Qualitative or quantitative?
Magnetic Resonance Imaging (MRI) is a useful tool for non-invasive clinical diagnosis and disease monitoring, with image contrast often used to qualitatively highlight tissue types and pathologies. However, MR contrast depends on factors such as patient anatomy, scan protocol settings, and hardware. This can lead to inconsistency between scans and scanning centres. There is growing interest in moving from qualitative images to quantitative measurements of tissue properties, to obtain more information on physiological changes and reduce the influence of other factors that affect the MRI contrast. Quantitative MRI can increase the robustness of tissue classification across different scans and imaging centres, allowing a greater range of perspectives on patient-specific pathological changes.
Weather reporting is a common analogy to highlight the benefits of quantitative information. For example, a qualitative description might say Rome is hotter than London today but would not tell us how much hotter. Alternatively, a quantitative assessment may reveal the temperature to be 25°C (77°F) and 19°C (66°F) for the two cities, respectively. The quantitative information gives us an understanding of the actual temperatures and how they compare. As a Meteorologist takes temperature measurements across an area or a Cartographer assesses a landscape to produce a map, quantitative MR aims to measure and map parameters of interest across a region of interest.
Ideally, any given QMRI method would establish the expected ranges for a particular property value, comparing “healthy” tissue with a specific pathology. Each range would describe the observed biological variation across patients and volunteers. Individual patient measurements could then be compared with these ranges to inform clinical decisions.
Two commonly mapped MR tissue properties are the longitudinal and transverse relaxation times of the tissue magnetisation vector: T1 and T2, respectively. Both properties depend on the local tissue environment and structure on a molecular level, which means the measured T1 and T2 can be used to indicate physiological and pathologic tissue changes.
Synthetic qualitative images
Synthetic qualitative images can be generated using parameter maps and MR physics signal models, mimicking the weighted images of conventional qualitative imaging. Such images remove the need for additional qualitative scans, reducing the total scan session duration. Qualitative images are already established and recognisable to those who are regularly working with clinical images, such as clinicians and radiographers. If desired, additional images with alternative contrast may be generated from the same parameter maps without extra scan time.
Measurement error
As with any measurement science, measurement error, accuracy, and precision are key factors to consider throughout the method design. Ideally, QMRI should be sufficiently sensitive and precise to detect the change caused by the physiological process of interest, while being insensitive to confounding sources of signal variation. Examples include system properties and imperfections, as well as subject variability and movement. System property maps may be used to improve tissue property fitting or may be useful by themselves, depending on the application.
Unfortunately, additional acquisitions or post-processing to mitigate measurement errors can further increase the scan duration. Fast and robust solutions to the problems associated with achieving accurate and precise quantitative parameter maps will help make quantitative MR more commonly used in clinics.
Single-parameter and multi-parameter
Conventional QMRI methods use a separate MR pulse sequence to target each parameter of interest. The gold standard parameter-mapping methods work in this way, running multiple MR scans in series to acquire a series of images, varying a single protocol parameter across the images. A model is fitted to each pixel data series to obtain the parameter of interest.
A major limitation of traditional single-parameter QMRI, particularly the gold standard methods, is that the required acquisition duration is often too long to be feasible for clinical applications. These long scan scans can be due to the need for signal-averaging and serial acquisitions to measure multiple parameters.
Multi-parameter QMRI uses a single acquisition to measure multiple parameters, offering a more efficient use of scanner time. However, these methods must be evaluated to determine any inherent measurement biases or relationships between parameter measurements. Gold standard methods can provide reference parameter maps to test and validate more complicated experimental measurement techniques such as multi-parameter QMRI.
Summary
Quantitative MR has the potential to improve clinical imaging in comparison to traditional qualitative MR but care must be taken to avoid bias and measurement error. For more details on the topics in this article, please view the other articles and links on this site, including the collection of QMRI publications. There is a search bar to search the entire site for any keywords or text. In particular, much of the content in my posts is based on sections of my PhD Thesis [Allen2019].
References and further reading:
- “An Optimisation Framework for Magnetic Resonance Fingerprinting” Jack Allen. Thesis. 2019.
- “The Perfect qMR machine: measurement variance much less than biological variance”. Paul Tofts et al. 2022
- P. S. Tofts. Quantitative MRI of the Brain: Measuring Changes Caused by Disease. John Wiley & Sons, 2003.