04-06 Practical Measurements of T1 and T2
Relaxation times can be measured in different ways with various degrees of accuracy.
04-06-01 In vitro Determination
High-resolution magnetic resonance spectroscopists have measured T1 values since the middle of the last century. The in vitro measurement is done on a small sample, approximately 0.1-1.0 ml or slightly larger in volume, in an extremely homogeneous magnetic field.
A variety of methods has been developed to obtain maximal precision with minimal time consumption. Typically, 15 to 30 magnetization measurements are performed on the sample for different time delays, TI in inversion-recovery experiments or TR in partial saturation experiments. Based on these results, an observed T1 value is calculated, and the error limits are usually better than 5%.
T2 can be calculated with a single multiecho sequence. The more echoes one uses, the more accurate the measurement will be.
Calculations based on fast pulse sequences (i.e., not "clean" sequences other than IR or SE) lead to rough estimates of T1, T2, T2* (and proton density) values. They might be "reproducible" when repeated, but the use of relaxation time values acquired with such pulse sequences is not advisable for scientific or clinical comparisons.
04-06-02 In vivo Determination
Magnet systems with larger bores allowed the examination of whole organisms, animals, and people, and a more physiological determination of relaxation time values than those of excised organs or tissues. Relaxation time measurements were considered very important during the first years of commercial MR imaging. All machines were programmed to create true T1 and T2 images, based on SE and IR sequences. However, soon it became clear that relaxation time values were not the claimed invaluable addition to diagnostics.
Localization. One of the major problems of in vivo relaxation time measurements is the localization of the volume to be observed. Details of such localization techniques are given in Chapter 6. Actual accuracy of in vivo measurements depends on the number of points acquired and the quality of localization. Localization is relatively uncomplicated in little or non-moving organs such as the brain, but demanding and partly impossible (in particular at high fields) in organs with complex movement patterns such as the heart.
Relaxation time values and proton density calculation. The current most dependable method used to obtain a T1 image ("T1 map"), i.e., an image whose picture elements represent pure T1 values, relies on a mathematical manipulation of separately obtained images with different T1 influence. Measurements are easier and more accurate at low and medium fields, because T1 values are shorter, ECG triggering is less complicated, and artifacts less pronounced at these fields.
Typically, two to four images are used and the signals mathematically processed to calculate pure T1 values. Bearing in mind that in vivo relaxation can be multiexponential, it is somewhat inadequate to perform the analysis by such a limited fit to an exponential curve. T2 images are calculated from the images of a multiecho series, e.g. CPMG. In clinical settings, usually four or eight echoes are applied.
Usually, diffusion, flow, and multiexponential decays are not taken into account in the fits and noise as well as motion artifacts add to inaccuracies.
Matrix size and slice thickness as well as partial volume effects are limiting factors in relaxation-time measurements in vivo. Partial volume effects and other factors influence the measurements. Variations within the same lesion related to vascularity, necrosis, and cell behavior (macroscopic compartmentilization) contribute to the overlapping of relaxation times values. All methods relying on slices through the examined object will have as additional error source partial volume effects from the edges of the slices; the only method which avoids this slice problem is the true 3D volume imaging method. Standard deviation in fitting, artifacts, and variations in the selection of volume elements by the operators are all possible sources of error (Figure 04-22).
Furthermore, similar lesions may have a more than single exponential relaxation rate, e.g., brain tumors and multiple sclerosis plaques. This is not unexpected, considering the heterogeneous nature of tumors. Reproducibility of such measurements is also limited.
 |
 |
Figure 04-22: |
The multilayered complexity of factors and features influencing and creating relaxation time and proton distribution changes is not completely understood yet [⇒ Springer]. A simplistic view offered by Koenig suggests that water molecules can wander, by thermally-induced diffusion, rather extensively throughout the intra- and extracullular regions of tissue, and that the exploration is rather thorough in a time of the order ot T1 (or even T2). Another concept is the highly structured water, restrained for a significant time in a geometry defined by various ionic and molecular constituents of the cytoplasm [⇒ Koenig 1985, 1988].
However, some features of the T1-dispersion do not fit easily into these concepts, for instance cross relaxation phenomena that lead to quadrupolar dips in the T1-dispersion plot. They are dependent on field strength and temperature (Figure 04-23).
More about the dependence of relaxation times on static field strength and its influence upon contrast can be found in Chapter 10.
|
|
Figure 04-23:
The dispersion of T1 in tissues (ms) versus field strength (log Tesla) is not as monotonic and smooth as shown e.g. in Figure 04-04 and Figure 10-16. This nuclear magnetic relaxation dispersion (NMRD) curve with higher resolution of a multiple sclerosis tissue sample reveals two dips (quadrupolar dips) at 0.0505 and 0.0660 T (2.1 and 2.8 MHz) where the otherwise steady increase of T1 is interrupted (after ⇒ Rinck 1988).
Faster data acquisition. Precise measurements require long acquisition times; the repetition time, TR, should be equal to or greater than 5×T1. At 0.15 T, the T1 of myocardium is around 380 ms, at 1.5 T it has climbed to around 1000 ms. Such measurements at low fields take approximately 5 minutes, at high or ultra-high field more than 10, perhaps 15 minutes. Thus, faster acquisition methods were sought and developed.
Fast acquisition of quantitative T1 maps can, e.g., be based on a series of snapshot fast low-angle shot (FLASH) images after inversion of the magnetization [⇒ Deichmann]. Such techniques were for instance used for estimating the concentration of paramagnetic contrast agents in an organ.
Since the acquisition of quantitative tissue data from a beating heart has to be very fast, lately much research is focused on modifications of a pulsed NMR sequence proposed by David C. Look and Donald R. Locker in 1969. MRI did not exist at that time, and Look and Locker used their time-saving one-shot method for NMR spectroscopy instead of the conventional methods to measure the T1 relaxation time. The spectroscopic “LL” method was within 10% of the conventionally precise-calculated value [⇒ Look+Locker].
In the 1980s, the method was further developed for MRI by Graumann and his colleagues [⇒ Graumann]. Others followed and precision was waived for speed. Among the modified sequences for cardiac and, e.g., brain MRI experimentally (and sometimes clinically) used today, one finds PURR [⇒ Lee], MOLLI [⇒ Messroghli], and ShMOLLI [⇒ Piechnik]. They all suffer to varyiing extent from errors, resulting in an underestimation of the true T1. "Apparent" T1 values of MOLLI and ShMOLLI measurements of, e.g., normal myocardium have an error range of 30% or higher. A number of different pulse sequences, e.g., SASHA, SAPPHIRE, DESPOT and many others were also being tested.
Imperfect spoiling of transverse magnetization at higher flip angles in gradient echo sequences has a negative effect on a precise estimation of signal intensity and other parameters, such as relaxation times [⇒ Zur].
Critical remarks. Cardiologists are using "apparent" T1 for "cardiac mapping" and T2* (in reality T2app) for measuring cardiac iron overload, which according to them is, so far, one of the most useful approaches of modern cardiac medicine. The sharp fall of T2* values at high and ultra-high fields is related to the drastic rise of magnetic susceptibility effects which grow linearly with magnetic field strength; pure T2 is not affected in the same way.
In some publications, additional terms are also used wrongly or confusingly, for instance T1* (a term which is not appropriate because T1 is not affected by susceptibility effects) for an apparent T1 (T1app or T1influx).
From a scientific point of view, these measurements are wrong because the margin of error is huge. However, in medicine to be imprecise does not preclude specific use.
Enter: the cardiologists. Welcome to the scientific arena.
A comment.
And a follow-up: The return of MR fingerprinting.
A comment.
Quantification of MR Parameters and Synthetic Images. Relaxation time and proton density values can be used to create synthetic or simulated images for training and teaching purposes.
04-06-03 T1 (T2) Image and Weighted Images
In clinical routine, people often talk about T1, T2 or proton-density images. The right terms should be T1-weighted, T2-weighted, and proton density (ρ)-weighted (or better, intermediately weighted) images, because these images have only a certain T1, T2, or proton-density dependence. However, they are not calculated pure relaxation time or proton density images. Chapter 10 will explain this in detail.
Figures 04-22 c and d are T1-weighted and T2-weighted images, to be compared with the pure T1 and T2 images of Figure 04-24a and 04-24b.
|
|
Figure 04-24:
Top row: Images of a patient with a polyp in the left paranasal sinus; bottom row: images of a cervical spine.
(a) Calculated (pure) T1, and (b) calculated (pure) T2 image. Figures (c) and (d) show T1- and T2-weighted images. Pure T1 and T2 images are of very limited diagnostic value. Multiparameter weighted images are far more valuable for clinical diagnosis and commonly used in patient studies.
Simulation software: MR Image Expert®
04-06-04 Measurements in Medical Diagnosis
Fifteen years after the first description of different relaxation behavior in tissues by Erik Odeblad [⇒ Odeblad], other researchers started postulating that relaxation times differentiate tumors from normal tissue since most T1 (and in a similar way T2) values of pathologic tissue can differ markedly from the T1 of the similar normal tissue [⇒ Damadian] (cf. History of MRI).
However, the ability to discriminate, type, or even grade tumors using relaxation time values has remained a dream, despite the sophisticated multi-point fits introduced over the years. Figure 04-25 shows that there are differences between, in this case, T2 of normal and diseased tissues. Although values of T2 are more accurate than those of T1 because more points are used for their calculation, these differences are not significant between T2 values of, for instance, tumors and edema or infarction.
|
|
|
Figure 04-25: |
Every year, the literature announces new attempts to exploit relaxation-time measurements in vivo. There are some positive reports about its successful use. Most concern follow-up of therapy, with patients being their own reference. Publications include, for instance, the report that relaxation times from leukemic bone marrow can be used for the differential diagnosis of this disease (Figure 04-26) [⇒ Jensen]. Similar results in high-grade gliomas have been published by another research group [⇒ Boesinger].
|
|
|
Figure 04-26: |
Another relaxation-time study is the measurement of normal appearing white brain matter in multiple sclerosis (MS) patients. Pixel-by-pixel mapping suggests that there could be minute invisible changes in the white matter, which might explain brain function deficits that cannot be explained by the size and location of visible MS plaques [⇒ Barbosa; ⇒ Lacomis; ⇒ Rinck 1987].
Yet, the follow-up of treatment based upon relaxation-time values is difficult and in most instances dubious (Figure 04-27). A rise and subsequent decline of relaxation time values after a local intervention might rather indicate edema and inflammation than successful treatment [⇒ Zhang 2014].
|
|
|
Figure 04-27: |
After absolute T1 and T2 values had been used unsuccessfully by researchers, combinations of T1 and T2, histogram techniques, and more sophisticated three dimensional display techniques of factor representations were used [⇒ Skalej].
Availibility of databases of in vivo relaxation-time measurements is very limited. The largest collection of data was published by Bottomley et al. [⇒ Bottomley 1984, 1987]. Unfortunately, these values are not very reliable. A comparison between in vivo and in vitro relaxation measurements is quite difficult because many T1 relaxation time values change rapidly after excision. Only brain tissues reveal a relatively stable relaxation behavior after they have been removed from the body [⇒ Fischer 1989].
Critical Remarks: Biomarkers. To add confusion to complicated science, changes of terms and terminology are common in contemporary bioscience. Many physiological and biological measurements and calculations are subsumed under new terms such a "radiomics", "molecular imaging", "celluar imaging", "personal imaging", "precision imaging", and others. Thirty years after the description of relaxation times and relaxation rates as possible or questionable biological indicators they are re-ranked among biological markers or biomarkers. None of these terms describes any novel scientific discipline, but rather existing specialized applications of certain measurement techniques.
In general, biomarkers are biological indicators of any kind; there are thousands of them. They are not specific for MR imaging or MR spectroscopy. Typical biomarkers are measurements or scores such as blood pressure, body temperature, or the body mass index.
In MR imaging, biomarkers break down into numerous subgroups where they can be applied standing alone or several combined, relaxation times being only one of them. Aside of T1 and T2, there are other possible indicators for the detection, diagnosis, and monitoring of treatment, i.e., of particular physiological or disease states. Quantification of MR parameters is discussed in Chapter 15. Biomarkers extracted through image segmentation and multispectral analysis are also described in Chapter 15, those acquired with the help of contrast agents in Chapter 13 and by dynamic imaging in Chapter 16.
