients with shorter overall survival (<15 months) showed thinner retinal nerve fiber layer and ganglion cell complex.
- Enlarged foveal avascular zone was observed in patients with poor survival.
- Machine learning accurately predicted long-term survival (≥15 months) with 78% accuracy using retinal data and visual fields.
Conclusions:
- Glioblastoma patients frequently experience visual field deficits, even with normal visual acuity.
- Retinal nerve fiber layer thinning and reduced microvascular density are associated with poor survival in glioblastoma.
- Machine learning models show promise in predicting glioblastoma survival based on ocular imaging and visual field data, warranting further validation.
