Abstract
1928
Objectives Quantitative analysis of clinical PET neuroimaging data involves a sequence of steps or processes that transform the primary data, from a number of sources, into parameters with biological significance. These outcome measures are used routinely in human studies of normo- and pathophysiology, disease progression and drug development. Until relatively recently, different components of the analysis workflow have been performed using separate applications, with significant human interaction and thus inefficiency with the potential for the introduction of errors and lack of reproducibility. MIAKAT addresses these issues and is being shared freely with the academic community to provide accessible software tools that will enable a broad set of users to conduct high quality, state of the art analysis of PET neuroimaging research data.
Methods MIAKAT (www.miakat.org) is a suite of analysis tools for PET neuroimaging data integrating state of the art tools in a single user friendly software environment making it easy for novice or advanced imaging scientists to efficiently and accurately analyse their data in a fully quantitative manner. The software includes functionality for motion correction of dynamic images, neuoranatomical parcellation of regions of interest (ROIs), blood/plasma input function modelling and tracer kinetic modelling for regional or parametric image analyses. MIAKAT is implemented in MATLAB[1], and has a central GUI that facilitates “point and click” operation. The standard brain pipeline configurations implemented in MIAKAT take the primary experimental data (dynamic PET and structural MR images, along with ancillary data such as arterial blood measurements) and perform a sequence of processes which ultimately produce results in the form of regional (or voxel-wise) parameters extracted via a comprehensive set of modelling techniques. MIAKAT combines in-house code with wrappers for SPM[2] and FSL[3] commands in order to provide start of the art functionality, including inbuilt audit trails, within a coherent framework.
Results Users can run preconfigured standard analysis workflows, or bespoke pipelines and processes to suit their needs. The software is distributed as MATLAB m-code, enabling users to understand and interact with the software at any level. The MIAKAT distribution consists of around 60,000 lines of Matlab code. MIAKAT implements a set of default image analysis workflows, as well as providing the ability to easily develop variations or new workflows as required. The standard PET neuroimaging analysis pipeline demonstrated here starts with a subject’s dynamic PET data and associated structural MRI in the NIfTI file format, along with associated ancillary data (such as frame times, injected radioactivity data, measure arterial blood, plasma and metabolite activity data) in a well-defined text-based .anc file format. The full analysis pipeline typically takes less than one hour per PET scan to perform all spatial and kinetic analysis processes, and has been used to analyse over 500 clinical PET scans.
Conclusions MIAKAT brings together state of the art tools for the quantitative analysis of PET molecular imaging data in a single user friendly software environment enabling high quality quantitative and reproducible analysis of PET neuroimaging studies. The inbuilt audit trails ensure quality and consistency. The modular nature of the software design easily permits the incorporation of additional algorithms as they become available, and the efficient implementation of alternative workflows. This will be a valuable resource for the academic community that currently does not have free access to such tools. [1] Matlab R2012b, The MathWorks Inc., 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) $$graphic_8B78BF3C-E78E-4266-9AC1-356D0D95ED55$$
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.