Topical sensor metrics for 18F-FDG positron emission tomography dose extravasation

Area of Science:

• Nuclear Medicine
• Medical Imaging
• Radiochemistry

Background:

• Extravasation of positron emission tomography (PET) tracers compromises PET image quality and standard uptake value (SUV) accuracy.
• Topical sensors offer a validated method for assessing injection quality in PET imaging.

Purpose of the Study:

• To explore factors contributing to PET tracer extravasation.
• To refine metrics using time-activity curves (TAC) from topical sensors for predicting extravasation.
• To develop an accurate algorithm for detecting extravasation.

Main Methods:

• Retrospective analysis of 18F FDG PET/CT data from 863 patients across 8 international sites.
• Utilized topical Lara sensors to determine dose migration metrics.
• Employed conventional statistical analysis and an artificial neural network (ANN) for deeper variable insights.

Main Results:

• Extravasation was predicted by tc50 (time to double injection sensor counts), ndAvgN (normalized difference at 4 min), and CEnd ratio (final sensor count ratio).
• An ANN algorithm, weighting and scaling these three metrics, achieved 100% sensitivity and 100% specificity for extravasation detection.
• The developed metrics effectively differentiate partial extravasation events.

Conclusions:

• Time-activity curve (TAC) metrics from topical sensors can reliably detect and differentiate partial PET tracer extravasation.
• These metrics offer potential for deeper insights into extravasation's impact on image quality and quantitation.
• Further validation in larger, diverse patient cohorts is recommended for the developed key metrics.