MRI Delta Radiomics to Track Early Changes in Tumor Following Radiation: Application in Glioblastoma Mouse Model

Area of Science:

• Neuro-oncology
• Medical Imaging Analysis
• Radiomics and Machine Learning

Background:

• Glioblastoma (GBM) is a lethal brain tumor with poor prognosis, resistant to conventional therapies due to tumor heterogeneity.
• Radiomics offers insights into tumor biology and treatment response by extracting features from MRI scans.
• Delta radiomics, analyzing temporal changes in radiomic features, shows promise for monitoring treatment effects.

Purpose of the Study:

• To evaluate the potential of delta radiomics in tracking temporal radiation-induced changes in glioblastoma (GBM) tumor morphology and microstructure.
• To develop a machine learning model for classifying irradiated tumors based on delta radiomic features.

Main Methods:

• 50 female CD1 nude mice with intracranial G7 glioblastoma were divided into irradiated (IR) and non-irradiated groups.
• MRI scans were acquired at baseline and post-radiation (weeks 12 and 14).
• Radiomic features (shape, histogram, texture) were extracted, analyzed for temporal changes, and used to train a machine learning classification model.

Main Results:

• Delta radiomics successfully captured significant temporal variations in tumor characteristics post-radiation.
• Distinct patterns in delta radiomics features were observed in the IR group, enabling high accuracy in machine learning classification.
• Feature selection was refined using intra-observer reproducibility assessment for reliability.

Conclusions:

• Delta radiomics provides a robust, non-invasive method for monitoring glioblastoma (GBM) treatment response after radiation therapy.
• Future research should focus on MRI delta radiomics for capturing short-term intratumoral radiation effects.
• This approach can significantly improve treatment monitoring and personalized therapeutic strategies for GBM.