In vivo multiple spin echoes imaging of trabecular bone on a clinical 1.5 T MR scanner

Transcription

1 Magnetic Resonance Imaging 20 (2002) In vivo multiple spin echoes imaging of trabecular bone on a clinical 1.5 T MR scanner S. Capuani a, G. Hagberg b, F. Fasano b, I. Indovina b, A. Castriota-Scanderbeg b,c, B. Maraviglia*, a a Istituto Nazionale Fisica della Materia (INFM) UdR Roma1, E. Fermi Center and Physics Department University La Sapienza, Rome, Italy b Laboratory of Functional Neuroimaging, Fondazione Santa Lucia, IRCCS, Rome, Italy c Department of Radiology, Fondazione Santa Lucia, IRCCS, Rome, Italy Received 30 June 2002; accepted 10 September 2002 Abstract In vivo multiple spin echoes (MSE) images of bone marrow in trabecular bone were obtained for the first time on a clinical 1.5 T scanner. Despite of a reduced sensitivity of the MSE trabecular bone images with respect to the cerebral matter ones, it is possible to observe some features in the MSE trabecular bone images that may be useful in the diagnosis of osteopenic states. Two different CRAZED-type MSE imaging sequences based on spin-echo and EPI imaging modalities were applied in phantom and in vivo. Preliminary experimental results indicate that EPI imaging readout seems to conceal the MSE contrast correlated with pore dimension in porous media. However it is still possible to detect anisotropy effects related to the bone structure in MSE-EPI images. Some strategies are suggested to optimize the quality of MSE trabecular bone images Elsevier Science Inc. All rights reserved. Keywords: MSE (multiple spin echoes); EPI; Trabecular bone; Bone marrow; Contrasted-imaging; Susceptibility effects; Porous spacing 1. Introduction Recently, there has been much interest in MRI based on the contrast arising from multiple spin echoes (MSE) [1] or intermolecular multi-quantum coherences (imqcs) [2], a manifestation of the dipolar field generated by residual intermolecular dipolar couplings in liquids. This occurs in the presence of a correlation magnetic field gradient with amplitude G c that breaks the magnetic isotropy of the sample so that intrinsic long-range dipolar couplings give rise to the refocusing of a signal. Hence the correlation gradient generates a magnetization helix resulting in spatial modulation of the sample magnetization. In heterogeneous systems, the amplitude of the MSE signal depends on sample heterogeneity over a correlation distance d c /( G c t) [2] (where t is the gradient pulse duration [1,3]) which is half a cycle of the magnetization helix, thus providing a novel contrast mechanism that can be tuned to a specific length scale. * Corresponding author. Tel.: ; fax: address: bruno.maraviglia@romal.infn.it (B. Maraviglia). Various MR techniques based on MSE effects, such as the CRAZED sequence [2], have been applied for characterizing tissues and heterogeneous systems [3 10]. Furthermore functional magnetic resonance imaging (fmri) and self-diffusion weighted imaging have been demonstrated using imqcs [11,12]. Despite of a lower signal-to-noise ratio (SNR) with respect to the single-quantum images, the imqcs images have revealed features that were undetectable with conventional MRI. IMQCs images are characterized by a contrast based on the dipolar interaction correlation distance as well as by a higher sensitivity to magnetic susceptibility distributions due to the dipolar demagnetizing field. For morphologic studies of trabecular bone, multiple spin-echoes (MSE) offers an alternative contrast mechanism that depends on the porosity of the bone, as recently demonstrated ex vivo at 7 T[8 10]. In particular, two competing contrast mechanisms are active in porous media. One is directly related to the mean number of coupled spins at the correlation distance d c since only the magnetization arising from spins at a distance d c is detected. The other mechanism depends upon the susceptibility differences between coupled spins at this distance. In X/02/$ see front matter 2002 Elsevier Science Inc. All rights reserved. PII: S X(02)

2 624 S. Capuani et al. / Magnetic Resonance Imaging 20 (2002) particular, susceptibility effects in porous media produce a drop of the signal intensity when the correlation distance d c corresponds to the pore diameter. In previous studies, we have shown a distinct relation between the porous structure and the correlation distance d c, by using bovine trabecular bone of the calf [8 10]. In particular, by varying the gradient strength and/or the duration of the gradient pulse, the MSE image contrast could be varied and unique information about the trabecular bone microstructure could be obtained. Especially, by using this approach, the pore size of the trabecular bone could be measured quantitatively and the data obtained from the MSE images were consistent with those obtained by scanning electron microscopy (SEM). Thus, these previous high-field studies [8 10] demonstrated the feasibility of such an approach, which however could become impractical at lower field strengths used in clinical routine. Particularly, the successful implementation of MSE imaging at 1.5 T, has the potential to provide useful information in bone disease. In the contest of trabecular bone imaging, an important issues regards what kind of image encoding that is to be used. The imaging version of the CRAZED-type sequence is obtained by adding a singleshot echo planar (EPI) [13,14] read-out [4,6] or a conventional sequential line-scan (SE) technique [3,7 9,15]. The advantage of the EPI read-out is based on its speed that compensates for the high number of signal averages necessary to obtain a sufficient SNR in the MSE image. However since k-space coverage with EPI is obtained by a series of image gradient reversals the resulting image will be weighted in T 2 * and signal drop-outs and distortions due to susceptibility artifacts at tissue interfaces may hamper an accurate interpretation of the image and its contrast. This effect may be particularly bothersome in tissues with rapid T 2 * decay, such as trabecular bone. In this paper, we outline some strategies that we have adopted to optimize MSE imaging of bone marrow in the trabecular bone at 1.5 T. A feasibility analysis of the MSEimaging method is presented and CRAZED-type sequences with different image encoding are discussed and applied on a phantom and in vivo on a clinical 1.5 T MR scanner. In trabecular bone and porous media phantom, we found that the T 2 * weighted image obtained by using EPI read-out seems to conceal the novel MSE contrast linked to the porous dimension in the CRAZED-type sequence. 2. Methods and material Fig. 1. CRAZED-type imaging sequences based on conventional sequential line-scan (SE) and EPI read-out. The primary sequence (a) consists of two slice-selective RF pulses, 90 and 120 spaced by the delay, combined with a pair of pulsed correlation gradients with amplitude G and 2G respectively. The modified sequence (b) has an additional 180 RF pulse at a delay TE/2 after the DQC signal. Two CRAZED-type imaging sequences, both based on a conventional sequential line-scan (MSE-SE) and EPI (MSE-EPI) read-out, were implemented on a Siemens Vision MR-scanner operating at 1.5 T (Siemens Medical System, Erlangen, Germany). The images were obtained by using a circular polarized volume head-coil for RF transmission and reception. The first sequence (Fig. 1a) consisted of two slice-selective RF pulses (90 and 120 ) spaced by a delay time combined with a pair of magnetic gradient pulses, with an integral ratio (gradient amplitude multiplied by time) of 1:2 in order to select multiple spin echo of order 2 (Double Quantum Coherences, DQCs) [3]. The modified CRAZED sequence (Fig. 1b), had an additional 180 RF pulse at a delay TE/2 after the second multiple spin echo that refocuses magnetic field in-homogeneities. The correlation gradients were applied in three directions with respect to the main magnetic field: along, perpendicular to and at the magic angle to the main field. Control MSE imaging experiments were performed on a phantom characterized by a uniform pore size distribution. The phantom was made by a 100 mm diameter hollow plastic sphere filled with 1 mm diameter glass beads and tap water. The interstitial space between the close-packed spheres formed a regular lattice of 0.5 mm diameter connected pores. K-space was sampled by line-scan and EPI read-outs on a grid and the field of view was 192 mm yielding a final resolution of mm. Other imaging parameters were: 20 ms, TR 10 s, number of averages 32 and G 1, 2.3, 4, 10 mt/m. MSE images were obtained both in vivo at the level of the distal femoral metaphysis in three healthy volunteers (2 males, aged 30 years and 35 years; 1 female, aged 28 years), after an informed consent, and ex vivo at the level of the proximal third of the femoral shaft of a calf. For the in vivo experiments, 8 or 16 signal averages were collected and a matrix was used. The correlation gradient strength G was 2, 4 or 10 mt/m and 20 ms. For the ex vivo experiments, between 16 and 256 signal averages were collected with a matrix size of or The correlation gradient was varied between 1 mt/m and 10 mt/m and the delay time between 20 ms and 40 ms. For both in vivo and ex vivo imaging the slice thickness was 8 10 mm, the field-of-view (FOV) 192 mm, and the repetition time (TR) 5 s.

3 S. Capuani et al. / Magnetic Resonance Imaging 20 (2002) A single type of MSE experiments was repeated several times on the same volunteer and on different volunteers to investigate intra and inter subject variability. Furthermore to make sure that the signal obtained with the imaging sequence described in Fig. 1 was unquestionably related to DQCs, the correlation gradient was varied. Specifically, we controlled that a zero signal intensity was obtained for the gradients applied at the magic angle and that half the signal intensity was obtained for the gradients perpendicular to the static magnetic field, as theoretically predicted. Conventional SE and EPI images with matrix-sizes and echo-times that matched those of the MSE images were also obtained with different matrix sizes. Furthermore the T 2 and T 2 * values were determined using conventional MRI. The imaging parameters used for the T 2 -weighted SE images, obtained by a double turbo spin echo sequence, were: TR 3800 ms, TE 22, 90 ms, resolution mm; while for the T 2 *-weighted FLASH (Fast Low Angles Shot) images the following parameters were used: TR 5000 ms, TE 4, 7, 10, 12 ms, resolution mm. Finally, diffusion weighted images (SE-EPI, TR 4000 ms, TE 98 ms, b-values 0, 500, 1000 s mm 2, resolution: mm) were also acquired. In vivo T 2 and T 2 * values for bone marrow vary from 20 ms to 80 ms and from 3 ms to 24 ms respectively, depending on the investigated region of trabecular bone. 3. Results To demonstrate the sensitivity of MSE imaging to microstructure, MSE images of a spherical phantom characterized by pore structures with a diameter of about 0.5 mm were obtained at 1.5 T. As recently demonstrated at 7 T [9], the MSE signal intensity of that phantom should exhibit a signal intensity drop when the correlation distance d c,related to the duration and amplitude of the applied correlation gradient, matches the interstitial pore diameter. To investigate the MSE signal variation, MSE-SE and MSE- EPI images without the 180 refocusing pulse (i.e., the CRAZED-type sequence shown in Fig. 1a) were obtained as displayed in the upper (MSE-SE) and lower (MSE-EPI) rows of Fig. 2. For the MSE-SE images, a bell-shaped signal variation was observed when the correlation distance was varied from 1.2 mm to 120 m, going from high to low to intermediate and back to high signal levels (Fig. 2a through c). This signal variation corresponds to a minimum when the correlation gradient duration multiplied by the gradient amplitude corresponds to a correlation distance that is on the order of the phantom pore size, i.e., d 0.5 mm. This signal intensity behavior would be unexpected in an ordinary spin echo experiment and confirms the ability of the MSE contrast mechanism to detect pores of dimension similar to d c. In the lower row of Fig. 2, MSE-EPI images are shown that were obtained with the same parameters as those used Fig. 2. Sensitivity of MSE contrast to microstructure dimension at 1.5 T. A phantom characterized by uniform pore size of 0.5 mm was measured by MSE-SE (a through d) and by MSE-EPI (e through h) at various settings of the correlation gradient. For MSE-SE a bell-shaped signal variation with varying correlation distance de, exhibiting a minimum at a de matching the phantom pore size (b) was obtained while for MSE-EPI the signal decreases for increasing correlation gradients. for the MSE-SE measurements. In this case, local field inhomogeneities caused by the magnetic susceptibility effects of the sample dominate the signal and consequently the expected MSE intensity modulation is obscured and the signal intensity simply decreases as the correlation gradient intensity increases (Fig. 2e, f, g, h). This EPI readout effect on MSE images was verified in ex vivo and in vivo. In Fig. 3, example in vivo MSE images obtained by the sequential readout (MSE-SE, first row) and the EPI readout Fig. 3. In vivo images of distal femoral metaphysis. Because of the low SNR of MSE images, all the images have a mask to emphasize the image and trabecular bone region. MSE images were obtained by MSE-SE (a through c) and by MSE-EPI (d through f) for various settings of the correlation distance: 650 m (a and d), 350 m (b, e) and 130 m (c, f). In the bottom row an MSE-EPI image obtained by modified CRAZED (Fig. 1b) and d c 650 m is shown (g). In h and i conventional spin echo images obtained at matrix size of (h) and (i) are shown.

4 626 S. Capuani et al. / Magnetic Resonance Imaging 20 (2002) Fig. 4. Signal intensity variation as a function of correlation distance measured in vivo. The MSE signal measured in the ROIs depicted in Fig. 3h is plotted as function of the correlation gradient for the MSE-SE (a) and MSE-EPI (b) sequences. Note the pronounced MSE signal variation, linked with the trabecular bone microstructure, that can be detected by MSE-SE. (MSE-EPI second row) at varying correlation gradient strengths can be compared. Depending on the trabecular bone region, the MSE-SE image obtained with a gradient strength of 4 mt/m showed a drop in signal intensity compared to the MSE-SE images obtained at 2 and 10 mt/m respectively. On the contrary, this effect was not observable in the MSE-EPI images. We repeated this series of experiments in the three volunteers several times and in all cases a minimum in signal intensity was detected in the MSE-SE image obtained with a correlation gradient strength of 4 mt/m that corresponded to a correlation distance of 300 m. In Fig. 4, two plots of the signal intensity as a function of the correlation gradient intensity obtained by MSE-SE (Fig. 4a) and MSE-EPI (Fig. 4b) in various regions of interest (ROIs) are shown. The selected ROIs are highlighted in the conventional high-resolution spin-echo image (bottom row of Fig. 3), together with a single quantum spin-echo image at a matrix size comparable to the MSE images (32 32 matrix). In the images obtained by sequential read-out gradients (Fig. 4a) a marked drop in signal intensity at a correlation gradient equal to 4 mt/m is evident in two of the investigated ROIs (depicted with red and green color). Furthermore, an increase in signal intensity as the gradient increased from 4 mt/m to 10 mt/m is visible in the cyan colored ROI. Contrasting these results, the measurements performed by MSE-EPI (Fig. 4b) yielded a decrease in signal intensity when the correlation gradient strength increased, as would be expected in a conventional experiment based on single-quantum coherences but not in a MSE measurements. This observation was confirmed in all investigated ROIs, excepted one that exhibited the MSE-specific contrast variation (red-colored ROI). Furthermore, we investigated the effect of a 180 refocusing pulse on the MSE-EPI contrast (CRAZED-type sequence shown in Fig. 1b). The effect of the refocusing pulse was an improved SNR (Fig. 3g) that can be compared with the SNR that is obtained without 180 pulse (Fig. 3a, sequence shown in Fig. 1a). Fig. 5 shows three MSE-SE images obtained with correlation gradient strength equal to 2 mt/m b), 4 mt/m c) and 10 mt/m d), corresponding to correlation distances equal to 600 m, 300 m and 120 m respectively. Conventional T 2 and T 2 *maps at the same anatomic locations are also depicted in a) and e) respectively. The different contrast shown by the MSE images compared with conventional T 2 and T 2 * weighted images is evident. More specifically, again the image obtained at 4 mt/m shows a drop in signal intensity compared with the MSE-SE images obtained at 2 and 10 mt/m respectively, depending on the trabecular bone region. Furthermore, in orange and green colored ROIs signal intensity at 10 mt/m is higher than that at 2 mt/m. Than, the trabecular bone of the ROIs depicted in Fig. 5f is in general characterized by more porous structures of 300 m diameter and secondary by more porous structures of 600 m diameter than that of 120 m diameter. Fig. 6 shows a set of representative MSE-EPI maps obtained at a correlation gradient equal to 2 mt/m along z (a), x (b) and y (c) direction along with conventional EPI (d), Flash (e), ADC [17] (f) and anisotropic diffusion images [17] along z (g), x (h), and y (i) direction. In order to show an image contrast dependence on sample anisotropy, NS 16 and NS 64 have been used for MSE images with G c along z and x, y direction respectively. The MSE images (a, b, c) show a net anisotropy of the signal depending on trabeculae orientation which is not evident in anisotropic diffusion weighted images (g, h, i). Only when the correlation gradient and the diffusion gradient are along the y direction (images c and i respectively) a minimum in signal intensity in trabecular bone region were visible. By comparing the images in Fig. 6a, b, c with those in Fig. 6d, e, f, g, h, i, it is evident that MSE contrast provides different information with respect to conventional EPI, Flash and diffusion imaging contrast. 4. Discussion and conclusion In the present work, CRAZED-type sequences have been used to obtain, for the first time, in vivo trabecular bone MSE images at 1.5 T. The intrinsic features of the MSE contrast, that enables the detection of the pore size of trabecular bone, may prove useful in the diagnosis of the osteopenic diseases. To start with, we verified the feasibility of detecting the relation between the correlation distance d c and the pore dimension in a porous system, at a magnetic field strength of 1.5 T, as previously observed at 7 T. However, while MSE imaging of the human brain is easy to obtain with a sufficient SNR and within a reasonable acquisition time [4,6,7, 15], trabecular bone images suffer from a low SNR, owing to the more rapid T 2 and T 2 * relaxation decay of the protons

5 S. Capuani et al. / Magnetic Resonance Imaging 20 (2002) Fig. 5. In vivo images of distal femoral metaphysis. MSE images obtained by MSE-SE for correlation distance equal to 600 m b), 300 m c) and 120 m d). Note the variation of contrast with the correlation distance and the different contrast with respect conventional T 2 and T 2 * images reported in a) and e) respectively. The images have a mask to emphasize the image and the trabecular bone region and they were obtained at matrix size of In f) high-resolution image with highlighting ROIs and signal intensity variation as a function of correlation distance. The MSE signal measured in the ROIs depicted in Fig. 5f is plotted as function of the correlation gradient strength. Note minima in signal intensity obtained at correlation gradient strength of 4 mt/m (that corresponded to a correlation distance of 300 m) in orange red and green colored ROIs. Particularly in orange and green colored ROIs, signal intensity at 10 mt/m is higher than that at 2 mt/m. in the bone marrow compared to those of gray and white matter of the brain. For this reason, appropriate MSE sequences, in terms of sequence timing and k-space encoding must be carefully selected. Among the CRAZED-type sequences that we analyzed, we found that an improvement in the SNR can be obtained by adding a 180 pulse; in fact, because of the refocusing pulse, the T 2 * decay does not affect the signal. Due to the strong susceptibility difference between bone marrow and trabecular bone, which causes a rapid single and double quantum T 2 * relaxation [16], it is mandatory to keep the delay time as short as possible and to limit the length of the correlation gradient pulse accordingly, in order to maximize the SNR and minimize magnetic field in-homogeneity effects. Regarding k-space encoding, we analyzed MSE images obtained by using both sequential line scan and EPI read-out. With the latter, imaging time is significantly reduced, but apart from this, EPI read-out does not seem to be indicated for MSE imaging of the bone marrow protons. In fact, because of the reversals of the read-out gradients, standard trabecular bone EPI images suffer from distortions and artifacts due to the large susceptibility difference between bone marrow and cancellous bone. This effect is so important that the expected MSE signal modulation arising as a result of the porosity of the porous medium is obscured. Despite this finding, dipolar demagnetizing field contrast is still considerable in MSE- EPI images, allowing the detection of a net dependence of the MSE contrast on the orientation of the correlation gradient with respect to the main magnetic field. Then the anisotropy of trabecular bone structure is well emphasized in the MSE-EPI images. A dip in MSE signal intensity has been observed in repeated experiments of the distal femoral metaphyses of the volunteers when the correlation distance d c was about ( ) m (where 200 m is due to lack of data). Likewise, a dip in the MSE signal intensity was observed in repeated experiments on the ex vivo trabecular bone sample at the level of the proximal third of the femoral shaft of a calf when the correlation distance d c was about m. In humans, the average values of trabecular bone ranged between 300 mto800 m depending on the exact location of the trabecular bone region (S. Majumdar, private communication). Thus, our procedure based on the MSE technique, seemingly underestimates the effective trabecular pore dimension. This may be explained by the presence of strong internal gradients, as demonstrated in previous works [9], where MSE measurements of the femoral shaft were performed before and after replacement with water of the bone marrow of the trabecular bone. This substitution resulted in a shift in the position of d c minima as a function of the correlation gradient duration pulse, due to the change in internal susceptibility at the bone-bone marrow and bonewater interfaces. Likewise, a similar mechanism linked with

6 628 S. Capuani et al. / Magnetic Resonance Imaging 20 (2002) Fig. 6. In vivo MSE-EPI maps of distal femoral metaphysis obtained at a correlation gradient equal to 2 mt/m along z (a), x (b) and y (c) direction along with conventional EPI (d), Flash (e), ADC (f) and anisotropic diffusion images along z (g), x (h), and y (i) direction. Note in MSE images a net anisotropy of the signal depending on trabeculae orientation which is not evident in anisotropic diffusion weighted images. the susceptibility mismatch between bone marrow and cancellous bone was observed in vivo, resulting in an interference with the effect of the correlation gradient resulting in a dip which is also a function of susceptibility effects. These results indicate that further investigations are required to validate the correspondence between signal minima and pore size distribution. The main limitation of MSE imaging at 1.5 T is the low SNR, leading to poor image quality or excessively long imaging time due to signal averaging with respect to single quantum imaging. As a result, the sensitivity at 1.5 T is dramatically reduced compared to that obtained at 7 T. Furthermore, in trabecular bone imaging, the sensitivity is lower than that obtained in the brain due to a more rapid T 2 and T 2 * decay. In this work a net difference between MSE images obtained with sequential line scans and EPI read-out has been outlined. To our knowledge this has never been proved before, because only when the correlation distance approaches the pores dimension in a porous media the effect is visible. More specifically, when the correlation distance d c is greater than the sample microstructures, the sample is considered an homogeneous one (as the brain tissues in MSE imaging experiments reported in literature [2 4,6 7]), and only the contrast mechanism related to the mean number of spins coupled at that correlation distance is present. In this case, MSE-SE and MSE-EPI images probably don t provide significative differences in MSE image contrast. On the contrary, when the correlation distance d c approaches the sample microstructures the contrast mechanism related to the pores dimension is also present. As this contrast mechanism is strictly linked to local susceptibility effects at the bone marrow-trabecular bone matrix interface, the interference of EPI read-out action disturbs the MSE signal, obscuring this particular contrast. Since one of the goals of the present work was to evaluate the difference between line scan and EPI read-out, the minimum delay time was set to 20 ms, due to limitations in maximal gradient strength (20 mt/m), and gradient switching time, that sets a lower limit to the effective echo time, defined as the time needed to cover half of the k space in a single-shot (8 ms). This time delay may not be optimal for trabecular bone imaging due to the rapid bone marrow T 2 * decay that varied between 3-24 ms, although it is still possible to observe important features in MSE trabecular bone images. In future studies, optimization of these parameters will allow an improvement in SNR, a sine-qua-non condition for clinical use of this method in the diagnosis of osteopenic disease states. Such optimization together with a more widespread clinical use of higher magnetic field strengths and more rapid and stronger gradients, may allow the development of MSE imaging into an important complementary tool for investigating bone pathologies. In conclusion, the results reported in this study suggest that new experiments have to be carried out to probe structural features in the trabecular bone. In particular, because MSE images show a contrast that depends upon the orientation of the correlation gradient, we suggest that both MSE and standard anisotropic diffusion sequences are used for investigating trabecular bone. Parallel information about proton diffusion and proton dipolar correlation distance along different directions may in this way be obtained that could reinforce the traditional interpretation of trabecular bone images. References [1] Bowtell R, Bowley RM, Glover P. Multiple spin echoes in liquids in a high magnetic field. J Magn Reson 1990;88: [2] Richter W, Lee S, Warren WS, He Q. Imaging with intermolecular multiple quantum coherences in solution nuclear magnetic resonance. Science 1995;267: [3] Mori S, Hurd RE, van Zijl CM. Imaging of shifted stimulated echoes and multiple spin echoes. Magn Reson Med 1997;37: [4] Warren WS, Ahn S, Mescher M, Garwood M, Ugurbil K, Richter W, Rizi RR, Hopkins J, Leigh JS. MR imaging contrast enhancement based on intermolecular zero quantum coherences. Science 1998;281: [5] Bifone A, Payne GS, Leach MO. In vivo multiple spin echoes. J Magn Reson 1998;135:30-6. [6] Rizi RR, Ahn S, Alsop DC, Garrett-Roe S, Mescher M, Richter W, Schnall MD, Leigh JS, Warren WS. Intermolecular zero-quantum coherence imaging of the human brain. Magn Reson Med 2000;43:

7 S. Capuani et al. / Magnetic Resonance Imaging 20 (2002) [7] Zhong J, Zhong C, Kwok E. In vivo intermolecular double-quantum imaging on a clinical 1.5 T MR scanner. Magn Reson Med 2000;43: [8] Capuani S, Alessandri FM, Maraviglia B, Bifone A. Multiple spin echoes for the evaluation of trabecular bone quality. Proceedings of the European Society for Magnetic Resonance in Medicine and Biology. Paris, September 14 17, 2000, p. 51. [9] Capuani S, Curzi F, Alessandri FM, Maraviglia B, Bifone A. Characterization of trabecular bone by dipolar demagnetizing field MRI. Magn Reson Med 2001;46: [10] Capuani S, Alessandri FM, Bifone A, Maraviglia B. Multiple spin echoes for the evaluation of trabecular bone quality. MAGMA 2002; 14(1):3-9. [11] Richter W, Richter M, Warren WS, Merkle H, Andersen P, Adriany G, Ugurbil K. Functional magnetic resonance imaging with intermolecular multiple-quantum coherences. Magn Reson Imaging 2000;18: [12] Zhong J, Zhong C, Kwok E, Kennedy S. Enhanced sensitivity to molecular diffusion with intermolecular double-quantum coherences: implications and potential applications. Magn Reson Imaging 2001; 19:33-9. [13] Mansfield P. Multi-planar image formation using NMR spin echoes. J Phys C 1977;10:L55. [14] Mansfield P, Morris PG. NMR imaging in biomedicine. New York: Academic Press, [15] Charles-Edwards GD, Payne GS, Leach MO, Bifone A. Intermolecular double-quantum coherence imaging of brain tumours at 1.5 T. Proc Intl Soc Mag Reson Med 2001;9:922. [16] Chen Z, Zheng S, Chen Z, Zhong J. Quantitation of transverse relaxation time of intermolecular multiple-quantum coherences. Proc Intl Soc Mag Reson Med 2001;9:852. [17] Le Biham D, editor. Diffusion, and perfusion magnetic resonance imaging. New York: Raven Press Ltd., 1995.