Radiomics signature based on FDG-PET predicts proliferative activity in primary glioma

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Glioma proliferative activity is crucial for prognosis.
• Current methods for assessing proliferation, like Ki-67 index, are invasive.
• Non-invasive methods are needed for glioma evaluation.

Purpose of the Study:

• To investigate a radiomics method using 2-[18F]-fluoro-2-deoxy-d-glucose (FDG) positron-emission tomography (PET) for non-invasive evaluation of glioma proliferative activity.
• To develop and validate predictive signatures for glioma prognosis.

Main Methods:

• Retrospective analysis of 123 primary glioma patients.
• Extraction and selection of 1,561 radiomics features from FDG-PET scans.
• Generation of radiomics, clinical, and fusion signatures; performance evaluation using ROC analysis.

Main Results:

• A radiomics signature comprising nine features achieved high accuracy (81.7% primary, 73.2% validation) and AUC (0.88 primary, 0.76 validation).
• Clinical and fusion signatures showed overfitting in the validation cohort.
• Both Ki-67 index and the radiomics signature effectively stratified patients into distinct prognostic groups.

Conclusions:

• Radiomics signature derived from 18F-FDG-PET is a promising non-invasive tool for measuring glioma proliferative activity.
• This method aids in predicting patient prognoses, offering a valuable alternative to invasive techniques.