GLOW-FDG

GLOW-FDG is an open-source AI model for whole-body cancer lesion segmentation in
¹⁸F-FDG PET/CT.

GLOW-FDG stands for:

Generalized cancer LesiOn Whole-body segmentation model for FDG-PET/CT.

Overview

Whole-body FDG-PET/CT is widely used in oncology, but manual lesion delineation is time-consuming, subjective, and difficult to scale. GLOW-FDG provides automated segmentation of FDG-avid cancer lesions across whole-body PET/CT scans.

The model was trained on 1,563 FDG-PET/CT scans spanning multiple cancer types and evaluated on 185 external validation scans from independent cohorts.

Key Features

• Whole-body FDG-PET/CT cancer lesion segmentation
• Trained on diverse multi-cancer data
• Validated on independent external cohorts
• Built on the nnU-Net framework with a ResEncL U-Net architecture
• Uses multitask organ supervision to reduce physiologic uptake false positives

Model Weights

Model weights are available on Hugging Face: mrokuss/GLOW-FDG

Intended Use

GLOW-FDG is intended for research use in automated FDG-PET/CT lesion segmentation.

Citation

If you use GLOW-FDG, please cite: TODO