TIP 2.0: A one-click software package for dynamic brain PET image processing and kinetic modeling

Ziel/Aim The pre-processing and quantitative analysis of dynamic brain PET data is a complex multi-step process requiring many manual steps and bearing a high potential for errors. The aim of this work was to develop a software package that completely automates and standardizes the pre-processing and kinetic modeling of dynamic brain PET data sets.

Methodik/Methods Application programming interfaces of Matlab, SPM 12 and QModeling were used to develop an image processing pipeline for dynamic PET data. A set of ten F-18-PI2620 second generation Tau-PET acquisitions (0-60 minutes post injection) was used to validate the automated data processing against a manual approach.

Ergebnisse/Results TIP 2.0 reads multi-timeframe PET images in DICOM format, performs movement correction by realignment of single timeframes, normalizes the data to a standard space by a perfusion template-based approach of early acquisition frames, performs kinetic modeling (Patlak, Logan, simplified reference tissue models) and calculates voxel-wise statistical deviation maps from average tracer binding in a norm cohort. Intermediate and output files are stored in NIFTI format and can be read by conventional image viewers. Automatically generated output files did not differ from manually created ones.

Schlussfolgerungen/Conclusions TIP 2.0 fully automates pre-processing and kinetic modeling of dynamic brain PET data sets and thereby assures for a standardized and accelerated process, reduces the need for specially trained technical personnel and might also enable the integration of a more reliable modeling-based quantification of PET tracer binding into the clinical routine. The software will be made publicly available.

Publication History

Article published online:
08 April 2021