Extension of the MIAKAT analysis software package to non-brain and pre-clinical PET analysis.

Objectives: The MIAKAT software package for quantitative PET analysis has been available, free for academic use, for over a year. Until now it has only been presented for application to clinical neuroimaging data using established workflows. Here we present extensions to the MIAKAT capabilities, and modified workflows which make it applicable to a broader set of PET experiments. The objective of this work is to provide researchers with access to an efficient, workflow-based software tool covering a broader set of applications including human and non-human, brain and non-brain, dynamic and static imaging.

Methods: MIAKAT (www.miakat.org) is a suite of analysis tools for PET imaging data implemented in MATLAB[1] and currently widely used for the analysis of human dynamic PET neuroimaging studies. MIAKAT provides a central graphical user interface which enables users of all backgrounds to apply robust analysis workflows with built-in audit trail and QC capabilities. The software takes primary image (in NIfTI format) and associated data, and executes a configured analysis pipeline with minimal human input, making it straightforward to analyse PET studies in an efficient, consistent and reproducible way. MIAKAT combines in-house MATLAB code with wrappers for SPM[2] and FSL[3] functions to apply state of the art image processing and kinetic modelling techniques to produce regional and voxel-wise results. The source code is provided with the distribution so that researchers may modify or add to the existing functionality as required.

Results: The existing workflows for dynamic human neuroimaging PET data include processes which are species and target-region specific. Alternative workflows were constructed and implemented which, together with new process modules, cover more general scenarios. The existing broad range of core kinetic modelling techniques has been extended to facilitate the application of image derived input function techniques, as well as semi-quantitative measures such as SUV, SUVr that are frequently used in areas such as oncology and neurodegeneration. The workflows presented cover a broad range of applications, but also act as examples, based upon which advanced users may develop alternative processes and pipelines.

Conclusion: By diversifying the set of available processes and workflows to cover common requirements in non-brain and pre-clinical PET data, the MIAKAT software package was generalised to a broader set of applications. This provides a valuable resource to enable the academic community to deliver reproducible research data. [1] MATLAB R2016a; The MathWorks In., Natick, MA, USA (www.mathworks.com) [2] SPM, Wellcome Trust Centre for Neuroimaging, London, UK (http://www.fil.ion.ucl.ac.uk/spm/) [3] FSL, FMRIB, Oxford, UK (http://fsl.fmrib.ox.ac.uk) Research Support: -