IRBM 29 (2008) 366–374
Original article
Low field magnetic resonance imaging in rat in vivo Imagerie in vivo du rat par résonance magnétique nucléaire à bas champ E. Breton, C. Goetz, P. Choquet, A. Constantinesco ∗ Service de biophysique et médecine nucléaire, CHU de Hautepierre, 1, avenue Molière, 67098 Strasbourg, France Received 31 August 2007; accepted 29 May 2008 Available online 10 July 2008
Abstract Objective. – This work aims at demonstrating the interest of low field magnetic resonance imaging (MRI) for in vivo imaging in rat. Material and methods. – MRI is performed using an open resistive 0.1 T magnet with a homogeneous zone measuring 6 cm × 10 cm × 10 cm. Rats under isoflurane gaseous anesthesia are placed in a technical cell for small animal imaging. Radio frequency solenoidal coils are developed for each application. Gradient echo T1 and T2 -weighted sequences are programmed to acquire 3D data in vivo in rats. Results. – Acquisition times are comprised between 1 min for scout sequences and 2 h 10 min to acquire volumes with in-plane pixel size ranging from 0.37 to 1 mm and a slice thickness between 0.37 mm and 1.88 mm. Information obtained from our acquisitions are comparable to those reported in the literature about five applications: 1: intracerebral glioma; 2: cerebral contrast enhancement using manganese; 3: description of knee joint anatomy; 4: cardiac dynamic acquisition synchronized on cardiac rhythm; 5: whole body (thorax and abdomen) acquisition. Conclusion. – Rat MRI using an 0.1 T device allows to answer to the biological questions studied in this work at the expense of long acquisition times compared to high field MRI. Low field MRI offers economical and technical characteristics that could make it attractive to a non-specialized scientific community aiming at realizing daily MRI in vivo in rat. © 2008 Elsevier Masson SAS. All rights reserved. Résumé Objectif. – Démontrer l’intérêt de l’imagerie par résonance magnétique (IRM) à bas champ pour l’imagerie in vivo du rat. Matériel et méthodes. – L’étude est réalisée à l’aide d’un aimant résistif ouvert à 0,1 T présentant un volume homogène de 6 cm × 10 cm × 10 cm. Les rats sous anesthésie gazeuse à l’isoflurane sont mis en place dans une cellule technique pour l’imagerie. Des antennes solénoïdales radiofréquences sont développées pour chaque application. Des séquences d’écho de gradient pondérées T1 ou T2 , adaptées à l’imagerie du rat, permettent de réaliser des acquisitions 3D. Résultats. – Des volumes 3D ayant une taille de pixel dans le plan comprise entre 0,37 et 1 mm et une épaisseur de coupe allant de 0,37 à 1,88 mm sont acquis dans des temps d’acquisition d’une minute pour les séquences de positionnement à deux heures dix minutes. Les informations déduites des acquisitions sont comparables à celles rapportées dans la littérature pour les cinq problématiques abordées : 1 : étude d’un gliome intracérébral ; 2 : amélioration du contraste cérébral par administration de chlorure de manganèse ; 3 : description de l’anatomie normale de l’articulation du genou ; 4 : acquisition dynamique synchronisée au rythme cardiaque ; 5 : acquisition corps entier (thorax et abdomen). Conclusion. – L’IRM du rat à 0,1 T permet de répondre aux questions biologiques posées dans ce travail moyennant des temps d’acquisition longs par rapport à ceux obtenus avec des appareils à haut champ. L’IRM à bas champ présente des caractéristiques économiques et techniques propre à une diffusion dans une communauté scientifique non spécialiste afin de réaliser au quotidien de l’IRM in vivo du rat. © 2008 Elsevier Masson SAS. All rights reserved. Keywords: Low field MRI; Rat; Small animal imaging; In vivo imaging; Cost Mots clés : IRM bas champ ; Rat ; Imagerie du petit animal ; Imagerie in vivo ; Coût
∗
Corresponding author. E-mail address:
[email protected] (A. Constantinesco).
1959-0318/$ – see front matter © 2008 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.rbmret.2008.05.002
E. Breton et al. / IRBM 29 (2008) 366–374
1. Introduction
2. Methods
Small animal models of human pathologies constitute a necessary step between in vitro experiments and first studies in human. Classical longitudinal follow-up methods are performed ex vivo and need regular sacrifice of animals. In comparison, the in vivo longitudinal imaging of individually identified animals greatly improves the understanding of therapeutics action while allowing to decrease the number of used animals [1]. Small mammals are well known animal models with cheap breeding costs. Rats, although their size is favorable for surgically induced models, remain small for in vivo imaging compared with humans. For all imaging modalities, technological improvements have been developed due to anatomical and physiological distinctive features of rats [1]. The need to check during acquisitions the physiological state of the animal under anesthesia is less taken into consideration but should also be addressed. Magnetic resonance imaging (MRI) with natural high soft tissue contrasts achieves a sub-millimetre three-dimensional (3D) resolution that makes it a proper method to study in vivo some models in rats. Nowadays, two kinds of MRI systems are used for rat imaging: dedicated high magnetic field magnets (hfMRI) [2–7] rather present in specialized imaging centres for small animal imaging and clinical MRI (cMRI) devices widely spread but with accessibility for small animal imaging limited in time [8–13]. Low field MRI devices have been very little considered yet for in vivo MRI in rat [14]. However, they have technical and economical characteristics that make them affordable for a daily use in a large research community [15] at the expense of long total acquisition times (TAT). Excluding spectroscopic studies, low field MRI devices allow in vivo imaging in rat in numerous routine biological problems, with TAT of the order of an hour. This article aims at showing the performance of a 0.1 T MRI device for in vivo imaging in rat. Some typical applications (linked with biological questions) were illustrated: visualization of cerebral tumor, anatomical description of knee joint, study of cardiac dynamic and whole body imaging.
2.1. Animal preparation
367
Experiments were led in agreement with the French legislation (authorization numbers A 67-482-20 and 67–104) in four Wistar male rats, weighting from 250 to 400 g, anesthetized with isoflurane (1.5 to 2% pushed by air). During acquisitions, animals were placed inside a technical cell for small animal imaging (Équipement vétérinaire Minerve, Esternay, France). This cell (Fig. 1B) does not contain ferromagnetic elements that could interact with the main magnetic field of the MRI. The animal head was placed in an anesthesia mask with a bite bar. The technical cell limits direct manipulations of the animal and facilitates its positioning in the imaging device. The cell is closed with a cylindrical tube and is warmed up thanks to air circulating in a closed circuit with controlled temperature. A solution of MnCl2 was used for cardiac dynamic imaging experiments. Manganese exhibits a cardiovascular toxicity upon chronic or acute exposure so that the injected manganese dose has to be limited and progressively administered [16–19]. In the literature, manganese doses ranging from 25 mol kg−1 [20] to 420 mol kg−1 [21] are used in rodent and no alteration in the cardiac dynamic is observed with slow intravenous injections. We chose a manganese dose of 200 mol kg−1 , injected via the intraperitoneal route as 1 mL of MnCl2 diluted in physiological serum, without observing any modification in the ECG. The cerebral tumor model was induced with a stereotaxic injection of C6 glioma cells two weeks before imaging. 2.2. 0.1 T MRI device, coils and acquisition parameters The 0.1 T main magnetic field is produced using a watercooled resistive magnet (Bouhnik SAS, Vélizy-Villacoublay, France) [22]. The 800 kg-magnet measures 70 cm × 70 cm in width and 105 cm in height. The electrical and water-cooling supplies of the magnet and gradient coils consume, respectively, 3230 W and 3 L min−1 . The magnet is open on three sides and its air gap measures 18 cm in height (Fig. 1A). The
Fig. 1. A: the low field 0.1 T MRI open magnet (1), its electrical supply (2), the technical cell for small animal imaging (3) and the gaseous anesthesia ensemble (4) placed next to the low field magnet. B: details of the technical cell for small animal imaging placed in the 0.1 T magnet. The RF coil (2) coupled to the pick-up coil (3) slided along the isolation chamber (1), face mask (4) for gaseous anesthesia, technical part (5) with the gaseous anesthesia supply (6) and the heating air circuit and regulation (7).
368
E. Breton et al. / IRBM 29 (2008) 366–374
Table 1 Characteristics of RF coils and acquisition parameters in each imaging experiment
RF coil Diameter (mm) Length (mm) Wire diameter (mm) Number of turns Q
Glioma
Knee
Cardiac gated
Whole body
36 41 5 6 430
25 26 2 5 260
60 34 5 4 550
65 130 5 14 600
Sequence
FLASH
CEFAST
FLASH
CEFAST
FLASH
FLASH
Sequence TE (ms) TR (ms) ␣ (◦ ) NEX TAT Acquisition matrix FOV (mm2 ) Number of slices Thickness (mm)
9 22 60 36 41 min 128 × 96 50 × 50 32 60
14 26 70 36 48 min 128 × 96 50 × 50 32 60
8 100 80 4 1 h 02 min 128 × 96 48 × 48 96 48
22 31 80 24 1 h 55 min 128 × 96 54 × 54 96 54
8 18 60 8 2 h 10 mina 128 × 48 128 × 64 64 64
8 100 80 4 1 h 02 min 128 × 96 128 × 128 96 128
0.39 × 0.39 × 1.875 mm3
375 m isotropic
375 m isotropic
422 m isotropic
1 mm isotropic
1 mm isotropic
Reconstructed voxel
Abbreviations used: RF coil quality factor (Q), number of excitation (NEX), total acquisition time (TAT), field of view (FOV). a The cardiac gated acquisition is performed in two interleaved acquisitions of eight cardiac phases (see Fig. 2).
homogeneous ± 5 ppm zone is ellipsoidal 10 cm wide and 6 cm high. The magnet 5-Gauss-line is located at 15 cm from the magnet side. Since the main magnetic field is vertical, solenoidal coils that offer proper homogeneity of the RF pulse can be used for imaging. Transmit/receive solenoid coils were designed for each application in order to maximize the filling factor (Table 1). Coils resonate at 4.26 MHz, which is the Larmor frequency of hydrogen at 0.1 T. They are coupled to preamplifier through inductive coupling with a loop slided along the cover of the technical cell for small animal imaging (Fig. 1B). 3D gradient echo T1 -weighted Fast Low Angle SHot (FLASH) [23] with radiofrequency (RF) spoiler and T2 -weighted Contrast Enhanced Fast Acquisition in STeady state (CE-FAST) [24] sequences were developed on a SMIS console (Guildford, UK), which replaced the original one in 1998. Acquisition parameters for the imaging experiments are detailed in Table 1. Acquisition parameters leading to isotropic or near isotropic voxels were preferred to limit partial volume effects and to allow for volume reorientation. Fast scout acquisitions lasting 1 to 2 min were used to check the animal positioning.
acquired for eight cardiac phases of the same cardiac cycle with TR equals 18 ms between acquisitions. When the next QRS complex was detected, the same line of the Fourier plane was acquired for the eight complementary cardiac phases delayed of TR/2 equaling 9 ms. This scheme was repeated every two cardiac cycles for each line of the Fourier volume, which brought to the acquisition of 16 cardiac phases. In every acquisition, the observation time was 8 ms ensuring the temporal independence of acquired cardiac phases. In order to minimize the influence of cardiac cycle variations, an approximate 10 ms margin remained free of acquisition at the end of the cardiac
2.3. Cardiac synchronization Two subcutaneous ECG needles were placed at chest level. The MRI acquisition was synchronized to the QRS complex through the TTL output of an electrocardiograph (Physiogard RSM784, Bruker/ODAM, Wissembourg, France). Taking into account a mean cardiac rhythm of 370 beats per minute (bpm), the cardiac cycle was divided in 16 phases separated by 9 ms. This delay was too short compared to our minimum achievable TR, so we chose that the acquisition took place in a two-interleaved-acquisition scheme (Fig. 2). After a first synchronization, a same line of the Fourier plane was
Fig. 2. Scheme of the cardiac gated acquisition. Sixteen cardiac phases are acquired in two interleaved acquisitions to describe the cardiac cycle. Acquisitions are synchronized to the rat ECG. Eight cardiac phases are acquired in each cardiac cycle (acquisition 1 and acquisition 2): Pi, with i ranging from 1 to 16; odd i in acquisition 1 (P1, P3 to P15); even i in acquisition 2 (P2, P4 to P16). The first RF pulse of acquisition 1 starts directly at the synchronization to acquire the first cardiac phase P1. The corresponding data of acquisition 2 are acquired after the next synchronization and delayed of TR/2 equaling 9 ms compared to acquisition 1. This scheme is repeated every two cardiac cycles for the acquisition of each line of the Fourier volume. The observation time is 8 ms ensuring the temporal independence of acquired cardiac phases.
E. Breton et al. / IRBM 29 (2008) 366–374
369
Fig. 3. Orthotopic glioma induced in rat. Four successive craniocaudal axial slices out of 32 T1 -weighted (A: FLASH sequence, TAT 41 min) and T2 -weighted (B: CE-FAST sequence, TAT 48 min) slices (reconstructed voxel: 390 m × 390 m in plane, ST 1.88 mm) showing, respectively, the deviation of brain ventricles (1) to the left and the inflammatory reaction (2).
cycle. This prospective synchronization brought directly a data set corresponding to determined phases of the cardiac cycle. 2.4. Data processing Images were processed using ImageJ (Rasband WS, ImageJ, US National Institute of Health, Bethesda, Maryland, USA, http://rsb.info.nih.gov/ij/, 1997–2007). The cerebral tumor presented a hyperintense signal on T2 -weighted images that allowed a selection with an upperthreshold. Voxels with a gray level higher than the threshold were selected in the successive reconstructed slices. The number of selected voxels multiplied by the voxel’s volume gave the volume of the tumor. The reconstructed volumes of knee, whole body and heart acquisitions, made of isotropic voxels, were reoriented, respectively, along their main reference axes. The heart was reoriented and sliced along its three anatomical axes in order to obtain three sets of slices (short axis, horizontal long axis and vertical long axis). Following the MnCl2 injection, the cardiac muscle appeared in a hypersignal in the data sets. A threshold scaled to 50% of the maximum was applied to the three sets. The plane of valves was manually drawn to close the ventricular contour. The internal surface of this delimited area was measured in all slices. The obtained total surface area corresponded thus to the mean of the surface area calculated along the three anatomical axes [25]. It was multiplied by the slice thickness to calculate the left ventricular volume. The ejection fraction was calculated as the difference between volumes of end of diastole and end of systole, divided by the end of diastole volume, and expressed as a percentage. 3. Results The natural T1 contrast images (Fig. 3A) show a deviation of brain ventricles to the left due to the development of the tumor. The T2 -weighted images (Fig. 3B) allow to visualize the inflammatory reaction and necrosis in the brain right hemis-
phere. We calculate a tumor volume of 505 mm3 based on the T2 images. Reoriented T1 -weighted slices (Fig. 4A) describe the anatomy of the right knee joint: epiphyses of femur and tibia, growth plates, patella, patellar ligament, infrapatellar fat pad. The T2 -weighted acquisition shows the synovial fluid (Fig. 4B). For cardiac acquisitions, the end of diastole and end of systole left ventricular volumes measured are, respectively, 640 and 120 L (Fig. 5A and B) and the corresponding ejection fraction is calculated as 81%. The 16 phases acquired after the injection of MnCl2 allow to follow the temporal evolution of the left ventricular volume during the cardiac cycle (Fig. 5C). Fig. 6 presents orthogonal slices from an isotropic acquisition of the rat thorax and abdomen. The field of view (FOV) measures about 10 cm and depicts the main anatomical structures from the skull basis to the lower part of the abdomen: heart, lungs, spinal cord, liver, gallbladder, perirenal fat, kidneys, bladder. 4. Discussion 4.1. Biological applications of low field MRI Examples detailed in this article concern typical fields of application of MRI in vivo in rat for which low field MRI can be used with TAT ranging from 41 min to 2 h 10 min. Natural T1 contrasts (Fig. 3A) depict the consequences for the cerebral tissue of the mechanical compression due to the tumor growth [3,26]. T2 -weighted images (Fig. 3B) reveal the inflammatory reaction and necrosis. 3D isotropic acquisitions make possible the simple measurement of the tumor volume [3,6,11,14,25]. This exam might be completed by an administration of gadolinium contrast agent that modifies the T1 tumor contrast when the blood brain barrier is broken [14,27]. The T1 and T2 -weighted 3D isotropic reconstructed volumes can be freely reoriented to study the anatomy of knee [4,28,29]. It is then useless to multiply scout sequences and the total positioning time is significantly reduced.
370
E. Breton et al. / IRBM 29 (2008) 366–374
Fig. 4. Rat knee. Reoriented sagittal, coronal and axial T1 -weighted (A: FLASH sequence, 128 slices, TAT 1 h 02 min, reconstructed voxel: 375 m isotropic) and T2 -weighted (B: CEFAST sequence, 128 slices, TAT 1 h 55 min, reconstructed voxel: 422 m isotropic) slices showing: epiphyses of femur (1) and tibia (2), their growth plates (3), patella (4), patellar ligament (5), infrapatellar fat pad (6), synovia (7). * Deformations of skin and subcutaneous fat due to the RF coil.
Studying cardiac dynamic remains difficult with MRI [30]. The mean cardiac rhythm of the normal rat anesthetized with isoflurane is approximately 350 bpm as reported in the literature [31–33] and its cardiac muscle measures about 15 mm in diameter [2]. In comparison to human cardiac dynamic MRI, both temporal and spatial resolutions have to be increased and the S/N ratio is consequently reduced [2]. The use of manganese as a contrast agent raises the T1 contrast of the cardiac muscle and facilitates the selection of the ventricle in order to calculate the volume of the left ventricular chamber (Fig. 5A and B). We obtain an ejection fraction of 81% in the normal rat, that is in agreement with values obtained in the litera-
ture using cardiac gated MRI [9,32,34,35]. Using hfMRI or cMRI, 2D slices of ten to 26 phases per cardiac cycle are acquired in TAT of 2 min per slice (Table 2) [2,9,31,34]. Even if the 3D isotropic acquisition takes a longer TAT, it allows to make a direct measurement of the left ventricular volume rather than an evaluation based on a geometric ellipsoidal model. The size of the homogeneous zone of the 0.1 T MRI device used in this study allows to acquire a FOV equal to the largest ones (10 cm in length) obtained in dedicated high field magnets for whole body acquisitions in rat (Table 2) [5,10,36,37]. Such an isotropic acquisition obtained in 1 h makes a whole body
Fig. 5. Cardiac gated acquisitions following MnCl2 injection. Mean cardiac rhythm 370 bpm. A: orthogonal reoriented slices in end of diastole (FLASH sequence, 64 slices, reconstructed voxel: 1 mm isotropic). Small axis (SAX) from apex to basis, horizontal axes (HLA) and vertical axis (VLA). B: orthogonal reoriented slices in end of systole (FLASH sequence, 64 slices, reconstructed voxel: 1 mm isotropic). Small axis (SAX) from apex to basis, horizontal axes (HLA) and vertical axis (VLA). C: temporal evolution of the left ventricular volume in 16 cardiac phases. Ejection fraction is 81%.
E. Breton et al. / IRBM 29 (2008) 366–374
371
Fig. 6. Anatomic “whole body” (thorax and abdomen) isotropic acquisition. Sagittal (A and B), coronal (C and E) and axial (D) slices obtained from the acquired volume (FLASH sequence, 128 slices, TAT 62 min, reconstructed voxel: 1 mm isotropic). Ribs (1), kidneys (2), liver (3), lungs (4), heart (5), fore paw (6), cerebellum and spinal cord (7), trachea (8), vertebra (9), perirenal fat (10), bladder (11).
anatomical registration with a functional information possible as we demonstrated in mouse [27].
to 0.5 T. The value of the main magnetic field used for rat imaging increased with the one of the cMRI devices used [41,42] up to 1.5 T. The first high magnetic field dedicated MRI devices appeared in the late 1980s [43,44] and quickly reached 7 T. While main magnetic fields increased, their homogeneity along with RF coils, acquisition electronics and RF pulse sequences improved and led to increasing S/N ratios in shorter acquisition times.
4.2. MRI systems used for in vivo rat imaging The first rat imaging experiments in the 1980s were done using early clinical systems [38–40] based on low fields inferior
Table 2 Rat imaging experiments and acquisition parameters depending on the main magnetic field of the MRI device: data collected from the cited literature compared with those obtained at 0.1 T W
Pixel size (m2 )
ST/gap (mm)
Vvoxel (mm3 )
SAT (s)
Nslices
TAT (min)
T2 T1 T2 T1 /T2 T1 /T2 T1 /T2
234 × 234 120 × 120 235 × 235 195 × 195 313 × 417 390 × 390
1/0 1/1.2 1/1.2 2/0 3/0 1.88/0
0.0548 0.0144 0.0552 0.0760 0.392 0.286
≈60 ? ? ? ≈60 76/90
9–11 11 11 13/19 8 32
11 ? ? ? 7 41/48
T1 T1 /T2 T1 T1 T2
109 × 109 547 × 547 391 × 391 375 × 375 422 × 422
0.146/0 1.5/0 1/0 0.375/0 0.422/0
0.00173 0.449 0.153 0.0527 0.0752
38 13/26 ? 38 71
96 32 64 96 96
61 7/14 ? 62 115
Cardiac 7 T [34] 2 T [2]
T1 T1
1.5 T [9] 0.1 T
T1 T1
310 × 310 195 × 195 175 × 175 560 × 560 1000 × 1000
1/0 2/? 1/? 2/X 1/0
0.0961 0.0760 0.0306 0.627 1
? ? ? 86 122
18–22 12 ppcc ? 16 ppcc 3 14 ppcc 64 16 ppcc
? 15 53 ≈6 130
Body 4.7 T [36] 2 T [5] 1.5 T [10] 0.1 T
T1 T1 T1 T1
391 × 391 254 × 254 391 × 463 1000 × 1000
3/1 3/0 ?/X 1/0
0.459 0.194 ? 1
2–4 ? 200 38
? 36 1 96
≈1 ? 3 62
B0 [ref.] Glioma 7 T [6] 2.35 T [3] 1.5 T [12] 1.5 T [11] 0.1 T Knee 7 T [4] 2 T [28] 1.5 T [13] 0.1 T
Abbreviations used: main magnetic field (B0 ), sequence weighting (W), slice thickness (ST), volume of voxel (Vvoxel ), slice acquisition time (SAT), number of slices acquired (Nslices ), total acquisition time (TAT). The number of cardiac phases acquired per cardiac cycle is specified in the Nslices column for the cardiac gated acquisitions (ppcc). A “?” indicates that the data is not given and cannot be derived from the cited reference. A “X” indicates that the interslice gap is not a relevant parameter because only independent slices are acquired.
372
E. Breton et al. / IRBM 29 (2008) 366–374
These improvements benefit to all MRI devices whatever their magnetic field value. In MRI, the maximum achievable spatial resolution is determined by the intensity of the gradients and by the S/N ratio. Independently of the main magnetic field, the maximum amplitude and rising time of the gradient coils play a major role in the acquisition speed and in the theoretical limit of resolution. On the contrary, the S/N ratio is almost proportional to the main magnetic field of the magnet. An increase in the magnetic field is directly reflected in the S/N ratio and allows to improve the resolution or to decrease the NEX. Since the theoretical S/N ratio is proportional to the square root NEX, a rise in the magnetic field allows to dramatically diminish the NEX and consequently the TAT, while keeping the S/N ratio constant. This property explains the usual technical choice of continuously increasing the main magnetic field values in MRI devices. The use of MRI devices with a magnetic field inferior to 1.5 T in rat is nowadays extremely rare. Concerning the example of in vivo applications exposed in this article, there is to our knowledge only one recent reference of 0.3 T MRI in vivo in rat [14]. Preclinical imaging in rat mainly takes place in dedicated hfMRI devices (ranging from 2 to 11.7 T in references cited in this article, see Table 2), or in a lower extend, in 1.5 or 3 T cMRI devices that are available for this use. A comparison between the MRI acquisition parameters for the applications exposed in this article is given Table 2 depending on the MRI device type. Dedicated hfMRI devices developed for small animal imaging allow to realize MRI acquisitions in vivo in rat with an in-plane pixel size ranging from 100 to 547 m (Table 2) and a slice thickness between 146 m and 3 mm (Table 2). To achieve these voxels’ size, the gradient coils of these dedicated devices are more intense and faster than those of cMRI. They benefit from the proportionality between the main magnetic field and the S/N ratio, but also from dedicated RF coils. TAT are in the order of magnitude of 10 min for 3D acquisitions covering the rat head. In parallel, the difficulty in maintaining the homogeneous zone increases with the main magnetic field value. In dedicated high magnetic field magnets, the size of the homogeneous zone is restricted to the minimum needed in small animal imaging and its accessibility is limited. Consequently, the animal positioning is difficult and direct visual control of the animal is impossible. Moreover, dedicated hfMRI devices are expensive in installation and operation. Thus, the generalization of this complex technique is limited. Dedicated hfMRI devices are mostly installed in specialized small animal imaging centres that can provide the necessary technical support for their use. Whole body cMRI devices are widely available in number and offer an alternative to dedicated magnets. They might be accessible for rat imaging outside of their clinical use time, but strict occupational hygiene must be observed [11]. Consequently, they do not offer a sufficient availability and their use is limited to punctual experiments. In practice, some RF coils designed for the clinical use with small field of view (small coil in diameter, surface coil or flexible coil) are adapted to rat imaging [11,12]. Specific receive RF coils might also be designed, while using the human whole body coil to ensure a homogeneous excitation. Acquisition parameters must be optimized for rat
imaging. The modifications of the sequence parameters consist in the reduction of the minimum voxel size and/or the increase of the maximum NEX and hence TAT. In-plane pixel sizes ranging from 195 to 560 m and slice thickness between 390 m and 3 mm are obtained with cMRI devices (Table 2). There is a twofold increase in the in-plane dimension of the smallest voxels compared to hfMRI. Using 1.5 or 3 T cMRI devices, the TAT are equivalent to the one obtained in dedicated hfMRI while the voxel size increases between two and 35 times. cMRI devices can perform imaging experiments that do not require ultra-high resolution (below 200 m in-plane), but their restricted accessibility strongly limits their use in rat. 4.3. Advantages and drawbacks of low field MRI in rat In this context, low field MRI devices have to be considered as an alternative to dedicated and clinical high field MRI for in vivo imaging in rat. Their principal and intrinsic limitation comes from the proportionality in MRI between the main magnetic field and the S/N ratio. Compared to higher magnetic field devices, the NEX must increase to compensate the lower intrinsic S/N ratio and TAT are then mandatory longer using a 0.1 T magnet. Considering that a similar acquisition protocol (sequence, coil, gradients. . .) is used at 1.5 and 0.1 T, the TAT should increase 152 times in order to compensate the difference in main magnetic field. The increase of the TAT (see Table 2) is indeed important with a median value of approximately 20, but is also strongly inferior to the theoretical expectation because of some technical adaptations observed and because we do not search to achieve the same S/N ratio. Designing adapted RF coils for each application (Table 1) allows to maximize the filling factor and to improve the S/N ratio, while the solenoidal geometry offers a proper RF pulse homogeneity [41]. As the console has only one receiver channel, we are limited to one coil operation and so cannot use modern acquisition sequences, which take advantage of phased-array coils. This is a way for possible improvements. In addition, 3D imaging sequences benefit from a better S/N ratio in each slice than 2D acquisitions with interslice gaps, at the expense of a longer TAT [4]. When considering the relative acquisition time for each slice at 0.1 T compared to higher magnetic field, we obtain a median ratio of 5 (1.5–19). In the case of isotropic 3D data, the reconstructed volume can be reoriented in any direction, thus limiting the use of scout sequences before the acquisition and reducing the total positioning time. Moreover, 3D acquisitions allow to measure cardiac or tumor volumes without requiring any geometric modeling of heart chambers or tumor. Thanks to these adaptations, low field 0.1 T MRI allows to obtain voxels with an in-plane size ranging from 375 m to 1 mm (Table 2). This almost twofold increase in in-plane size compared to cMRI comes from our choice of isotropic voxels or voxels with small ST in order to limit the partial volume effects. Consequently, the voxel volumes are almost equivalent (ranging from 0.053 to 1 mm3 ) to those obtained in 3D cMRI, in consideration of about 1 h TAT (Table 2). The second main limitation of a low field device is that it produces small chemical displacements, making any spectroscopic
E. Breton et al. / IRBM 29 (2008) 366–374
measurement impossible. However, low field MRI also causes less susceptibility artifacts in images than higher magnetic field MRI. The initial rationale for the use of a resistive 0.1 T low field magnet is its much lower cost in terms of initial expenditure, installation and operating costs than high magnetic field devices [15]. The limited magnet floor space requirement, weight and 5Gauss-line allow to settle the whole in a standard room without structural modifications. If necessary, a small Faraday cage can be built just surrounding the magnet to suppress outer noise. The electrical supply of the magnet and gradient coils needs a three-phase outlet and consumes 3.2 kW, while their cooling is ensured using a continuous water circuit (3 L min−1 ) but could also be done using a closed water circuit regulated in temperature (10 thousand euros). In addition, the system can be shut down during periods of inactivity. The purchase of a 0.1 T low field magnet similar to the one used in this study, its electrical supply and an MRI acquisition console costs approximately 200 thousand euros without VAT. Moreover, specific costs for in vivo small animal imaging have to be added to the MRI equipment. Gaseous anesthesia is considered as the most reliable anesthesia method in term of dosage and perturbation of cardiac and respiratory rhythms [45]. A gaseous anesthesia set costs approximately 15 thousand euros. An animal holder with a warming-up solution is mandatory to maintain homeostasis of the animal during the acquisitions and to standardize experiments. In addition, a small animal monitoring system, which is adapted in cardiac and respiratory frequencies, costs approximately 10 thousand euros. The global cost related to small animal in vivo imaging is of the order of magnitude of 40 thousand euros without VAT and is thus not immaterial compared to the 0.1 T MRI equipment. In comparison, the initial expenditure for a high magnetic field magnet represents approximately 1 million euros, excluding structural installation and operation costs. Low field 0.1 T MRI constitutes a solution to perform low cost MRI at the expense of long TAT, in the range of 1 to 2 h. A dedicated low field device also exhibits some practical advantages. In an open magnet, the animal remains accessible and visible all along the acquisition. The large air gap (18 cm in our 0.1 T device) also makes the animal handling and positioning easy. Since the 5-Gauss-line is located at 15 cm from the inner magnet, the metallic piece hazard is almost immaterial using a 0.1 T magnet. The devices needed for the MRI acquisition, gaseous anesthesia and the monitoring of physiological parameters can be placed in close proximity to the magnet (Fig. 1A). Consequently, a low field MRI device can be installed directly inside the animal facilities, including already existing ones without any structural modification. MRI follow up experiments can therefore be performed while the sanitary status of animals is kept all along the study. 5. Conclusion Considering the various in vivo rat-imaging experiments presented in this article, we obtained using a 0.1 T MRI, biological information comparable to those obtained in the literature using cMRI and hfMRI devices. Considering a 1 h acquisition time,
373
0.1 T MRI offers sufficient S/N to answer most of basic in vivo problematics, excluding spectroscopic studies. The low magnetic field allows to design practical small open magnets with large air gaps and significantly reduces metallic pieces hazard. Low field MRI offers a practical solution to perform routine imaging studies in rat models at very low initial and operating costs. Such low field devices can obviously not replace dedicated high magnetic field devices in studies requiring the highest resolution in vivo in rat. However, dedicated low field devices may reach a non-MRI-specialized scientific community interested in performing low cost MRI in order to answer daily biological imaging needs at the expense of long acquisition times. Acknowledgments E. Breton benefits from a grant cofinanced by General Electric Healthcare and the region of Alsace. Authors thank the Équipement vétérinaire Minerve Company for their technical support. Authors also wish to thank Pr. Izzie Namer for letting them use the glioma model and Dr. Julien Detour for pharmaceutical preparations. References [1] Koo V, Hamilton PW, Williamson K. Non-invasive imaging in small animal research. Cell Oncol 2006;28:127–39. [2] Brau ACS, Hedlung LW, Johnson GA. Cine magnetic resonance microscopy of the rat heart using cardiorespiratory-synchronous projection reconstruction. J Magn Reson Imaging 2004;20:31–8. [3] Brekke C, Lundervold A, Enger PO, Brekken C, Stalsett E, Pederson TB, et al. NG2 expression regulates vascular morphology and function in human brain tumours. Neuroimage 2006;29:965–76. [4] Faure P, Doan BT, Beloeil JC. In-vivo high resolution three-dimensional MRI studies of rat joints at 7 T. NMR Biomed 2003;16:484–93. [5] Fissoune R, Pellet N, Chaabane L, Contard F, Guerrier D, Briguet A. Evaluation of adipose tissue distribution in obese fa/fa Zucker rats by in vivo MR imaging: effects of peroxisome proliferator-activated receptor agonists. MAGMA 2004;17:229–35. [6] Lemaire L, Roullin VG, Franconi R, Vennier-Julienne MC, Menei P, Jallet P, et al. Therapeutic efficacy of 5-fluorouracil-loaded microspheres on rat glioma: a magnetic resonance imaging study. NMR Biomed 2001;14:360–6. [7] Maï W, Badea CT, Wheeler CT, Hedlung LW, Johnson GA. Effects of breathing and cardiac motion on spatial resolution in the microscopic imaging of rodents. Magn Reson Med 2005;53:858–65. [8] Blanchard J, Mathieu D, Patenaude Y, Fortin D. MR-pathological comparison in F98-Fischer glioma model using a human gantry. Can J Neurol Sci 2006;33:86–91. [9] Montet-Abou K, Daire JL, Hyacinthe JN, Nguyen D, Jorge-Costa M, Morel DR, et al. Optimization of cardiac cine in the rat on a clinical 1.5-T MR system. Magn Reson Mater Phys 2006;19:144–51. [10] Park MK, Lee HJ, Hong SH, Choi SS, Yoo YK, Lee K, et al. The increase in hepatic uncoupling by fenofibrate contributes to a decrease in adipose tissue in obese rats. J Korean Med Sci 2007;22:235–41. [11] Raila FA, Bowles AP, Perkins E, Terrel A. Sequential imaging and volumetric analysis of an intracerebral C6 glioma by means of a clinical MRI scanner. J Neurooncol 1999;43:11–7. [12] Thorsen F, Ersland L, Nordli H, Enger PO, Huszthy PC, Lundervold A, et al. Imaging of experimental rat gliomas using a clinical MR scanner. J Neurooncol 2003;63:225–31. [13] Wang D, Miller SC, Sima M, Parker D, Buswell H, Goodrich KC, et al. The athrotropism of macromolecules in adjuvant-induced arthritis rat model: a preliminary study. Pharm Res 2004;21:1741–9.
374
E. Breton et al. / IRBM 29 (2008) 366–374
[14] Yamamoto J, Hirano T, Li S, Koide M, Kohno E, Inenaga C, et al. Selective accumulation and strong photodynamic effects of a new photosensitizer, ATX-S10. Na(II), in experimental malignant glioma. Int J Oncol 2005;27:1207–13. [15] Hayashi N, Watanabe Y, Masumoto T, Mori H, Aoki S, Ohtomo K, et al. Utilization of low field MR scanners. Magn Reson Med Sci 2004;3:27–38. [16] Brurok H, Schojtt J, Berg K, Karlsson JO, Jynge P. Manganese and the heart: acute cardiodepression and myocardial accumulation of manganese. Acta Physiol Scand 1997;159:33–40. [17] Crossgrove J, Zheng W. Manganese toxicity upon overexposure. NMR Biomed 2004;17:544–53. [18] Natanzon A, Aletras AH, Hsu LY, Arai AE. Determining canine myocardial area at risk with manganese-enhanced MR imaging. Radiology 2005;236:859–66. [19] Wolf GL, Baum L. Cardiovascular toxicity and tissue proton T1 response to manganese injection in the dog and rabbit. Am J Roentgenol 1983;141:193–7. [20] Krombach GA, Saeed M, Higgins CB, Novikov V, Wendland MF. Contrastenhanced MR delineation of stunned myocardium with administration of MnCl2 in rats. Radiology 2004;230:183–90. [21] Hu TCC, Pautler RG, MacGowan GA, Koretsky AP. Manganese-enhanced MRI of mouse heart during changes in inotropy. Magn Reson Med 2001;46:884–90. [22] Arbogast-Ravier S, Xu F, Choquet P, Brunot B, Constantinesco A. Dedicated low field MRI: a promising low-cost technique. Med Biol Eng Comput 1995;33:735–9. [23] Haase A, Matthaei J, Hänicke W, Merboldt K-D. FLASH imaging. Rapid NMR imaging using low flip-angle pulses. J Magn Reson 1986;67:258–66. [24] Gyngell ML. The application of steady-state free precession in rapid 2DFT NMR imaging: FAST and CE-FAST sequences. Magn Reson Imaging 1988;6:415–9. [25] Goldbrunner RH, Bendszus M, Sasaki M, Kraemer T, Plate KH, Roosen K, et al. Vascular endothelial growth factor-driven glioma growth and vascularization in an orthotopic rat model monitored by magnetic resonance imaging. Neurosurgery 2000;47:921–30. [26] Young RJ, Knopp EA. Brain MRI: tumor evaluation. J Magn Reson Imaging 2006;24:709–24. [27] Goetz C, Breton E, Choquet P, Israel-Jost V, Constantinesco A. SPECT lowfield MRI system for small-animal imaging. J Nucl Med 2008;49:88–93. [28] Simon GH, von Vopelius-Feldt J, Wendland MF, Fu Y, Piontek G, Schlegel J, et al. MRI of arthritis: comparison of ultrasmall superparamagnetic iron oxyde vs. Gd-DTPA. J Magn Reson Imaging 2006;23:720–7. [29] Wang YXJ, Bradley DP, Kuribayashi H, Westwood FR. Some aspects of rat femorotibial joint microanatomy as demonstrated by high-resolution magnetic resonance imaging. Lab Anim 2006;40:288–95.
[30] Vallée JP, Ivancevic MK, Nguyen D, Morel DR, Jaconi M. Current status of small cardiac MRI in small animal. MAGMA 2004;17:149–56. [31] Iltis I, Kober F, Dalmasso C, Lan C, Cozzone PJ, Bernard M. In vivo assessment of myocardial blood flow in rat using magnetic resonance imaging: effect of anesthesia. J Magn Reson Imaging 2005;22:242–7. [32] Nahrendorf M, Wiesmann F, Hiller KH, Hu K, Waller C, Ruff J, et al. Serial cine-magnetic resonance imaging of left ventricular remodeling after myocardial infarction in rats. J Magn Reson Imaging 2001;14:547–55. [33] Stein AB, Tiwari S, Thomas P, Hunt G, Levent C, Stoddard MF, et al. Effects of anesthesia on echocardiographic assessment of left ventricular structure and function in rats. Basic Res Cardiol 2007;102:28–41. [34] Nahrendorf M, Hu K, Fraccarollo D, Hiller KH, Haase A, Bauer WR, et al. Time course of right ventricular remodeling in rats with experimental myocardial infarction. Am J Physiol Heart Circ Physiol 2003;284:241–8. [35] Thomas D, Ferrari VA, Janik M, Kim DH, Pickup S, Glickson JD, et al. Quantitative assessment of regional myocardial function in a rat model of myocardial infarction using tagged MRI. MAGMA 2004;17:179–87. [36] Seki Y, Naruse S, Seo Y, Kitagawa M, Ishiguro H, Wang Y, et al. Timecourse magnetic resonance imaging of rat pancreatic cyst after experimental pancreatitis. Magn Reson Imaging 2000;18:1003–10. [37] Tang H, Vasselli JR, Wu EX, Boozer CN, Gallagher D. High-resolution magnetic resonance imaging tracks changes in organ and tissue mass obese and aging rats. Am J Physiol Regul Integr Comp Physiol 2002;282:R890–9. [38] Fiel RJ, Alletto JJ, Severin CM, Nickerson PA, Acara MA, Pentney RJ. MR imaging of normal rat brain at 0.35 T and correlated histology. J Magn Reson Imaging 1991;1:651–6. [39] Herfkens RJ, Davis P, Crooks L, Kaufman L, Price D, Miller T. Nuclear magnetic resonance imaging of the abnormal live rat and correlations with tissue characteristics. Radiology 1981;141:211–8. [40] Herfkens RJ, Sievers R, Kaufman L, Sheldon PE, Ortendahl DA, Lipton MJ. Nuclear magnetic resonance imaging of the infarcted muscle: a rat model. Radiology 1983;147:761–4. [41] Ballon D, Graham MC, Miodownik S, Koutcher JA. Doubly tuned solenoidal resonators for small animal imaging and spectroscopy at 1.5 T. Magn Reson Imaging 1989;7:155–62. [42] Smith DA, Clarke LP, Fiedler JA, Murtagh FR, Bonaroti EA, Sengstock GJ, et al. Use of a clinical MR scanner for imaging the rat brain. Brain Res Bull 1993;31:115–20. [43] Rudin M. MR microscopy on rats in vivo at 4.7 T using surface coils. Magn Reson Med 1987;5:443–8. [44] Suddarth SA, Johnson GA. Three-dimensional MR microscopy with large arrays. Magn Reson Med 1991;18:132–41. [45] Janssen BJA, De Celle T, Debets JJM, Brouns AE, Callahan MF, Smith TL. Effects of anesthetics on systemic hemodynamics in mice. Am J Physiol Heart Circ Physiol 2004;287:1618–24.