Synopsis
This syllabus outlines the teaching objectives of the Sunrise Session "DSC and DCE Basics and Analysis." This is essentially a large-scale overview of Dynamic MRI studies, intended to provide directions for further study in a general audience.
Target Audience:
Clinicians, Radiologists, Students, and Physicists exploring, performing and/or analyzing assessments of tissue vascular function.Outcome/Objectives:
The main purpose of this presentation is to transmit an intuitive understanding of how inferences about vascular physiology can be formed from MRI experiments using contrast agents in dynamic studies. I intend to minimize the presentation of equations and maximize intuitive approaches to time-dependent indicator dilution studies using MRI contrast agents.
Attendees should gain an understanding of the following:
1. How the principle of conservation of mass applies to indicator dilution studies.
2. Indicator dilution studies in tissue considered as a stationary two-port system.
3. The central volume theorem and the frequency function of transit times as framed by the two-port model.
4. Convolution as an operator in two-port systems.
5. The internal structures of useful two-port models and related inferences about tissue physiology.a. Exponential kernels and the inverse problem.
6. The relation between MRI signal and tissue indicator concentration.
a. T2* contrast versus T1 contrast
b. The effect of water exchange across tissue boundaries
7. Model selection in stochastic processes
8. Where the challenges are:
a. Arterial input sampling
b. Nonlinear contrast mechanisms
c. Tapering effects in biological systems
Purpose:
The attendees should get an understanding of both the inferences possible, and the limits of knowledge in the DSC and DCE dynamic MRI experiment. The end-product of a dynamic MRI study is the formation of an inferenceabout the state of the tissue being examined. Inferences and models are intimately connected; model choice is essentially a statement by the investigator as to what state of the tissue is to be inferred, via parametric estimates, from the physical measurement. Elements that define the possible inferences are model choice, the time constants of the system under consideration, the sampling times of the dynamic study, contrast-to-noise in the study, and bias and variance in the model’s parametric estimates. Because of the limited contrast to noise and multiple compartments for contrast leakage, “tapering effects” are an important and substantial source of bias. Data-driven model choice attempts to balance bias and variance.Methods:
All models for estimating vascular parameters using indicator dilution in tissue depend on an estimate of contrast agent concentration, usually (but not always) on a compartmental basis. Contrast agent concentration and MRI contrast are not directly coupled in most tissues. Generally, T2* and T1 contrast mechanisms contribute to MRI contrast and compete with each other. A list of ‘tapering’ effects includes water exchange across cellular membranes, exchange across capillary walls, and blood flow. Sampling strategies include shortening echo times, increasing flip-angles, dual-echo experiments. A major bias in most estimates of vascular parameters is produced by uncertainties in estimating the amplitudeof the AIF. This bias bleeds into every parametric estimate produced by both DSC and DCE experiments. Strategies include automated selection of AIF and selection of reference tissues with known vascular volumes.Results:
A critical survey of recent clinically relevant results for both DSC and DCE, focused on assessments of physiology in tumors, will be presented. Emphasis will be placed on sources of bias and the bias/variance tradeoff. One major difficulty in DCE-MRI in particular is the lack of sampling variance determinations. Where are the test-retest measures?Discussion and Conclusion:
George Box: “All models are wrong – some are useful.”Acknowledgements
No acknowledgement found.References
Suggested reading:
1. Jackson, D.L. Buckley, G.J.M. Parker, Editors. Dynamic Contrast-Enhanced Magnetic Resonance Imaging in Oncology Springer-Verlag – Medical Radiology Series (2005).
2. Sourbron SP, Buckley DL. Tracer kinetic modelling in MRI: estimating perfusion and capillary permeability. Phys Med Biol 2012;57(2):R1-33.
3. Ewing JR, Bagher-Ebadian H. Model selection in measures of vascular parameters using dynamic contrast-enhanced MRI: experimental and clinical applications. NMR Biomed 2013;26(8):1028-1041.