Functional magnetic resonance imaging with intermolecular multiple-quantum coherences

Magnetic Resonance Imaging 18 (2000) 489 – 494

Rapid communication

Functional magnetic resonance imaging with intermolecular multiple-quantum coherences Wolfgang Richtera,*, Marlene Richtera,b, Warren S. Warrenc, Hellmut Merkleb, Peter Andersenb, Gregor Adrianyb, Kamil Ugurbilb a

Institut du biodiagnostic, Conseil national de recherches Canada, Winnipeg, Manitoba, Canada b Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, USA c Department of Chemistry, Princeton University, Princeton, NJ, USA Received 28 December 1999; accepted 3 February 2000

Abstract For the first time, we demonstrate here functional magnetic resonance imaging (fMRI) using intermolecular multiple-quantum coherences (iMQCs). iMQCs are normally not observed in liquid-state NMR because dipolar interactions between spins average to zero. If the magnetic isotropy of the sample is broken through the use of magnetic field gradients, dipolar couplings can reappear, and hence iMQCs can be observed. Conventional (BOLD) fMRI measures susceptibility variations averaged over each voxel. In the experiment performed here, the sensitivity of iMQCs to frequency variations over mesoscopic and well-defined distances is exploited. We show that iMQC contrast is qualitatively and quantitatively different from BOLD contrast in a visual stimulation task. While the number of activated pixels is smaller in iMQC contrast, the intensity change in some pixels exceeds that of BOLD contrast severalfold. © 2000 Elsevier Science Inc. All rights reserved. Keywords: fMRI; Intermolecular multiple quantum coherences; Brain function

1. Introduction 1.1. Intermolecular multiple-quantum coherences The phenomenon of intermolecular multiple quantum coherences (iMQCs) was first described a few years ago [1– 4]. In NMR, we can directly observe only single-quantum single-spin coherences (this corresponds to magnetization). Two-dimensional NMR methods [5] make it possible to observe other coherences between states in a multi-spin system as well; these coherences are made to evolve ‘silently’ during a successively incremented time interval and are detected after transformation into magnetization. In order for this transformation to take place, a net coupling between the spins involved must exist. In the case of spins 1/2 (such as the hydrogen nucleus), we can distinguish two types of couplings: scalar couplings, which act through * Corresponding author. Institute for Biodiagnostics, National Research Council, 435 Ellice Ave., Winnipeg MB R3B 1Y6 Canada. Tel.: ⫹1-204-984-6564; fax: ⫹1-204-984-7036. E-mail address: [email protected] (W. Richter).

chemical bonds, and dipolar couplings, which are much stronger than scalar couplings and act through space. Scalar couplings are commonly used to effect the abovementioned transformation of multi-spin coherences into magnetization; hence 2-D NMR method can provide information about the connectivity between spins in a molecule. On the other hand, dipolar couplings are normally not observable in liquids. The dipolar coupling strength between two spins scales as D ij ⬀ 3cos2␪ ⫺ 1 where ␪ is the angle between the interspin vector and the main magnetic field. This coupling averages to zero when integrated over all directions (the surface of a sphere). In the case of short-range dipolar interactions, the interspin vector samples all directions on an NMR time scale through molecular diffusion. This is not true for long-range dipolar interactions, where ␪ is almost constant in time. However, in that case the distribution of spins is quasi-continuous, and the dipolar interactions average to zero in space as long as the liquid is magnetically isotropic. Magnetic field gradient pulses ap-

0730-725X/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S 0 7 3 0 - 7 2 5 X ( 0 0 ) 0 0 1 3 3 - 8

490

W. Richter et al. / Magnetic Resonance Imaging 18 (2000) 489 – 494

plied during the experiment can break this isotropy; hence long-range dipolar couplings can be reintroduced by the experimenter in a controlled fashion. In this manner, the structure of a sample can be probed on the distance scale on which the dipolar couplings act. Most importantly, this distance may be tuned through the choice of experimental parameters. The gradient pulse causing the reappearance of dipolar couplings (the so-called ‘correlation gradient’) can be thought to wind up a helix of magnetization along its axis; the dipolar interactions that are reintroduced in this manner act over a distance scale of approximately one half pitch of that helix. Hence the larger the area under the gradient pulse is, the shorter is the correlation distance. In practice, the correlation distance ranges from tens to hundreds of micrometers, which is of course far above the microscopic range that is provided by scalar couplings, yet below the size of an imaging voxel. The method described here is therefore among the few NMR methods that can provide structural information on a mesoscopic distance scale. 1.2. Functional MRI (fMRI) FMRI was first demonstrated a few years ago [6 – 8] and is today one of the most powerful neuroimaging techniques. FMRI can measure brain activity noninvasively, with a spatial resolution of millimeters and a temporal resolution of seconds. Most fMRI techniques are based on the ‘blood oxygen level dependent’ (BOLD) effect. The BOLD effect is thought to arise from localized changes in the concentration of the strongly paramagnetic deoxyhemoglobin molecules in the brain, which are coupled to alterations in neuronal activity. Blood flow increases within seconds near the site of activation and overcompensates for the increased metabolic demand, resulting in decreased deoxy- and increased (diamagnetic) oxy-hemoglobin contents. The ensuing changes in susceptibility gradients across capillaries and venous blood vessels result in an increase of the apparent transverse relaxation time (T*2) of the spins [9]. Therefore, an image whose intensity is weighted by T*2 will show neuronal activation, through the secondary effect of blood oxygenation, as an increase in signal intensity. The typical signal increase upon activation is on the order of a few percent; hence this method is relatively insensitive. The application of iMQCs to fMRI is based on the rationale that multi-spin coherences themselves do have a different, and possibly higher, sensitivity to susceptibility gradients than single-spin coherences. They may also be more specific to the site of activation. For example, an intermolecular zero-quantum coherence (iZQC) evolves at the difference of the single-quantum frequencies of the two spins involved, hence the zero-quantum signal intensity is a function of the distribution of susceptibility gradients. The BOLD signal, on the other hand, is a function of the average strength of those gradients within a voxel. Hence iZQC contrast is fundamentally different from T*2 contrast [this has been shown previously by our group [3]], but still a function

Fig. 1. iDQC pulse sequence. The iDQC preparation period (including the pair of correlation gradients) is followed by a double spin echo and a segmented EPI readout sequence. Note that the timing is modified from the simplified version that is described in the text, on account of the double spin echo. The effective evolution time after the second gradient is 2␶dq, as required.

of blood oxygenation. Furthermore, the choice of correlation distance provides an additional degree of freedom to optimize contrast, which conventional methods do not possess. Susceptibility gradients due to deoxyhemoglobin arise from blood vessels ranging from densely distributed capillaries to low density large veins. The latter are clearly undesirable in fMRI experiments since they will not have accurate correspondence with the actual sites of enhanced neuronal activity. iMQCs offer the possibility of altering the distance scale over which the dipolar-couplings lead to observable signal, and hence the sensitivity to blood vessels of different sizes. Here, we provide initial evidence that iMQCs (here actually intermolecular double quantum coherences that are, in our pulse sequence, made to depend on the frequency difference between the spins, like iZQCs) can indeed generate large contrast related to functional activation of the brain.

2. Methods 2.1. Hardware Experiments were carried out with a whole-body 7 Tesla imaging system (Varian/Magnex) with a head gradient insert and a double-loop surface coil. 2.2. Pulse sequences The pulse sequence used for iMQC contrast is shown in Fig. 1. It is a modification of previously used pulse sequences, with the goal to minimize signal fluctuations. This sequence was discussed in ref. [3] and selects for doublequantum coherences (iDQCs). Hence, after the first r.f. pulse a double quantum coherence between spins 1 and 2, such as (Ix1Ix2 ⫺ Iy1Iy2) evolves at the sum of the individual resonance frequencies (⌬␻1 ⫹ ⌬␻2), for a duration of ␶dq.

W. Richter et al. / Magnetic Resonance Imaging 18 (2000) 489 – 494

The second r.f. pulse transforms this into two-spin, singlequantum coherences, such as (Ix1Iz2 ⫹ Ix2Iz1); those coherences evolve at the single-quantum frequencies ⫾ ⌬␻1 and ⫾ ⌬␻2, respectively. After a time interval of 2*␶dq, a partial coherence transfer echo forms, with a total phase evolution of ⫾ ␶dq*(⌬␻1 ⫺ ⌬␻2). In order to read out the image, we then used a conventional (segmented) echo planar imaging sequence. Note that the detected signal (including evolution over both time periods) is a function of the resonance frequency difference of the two spins (even though the DQ coherence evolves at the sum of the resonance frequencies). This functional dependence is the same as that of a zero-quantum coherence (iZQC). However, the second gradient, which is not present in the previously published pure zero-quantum version of this pulse sequence, helps cancel spurious magnetization. Relevant pulse sequence parameters were the following: matrix size ⫽ 64 ⫻ 64; TR ⫽ 3.5 s; NEX ⫽ 2 (with phase cycling of the first pulse); 2 segments with linear k-space sampling (hence the total imaging time was 14 s per volume); ␶dq ⫽ 0 (this means that the double quantum signal evolution takes place entirely during the correlation gradient); ␶1 ⫽ 3.8 ms; ␶2 ⫽ 23.7 ms; ␶3 ⫽ 27.5 ms; duration of first correlation gradient ⫽ 8.27 ms; G ⫽ 1.5 G/cm. The correlation distance was, therefore, on the order of 100 ␮m [1]. Conventional BOLD experiments were carried out with a segmented Gradient-Echo EPI pulse sequence: matrix size ⫽ 64 ⫻ 64; TR ⫽ 3.5 s; NEX ⫽ 1; 2 segments with center-out k-space sampling (for an imaging time of 7 s per volume); an effective echo time (to the first echo) of 5 ms; and a nominal echo time (to the center of the echo train) of 19 ms. 2.3. Human subjects Seven normal subjects (six men and one woman) were studied, in accordance with the guidelines of the Institutional Review Board of the University of Minnesota. Informed consent was obtained from all subjects. Technically acceptable data were obtained from five subjects (four men and one woman); only those were considered further. 2.4. Visual stimulation task Monocular or binocular visual stimulation (red LEDs, flashing at 8 Hz) was provided through commercial goggles (Grass Instruments, Quincy, MA, USA). 2.5. Experimental protocol First, a series of anatomic FLASH images was obtained. From those images, a single slice (along one of the Cartesian axes) containing part of the primary visual cortex was selected. Next, a series of iDQC images with different correlation gradients was prepared, in order to assess and evaluate the properties of the iDQC signal. Then, a conven-

491

tional (BOLD) gradient-echo EPI functional imaging experiment was carried out (50 volumes, three rest periods and two stimulation periods of 10 volumes each). After that, an iDQC functional imaging experiment was carried out (25 volumes, three rest periods and two stimulation periods of 5 volumes each). For several subjects, another pair of BOLD and iDQC experiments was carried out, with a different slice orientation. In total, we obtained four coronal, three axial, and one sagittal data set from the five subjects. 2.6. Data analysis Data were analyzed using the software package Stimulate [10] and various software routines written in Interactive Data Language (Research Systems, Boulder, CO, USA). Functional maps were obtained by calculating Pearson’s correlation coefficient between each pixel’s time course and the stimulation (off/on) waveform. Hemodynamic lag was taken into account by shifting the reference waveform by one or two volumes; the highest correlation coefficient resulting from this multiple comparison was retained. Maps were thresholded to an uncorrected confidence interval of 0.01. Additionally, spatial filtering was applied so that only spatial clusters of at least two activated pixels were included in the maps.

3. Results 3.1. Authenticity of the iDQC signal An important property of iMQCs is the overall scaling of the signal by a factor of (3 cos2␪ ⫺1) (this is similar to the scaling factor for dipolar couplings), where in this case ␪ is the angle between the main magnetic field and the correlation gradient direction. Accordingly, we expect that the signal is twice as large for the longitudinal (z) gradient direction as for the transverse (x or y) gradient direction, and that the phase difference between them is ␲. This is shown in Fig. 2. Shown on the top are two phase images with z and x correlation gradients, respectively. Note that the phase difference pixel-by-pixel is approximately ␲ indeed. On the bottom, the average magnitude of the signal in the whole slice is shown. Note that the magnitude for the z gradient is indeed twice that of the transverse gradients. Furthermore, an analysis of signal-to-noise ratios (SNRs) shows that the SNR for the iDQC images is, on average, reduced by a factor of 6.9 from the GE-EPI images; this is on the order of magnitude of our expectations from theoretical considerations [3]. Hence we are confident that the signal comes essentially from iDQCs. 3.2. Anatomic and functional contrast All data sets evaluated showed activation throughout the occipital lobe in both the BOLD and the iDQC experiments. A representative data set is shown in Fig. 3. On the left, we

492

W. Richter et al. / Magnetic Resonance Imaging 18 (2000) 489 – 494

Fig. 2. Phase and magnitude of the iDQC signal for two different correlation gradient directions. Note that the the signals for longitudinal and transverse gradients are ␲ out of phase, and that their intensity is different by approximately a factor of two, in accordance with the theoretical model for iDQCs.

show a conventional GE-EPI image in gray scale. Superimposed on it, in color, is the corresponding BOLD activation map. On the right, we show an iDQC image in gray scale, with the iDQC activation map in color superimposed. In both images and maps, Gaussian smoothing (interpolation) was performed for display purposes only. Note that the

anatomic contrast is fundamentally different between the two images, as we expect according to the mechanism postulated. The two activation maps are qualitatively different from one another as well; the iDQC map is considerably more sparse, and the foci of activation are only partially overlapping.

Fig. 3. BOLD and iDQC activation maps on a background of GE-EPI and iDQC contrast. Note that the anatomic contrast is fundamentally different, as are the functional maps.

W. Richter et al. / Magnetic Resonance Imaging 18 (2000) 489 – 494

493

Table 1 Activited pixels in the ROI with either method alone and in both methods simultaneously Subject #

Total ROI iDQC BOLD Overlap

1 (sag) both eyes

1 (cor) both eyes

1 (ax) both eyes

2 (cor) both eyes

3 (ax) left eye

3 (ax) right eye

4 (cor) both eyes

5 (cor) both eyes

557 18 241 15

694 37 276 26

1091 12 320 7

962 5 83 1

927 11 119 4

927 6 115 4

806 28 288 12

723 2 67 0

In order to quantify this result, we calculated four parameters for each map. These are shown in Table 1. Here we indicate the slice orientation (coronal, axial, sagittal) and the stimulation (left eye, right eye, both eyes). We also give the number of pixels in the region of interest (ROI), which was the part of the occipital lobe intersected by the slice. We then give the number of pixels activated in the iDQC and BOLD experiment, respectively, and the overlap of those two sets. Note that in all subjects but one, some but not all pixels activated in the iDQC experiment are also activated in the BOLD experiment. An important issue of interest is the intensity change upon activation. Inspection of the data sets revealed that the baseline fluctuations were considerably larger for the iDQC signal than for the BOLD signal. This is not unexpected, given the presence of the two gradient pulses, introducing a potential source of instability that is absent in the BOLD experiment. At the same time, the signal intensity change for the activated pixels was larger (on the order of 10% increase from baseline) for the iDQC signal as well. Thus, the smaller “activation” volumes in the iDQC data may initially be thought to be an artifact of the method of analysis, arising from the larger overall signal fluctuations. We investigated this by comparing the pixels that were commonly activated in both experiments (corresponding to the last row in Table 1). In Fig. 4, we graph the relative intensity change of

Fig. 4. BOLD and iDQC signal intensity changes in commonly activated pixels. Note that the iDQC intensity changes are generally larger, at least for this set of parameters.

those pixels in both methods. Note that for almost all pixels the iDQC intensity change is larger than the BOLD intensity change (these are the points below the x ⫽ y line). The analysis reveals no significant correlation between these two quantities (confidence interval ⬎ 0.2). However, the average intensity change in these pixels was 10.0% for the iDQC signal, as compared to 4.2% for the average BOLD signal change. In Fig. 5, we show timecourses from commonly activated pixels in one subject. Horizontal bars denote the activation periods. Again, the signal intensity change is much larger in the iDQC experiment than in the BOLD experiment, while the baseline fluctuation is larger as well. Note that these time courses are not representative of pixels with the highest correlation in either method. 4. Discussion This experiment constitutes the first demonstration of functional activation revealed by iMQCs. At this moment, we lack a physiological model to relate the observed activation to blood flow, blood oxygenation, blood volume, vascular structure, oxygen consumption, and other potentially relevant parameters quantitatively; this is also largely true for the BOLD mechanism. We find it particularly noteworthy that the foci of activation are partially, but not completely, congruent in the two methods. This observation strongly suggests that iDQC contrast is fundamentally different from BOLD contrast, but is also sensitive to changes subsequent to neuronal activation. These results are in accordance with our expectations. The functional signal change in commonly activated pixels was on average two to three times larger in the iDQC method compared to BOLD for the particular echo time used in the BOLD experiments. This number has to be considered carefully, as a fair comparison between the two methods is by no means straightforward. Signal intensity changes depend on a variety of factors, many of which are rooted in the technical details of each experiment, and not in a physiological phenomenon. In this preliminary effort, a systematic comparison of “contrast-to-noise” was not performed either for the BOLD or for the iDQC studies. This would also require that differential sensitivity to different vessel sizes is considered, and that the comparison is performed when the two methods are displaying changes cou-

494

W. Richter et al. / Magnetic Resonance Imaging 18 (2000) 489 – 494

neuronal activity. Future work will have to explore methods to increase the stability of the signal toward physiological and instrumental fluctuations, and the influence of the correlation distance on the observed contrast. The iMQC method may become an invaluable tool for the elucidation of brain structure and function.

Acknowledgments

Fig. 5. Time courses from activated pixels for BOLD (bottom) and iDQC (top) methods. Note that the signal intensity change is larger in the iDQC time course, as are the baseline fluctuations.

Supported by NIH National Resources grant RR08079 (University of Minnesota), the Keck Foundation, and NIH GM 35253 and the McKnight Foundation (Princeton University).

References pled to neuronal activity for similar types of blood vessels. In this respect, it would be more appropriate to compare spin-echo BOLD studies with the current iDQC experiment. However, we can confidently state that there exist pixels that are functionally activated, in which the iDQC signal shows a larger intensity change than the BOLD signal. The activation maps are much more sparse in the iDQC method. The reason for this may be purely that the signal is much less stable in the present implementation of the iDQC method, thereby obscuring true activation. This could be rectified in future studies, for example with the use of navigator-echo type approaches. However, this observation may also reflect true differences in the physiological parameters that actually influence the signal. It should also be pointed out that the correlation distance, which is possibly the most crucial parameter determining contrast, was not optimized or investigated in this preliminary study. A careful investigation and optimization of the iDQC approach in detecting signal changes coupled to neuronal activity remains to be performed. However, we find it encouraging that a large signal change was observed in the iDQC method.

5. Conclusion We demonstrate here that the iMQC method yields signal changes in the brain that are coupled to alterations in

[1] Warren WS, Richter W, Hamilton Andreotti A, Farmer BT II. Generation of impossible cross peaks in solution nuclear magnetic resonance. Science 1993;262:2005. [2] Richter W, Lee S, Warren WS, He Q. Imaging with intermolecular multiple quantum coherences. Science 1995;267:654 –7 . [3] Warren WS, Ahn S, Rizi R, Hopkins J, Leigh JS, Richter W, Mescher M, Garwood M, Ugurbil K. MR imaging contrast enhancement based on intermolecular zero quantum coherences. Science 1998;281:247– 51. [4] Lee S, Richter W, Vathyam S, Warren WS. Quantum treatment of the effects of dipole-dipole interactions in liquid nuclear magnetic resonance J Chem Phys 1995;105:874 –900. [5] Jeener J. Ampere International Summer School: Baski Polje, Yugoslavia, 1972. [6] Bandettini PA, Wong EC, Hinks RS, Tikofsky RS, Hyde JS. Time course EPI of human brain function during task activation. Magn Reson Med 1992;25:390 –7. [7] Kwong KK, Belliveau JW, Chesler DA, Goldberg IE, Weisskoff RM, Poncelet BP, Kennedy DN, Hoppel BE, Cohen MS, Turner R, et al. Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc Natl Acad Sci USA 1992;89: 5675–9. [8] Ogawa S, Tank DW, Menon R, Ellermann JM, Kim S-G, Merkle H, Ugurbil K. 4 Tesla gradient recalled echo characteristics of photic stimulation-induced signal changes in the human primary visual cortex. Proc Natl Acad Sci USA 1992;89:5951–5. [9] Ugurbil K, Ogawa S, Kim S-G, Chen W, Zhu XH. Imaging brain activity using nuclear spins. In: Maraviglia B, editor. Magnetic Resonance and Brain Function: Approaches from Physics. Amsterdam: BIOS Press, 1999. p. 261–310. [10] Strupp, JP. A GUI based fMRI analysis software package. Neuroimage 1996;3:S603.