Segmentation of gliomas in ¹⁸F-fluorocholine PET/CT. A multiapproach study

Area of Science:

• Nuclear Medicine
• Oncology
• Radiology

Background:

• Glioma segmentation is crucial for treatment planning and response assessment.
• ¹⁸F-fluorocholine Positron Emission Tomography/Computed Tomography (PET/CT) shows promise for visualizing glioma.
• Standardized segmentation methodologies are needed for reliable interpretation of PET/CT scans.

Purpose of the Study:

• To evaluate interobserver agreement in glioma segmentation using ¹⁸F-fluorocholine PET/CT.
• To identify a reliable method for segmenting gliomas based on PET/CT imaging.

Main Methods:

• 25 glioma patients underwent ¹⁸F-fluorocholine PET/CT.
• Tumor segmentation was assessed using fixed thresholds, semi-automatic methods with optimal SUVmax-% thresholds, and variable SUVmax-% thresholds.
• Interobserver variability was quantified, and segmentation accuracy was evaluated by consensus.

Main Results:

• Significant variability was observed among observers in tumor segmentation.
• Optimal SUVmax-% threshold demonstrated a correlation with SUVmax, but a single regression model was insufficient.
• A fixed SUVmax threshold (mean background + 3 SD) segmented a portion of tumors but not all.

Conclusions:

• High interobserver variability complicates reliable glioma segmentation with ¹⁸F-fluorocholine PET/CT.
• No single tested method achieved consistent and accurate segmentation of all gliomas.
• Tumor heterogeneity on ¹⁸F-fluorocholine PET/CT likely contributes to segmentation challenges.