Click here to


Are you sure ?

Yes, do it No, cancel

Test-Retest Repeatability of 18F-FET PET in Small Volumes

P Ferjancic1*, MA Ebert2,3, AK Nowak2,3, RJ Francis2,3, R Jeraj1,4, (1) University of Wisconsin, Madison, WI, (2) The University of Western Australia, Nedlands, Australia, (3) Sir Charles Gairdner Hospital, Perth, Australia, (4) University of Ljubljana, Ljubljana, Slovenia


(Tuesday, 7/16/2019) 4:30 PM - 6:00 PM

Room: 304ABC

Purpose: ¹�F-fluoro-ethyl-tyrosine (FET) is a common PET tracer in glioblastoma patients but uncertainties of its uptake have not been evaluated. Here, a test-retest analysis of FET PET images was performed by quantifying repeatability of small volumes within lesions and healthy brain, and determining whether normalization by healthy-brain reference regions improves repeatability.

Methods: Eight glioblastoma patients received pre-treatment test-retest FET PET/CT scans within 6-9 days of each other. Test and retest images were rigidly registered and lesions were segmented using local PET thresholds. Bland-Altman 95% Limits of Agreement (LoA) were calculated for SUVmax, SUVmean, and SUVtotal for lesions and entire healthy brain. SUVmean LoA of small volumes within both lesions and healthy brain (range: single voxel to 4 cm³) were calculated for both unchanged and mean-normalized voxel values. LoA were calculated in individual patients and evaluated across the population. Repeatability of lesion SUVmean normalized by mean uptake in varying sized healthy brain reference regions was compared to non-normalized SUVmean repeatability using LoA width.

Results: LoA for lesion-level SUVmean, SUVmax, and SUVtotal were [0.84,1.25], [0.79,1.23] and [0.84,1.25], respectively. For healthy brain, LoA were [0.85,1.15], [0.65,1.13] and [0.85,1.15]. For mean-normalized volumes within lesions, LoA ranged from [0.75,1.34] for single voxels to [0.87,1.15] for the largest considered neighborhood. For healthy brain, LoA ranged from [0.56,1.80] for single voxels to [0.74,1.34] for the largest neighborhood. Healthy-brain normalized lesion SUVmean LoA were wider than non-normalized lesion SUVmean when the volume considered for normalization was smaller than 11 cm3.

Conclusion: This study was the first to evaluate test-retest repeatability of FET PET imaging. Results showed that small volumes in healthy brain tissue were less repeatable than volumes of the same size in tumor. Normalization of SUVmean using reference regions in healthy brain was found to worsen repeatability when reference volumes of less than 11 cm³ was used.

Funding Support, Disclosures, and Conflict of Interest: Work funded in part by the University of Wisconsin Carbone Cancer Center Support Grant P30 CA014520 and by the Cancer Council of WA research grant. Robert Jeraj is a co-founder of AIQ solutions.




IM- PET : Quantitative imaging/analysis

Contact Email