Hands-On Perfusion Modelling
Michael Chappell1,2 and Patricia Figueiredo3
1Institute of Biomedical Engineering & Wellcome Centre for Integrative Neuroimaging, University of Oxford, United Kingdom, 2Beacon of Excellence in Precision Imaging & Sir Peter Mansfield Imaging Centre, University of Nottingham, Nottingham, United Kingdom, 3ISR-Lisboa/LARSyS and Department of Bioengineering, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal

Synopsis

This session aims to provide a concise but comprehensive overview of models of arterial spin labelling (ASL) perfusion signals. The basic principles of ASL perfusion imaging based on tracer kinetics will be presented, and common ASL modelling scenarios will be described focusing on perfusion quantification as well as on the exploration of other available haemodynamic information. The session will be organized in three parts, each consisting of a description of the theory underpinning a specific modelling approach, followed by a step-by-step hands-on demonstration of its practical application on real datasets using a freely available tool.

Target audience

Researchers who want to get started with ASL perfusion signal modelling; clinicians and basic scientists interested in quantifying perfusion and other haemodynamic parameters from ASL data.

Objectives / outcomes

To provide a concise but comprehensive overview of models of ASL perfusion signals, together with step-by-step live demonstrations of common ASL modelling scenarios focusing on perfusion quantification and exploring what other haemodynamic information is available using ASL.
By attending this course, participants should be able to:
- Describe the standard tracer-kinetic models of ASL perfusion signals, including their main assumptions and limitations, and be familiar with advanced models (including a macrovascular compartment, partial volume estimation, bolus dispersion, or restricted water exchange);
- Describe how ASL modelling can be used to quantify perfusion as well as other haemodynamic parameters (including arterial transit time or arterial blood volume);
- Undertake simple kinetic modelling tasks on real ASL datasets, including the main data analysis steps leading to the quantification of perfusion and other haemodynamic parameters, using freely available tools.

Overview

The session will be organized in three parts, each consisting of a description of underpinning theory introducing a specific modelling approach, followed by a hands-on demonstration illustrating its practical application on real datasets using the freely available tool BASIL, part of the FMRIB Software Library (FSL) (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/BASIL). The three parts will cover the following theoretical topics and hands-on demonstrations:
Theory 1
Following the definition of perfusion, the basic principles of perfusion imaging based on tracer kinetics will be presented, including the concepts of arterial input function (AIF) and tissue residue function. The basic principles underlying ASL perfusion imaging will then be introduced, by describing how an endogenous perfusion tracer can be created through the inversion of the magnetisation of arterial blood water spins. The main labelling strategies of pulsed ASL (PASL) and continuous and pseudo-continuous ASL (pCASL) will be presented, as well as the associated timing parameters inversion time (TI) and post-labelling delay (PLD), respectively. The main issues regarding the acquisition of ASL data, consisting of both label and control images for subsequent subtraction, will be discussed, including background suppression, readout schemes, and signal averaging.
The standard kinetic model that is most commonly used to explain ASL perfusion signals will be described, including: the boxcar AIF resulting from the assumption of no label dispersion, and associated parameters of labelling efficiency, label duration, and arterial transit time (ATT); and the single, well-mixed compartment residue function resulting from the assumption that water is a freely diffusible tracer, and associated parameters blood-tissue partition coefficient of water and apparent tissue T1. The kinetic curves obtained for both the PASL and pCASL labelling methods will then be presented, highlighting their sensitivity to perfusion as well as the ATT.
Perfusion quantification based on single-delay (TI/PLD) ASL measurements will be described (following the ASL consensus paper recommendations), including: the choice of TI/PLD value, the model inversion equation, and the calibration method for the estimation of the equilibrium magnetisation of arterial blood.
Hands-on 1
Perfusion quantification based on the analysis of a single-PLD pCASL dataset will be demonstrated, following the ASL consensus paper recommendation, including: the main pre-processing steps; label-control subtraction; standard kinetic model inversion; calibration; and perfusion quantification.
Theory 2
The potentially confounding effects of variable, and in particular unexpectedly long, ATT values will be described, motivating the need for multi-delay ASL measurements. The main multi-delay sampling approaches will be presented, including Look-Locker and time-encoded sequences, as well as associated optimal sampling strategies. Methods for fitting the standard kinetic model to multi-delay ASL data in order to estimate both perfusion and ATT will be introduced, and the utility of ATT as a haemodynamic parameter in its own right will be considered.
The contribution of water spins in arteries and arterioles to the ASL signal, particularly when measured at short delays, will be discussed. Methods for minimising these macrovascular signal contributions through the use of flow-suppression gradients will be described. An extended kinetic model including an additional component explaining macrovascular signal contributions will then be presented, and how to quantify the arterial blood volume (aBV), as an extra haemodynamic parameter, will be described.
Hands-on 2
A multi-PLD pCASL dataset will be analysed, including fitting of a standard kinetic model with and without macrovascular signal component. Approaches for non-linear model fitting will be explored including Bayesian inference with different types of prior information, for the quantification of perfusion, ATT and aBV. An optimal multi-delay ASL acquisitions strategy will be generated.
Theory 3
Some advanced ASL kinetic modelling approaches will be briefly introduced, in order to illustrate potential future directions in specific situations where the standard or extend models may not be appropriate. These include: modelling the AIF in the presence of bolus dispersion, with potentially significant impact on ATT estimates, when sufficient temporal resolution is available; modelling restricted water exchange in the residue function, potentially allowing for the measurement of water permeability; and modelling partial volume effects by considering multiple model components for the different tissues.
Hands-on 3
A multi-PLD pCASL dataset will be analysed with the inclusion of dispersion in the kinetic model to explore the difference this makes to the resulting parameter estimation. Finally, the process of partial volume correction will be explained and applied to this data to attempt to separate grey and white matter perfusion.

Software

The following is a non-exhaustive list of software tools available for the analysis of ASL data:
- The main scanner manufacturers have in-built tools for basic perfusion-weighted image generation and also for perfusion quantification; however, more options are generally available if you export the data and analyse it yourself.
- BASIL, part of the FMRIB Software Library (FSL) (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/BASIL). BASIL will be used for the demonstrations in this session.
- ASLtbx, a SPM plugin (https://cfn.upenn.edu/~zewang/ASLtbx.php).
- ASAP, a SPM plugin that also exploits FSL tools (https://sites.google.com/site/asltoolbox/).
- ExploreASL, a Matlab pipeline focused on large multi-center studies (https://sites.google.com/view/exploreasl).

Acknowledgements

Engineering & Physical Sciences Research Council UK (EP/P012361/1)

References

Books

Buxton R, Introduction to Functional Magnetic Resonance Imaging: Principles and Techniques (2nd ed.). Cambridge University Press, 2009.

Chappell, MacIntosh & Okell, Introduction to Perfusion Quantification using Arterial Spin Labelling. Oxford University Press, 2017. http://www.neuroimagingprimers.org/ (freely available tutorial exercises for ASL using BASIL tools available via the website).

Reviews

Alsop D, et al., Recommended implementation of arterial spin‐labeled perfusion MRI for clinical applications: A consensus of the ISMRM perfusion study group and the European consortium for ASL in dementia. Magnetic Resonance in Medicine 2015; 73(1):102–116.

Jezzard P, et al., Arterial spin labeling for the measurement of cerebral perfusion and angiography. Journal of Cerebral Blood Flow & Metabolism 2018; 38(4):603–626.

Papers

Buxton R, et al., A general kinetic model for quantitative perfusion imaging with arterial spin labeling. Magnetic Resonance in Medicine 1998; 40(3):383–396.

Chappell MA, et al., Separation of Macrovascular Signal in Multi-inversion Time Arterial Spin Labelling MRI. Magnetic Resonance in Medicine 2010; 63(5):1357–1365.

Chappell MA, et al., Partial Volume Correction of Multiple Inversion Time Arterial Spin Labeling MRI Data. Magnetic Resonance in Medicine 2011; 65:1173–1183.

Chappell MA, et al., Modeling Dispersion in Arterial Spin Labeling: Validation Using Dynamic Angiographic Measurements. Magnetic Resonance in Medicine 2013; 69:563–570.

Sousa I, et al., Reproducibility of the Quantification of Arterial and Tissue Contributions in Multiple Postlabeling Delay Arterial Spin Labeling. Journal of Magnetic Resonance Imaging 2014; 40:1453:1462.

Woods et al., A general framework for optimizing arterial spin labeling MRI experiments. Magnetic Resonance in Medicine 2018; 81(4):2474-2488.

Pinto J, et al., Calibration of arterial spin labeling data — potential pitfalls in post‐processing. Magnetic Resonance in Medicine2020; 83 (4),1222-1234.

Proc. Intl. Soc. Mag. Reson. Med. 28 (2020)