Radiomics Biomarker for Predicting Tumor Infiltration and Recurrence/Locations in Glioblastoma

Grant Application

Primary Investigator: Murat Ak, MD

Proposed Innovation

Glioblastoma is the deadliest and most aggressive form of brain cancer. Despite the best available treatments, recurrence occurs in up to 90% of cases, typically within a few centimeters of the resection cavity. Although more extensive surgery can improve prognosis, there is a risk that it can cause damage to critical areas of the brain responsible for speech, movement, and sensation.

This project aims to develop an artificial intelligence (AI) model to look at medical scans to predict recurrence of glioblastoma. It offers patients the best chance of personalized treatment planning with the least risk of impacting neurologic function.

Improvements in Action

Researchers showed that microscopic tumor infiltration in glioblastoma causes subtle variations that can be detected using radiomics — a research field that extracts quantitative data from standard of care MRIs.

Through this project, MRI studies from patients with a confirmed glioblastoma diagnosis will be used to generate a machine-learning model. A radiomics-based biomarker algorithm will be developed to predict recurrence locations by detecting tumor infiltration that is not visible to the human eye but has already breached the blood-brain barrier.

Intended Outcomes

This project has the potential to revolutionize the management of glioblastoma by improving earlier detection of recurrence, enabling more timely intervention, informing personalized treatment plans, and ultimately improving patient survival rates.