A linearised fit model for robust shape parameterization of FET-PET TACs

18th March 2021

Christoph Lerche, Timon Radomski, Philipp Lohmann, Liliana Caldeira, Cláudia Régio Brambilla, Lutz Tellmann, Jürgen Scheins, Elena Rota Kops, Norbert Galldiks, Karl-Josef Langen, Hans Herzog, Jon Shah

Positron emission tomography (PET) using the tracer 18F-FET is a proven imaging method for assessing brain tumour patients. In particular, the kinetic analysis of 18F-FET time-activity curves (TAC) can provide valuable diagnostic information in glioma patients. Consequently, an analysis method that provides robust information that is as much detailed i as possible within a clinically practicable timeframe is advantageous in diagnosis and treatment planning.

The aim of this work was to examine 18F-FET uptake using an explicit linearised model that can reproduce the different uptake kinetics of 18F-FET with high accuracy and robustness, using as much TAC data as possible while also allowing the different kinetic behaviours of healthy brain tissue and glioma tissues to be mapped objectively and with low numerical uncertainty.

The model was validated in a group of 33 adult patients with histomolecularly characterised primary (n=32) or recurrent cerebral glioma (n=1) according to the revised WHO classification 2016 of Tumours of the Central Nervous System. The feasibility and predictive value of this approach was also evaluated on the same patient cohort.

The results show that, when compared to alternative fit-models, the linearised model fitted to the averaged TAC values from dynamic FET PET data in the time interval 4-50 min p.i. was able to achieve the best classification of gliomas.

It was also found that, compared to other possible models, the proposed linearised model achieved good agreement with measured TACs, has only two free parameters, and can be linearised, thus providing increased numerical stability and lower parameter uncertainties, especially in the case of voxel-wise analysis.

Moreover, as the classification performance of the linearised fit model and the standard linear fit model were only slightly lower for the reduced acquisition interval 20–40 min p.i, the method is suitable for clinical applications. Consequently, following further investigations, the use of this model is expected to enhance clinical understanding and treatment of gliomas.

Origional publication:

A Linearised Fit Model for Robust Shape Parameterization of FET-PET TACs