Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Magn Reson Med. Author manuscript; available in PMC 2010 Aug 1.
Published in final edited form as:
PMCID: PMC2742771

On the Dual Contrast Enhancement Mechanism in Frequency Selective Inversion Recovery Magnetic Resonance Angiography (IRON-MRA)


The susceptibility of blood changes after administration of a paramagnetic contrast agent which shortens T1. Concomitantly the resonance frequency of the blood vessels shifts in a geometry-dependent way. This frequency change may be exploited for incremental contrast generation by applying a frequency selective saturation pre-pulse prior to the imaging sequence. The dual origin of vascular enhancement depending firstly on off-resonance and secondly on T1 lowering was investigated in vitro together with the geometry dependence of the signal at 3T. First results obtained in an in vivo rabbit model are presented.

Keywords: Magnetic Resonance Angiography, Off-resonance imaging, Blood-pool contrast agent


In a recent paper by Edelman et al., a novel method for Magnetic Resonance Angiography (MRA) was presented based on the off-resonance properties of blood after the administration of gadolinium [1]. Because of the contrast agent, a local magnetic susceptibility change is induced in the blood. This results in an off-resonance signal behavior: the Larmor frequency shifts, resulting in changes of the sensitivity to frequency selective radiofrequency pulses, which is exploited to generate the angiographic contrast.

Furthermore, there is the well-known contrast enhanced T1-weighted MRA, typically with bolus injection of gadolinium-based compounds. This is a clinically widespread and accepted method to obtain exquisite images of the vasculature [2]. With short echo-times, large flip-angles and short TR, the background is well suppressed and the images are relatively insensitive to flow-artifacts. However, because the contrast agent quickly extravasates into the extracellular space, scan time is limited to the order of one minute. Even with advanced k-space trajectories such as randomized centric [3], this is a limiting factor for high-resolution imaging with large volumetric coverage. For this purpose, blood-pool agents are attractive: the contrast media resides in the blood during the whole imaging session, keeping the contrast at maximum, while concomitant background enhancement can largely be avoided [4].

Superparamagnetic iron-oxide nanoparticles have great potential to be utilized for MRA combining the aforementioned methodologies: On the one hand there is a very strong off-resonance effect at equilibrium concentrations in the blood-pool. On the other hand, these nanoparticles act as blood-pool agents, thereby enabling imaging with a longer scan time. Finally, the R1-relaxivity is very high, resulting in a significant T1-shortening of the blood-pool at equilibrium.

Exploiting the above features, a sequence was designed for high spatial resolution imaging of vessels with a blood-pool agent in the equilibrium phase. The method suppresses the on-resonant background signal with a frequency selective saturation pre-pulse. Off-resonant spins in the blood pool will retain high signal since they do not undergo saturation. Blood that is on-resonance due to unfavorable vessel geometry will also appear bright due to rapid T1 recovery. For in vivo imaging, additional fat suppression was added. Below we describe the general design of the sequence and the resultant signal enhancement properties of the vascular signal in vitro. An in vivo application is also presented.


Frequency shift of blood after contrast

After injection of a paramagnetic contrast agent, the susceptibility of the blood-pool changes. With the approximation to an infinitely long cylindrical geometry in the external magnetic field, there will be a change of the magnetization inside the vessels. This will induce frequency shift of the resonance frequency of the blood depending on the contrast agent concentration [5]. This frequency shift Δν is also dependent on the orientation of the vessel to the external main magnetic field, as


Here θ is the angle between the main magnetic field and the axis of the cylinder, and Δν0 is proportional to the change in susceptibility induced by the contrast agent (see e.g. [5] for a full explanation). The frequency shift is maximized if the vessel is aligned in parallel with the main magnetic field and becomes negative if perpendicular. The effect is nulled when θ=cos-1(1/√3) ≈ 55°.


The schematic of the radio frequency (RF) part of the proposed frequency selective inversion recovery magnetic resonance angiography (IRON-MRA) pulse sequence is shown in Fig 1. The first component is a spectrally selective saturation pulse centered on the water frequency. By choosing its duration, the bandwidth can be adjusted. Optionally, this pulse is followed by a frequency selective fat suppression pulse. These frequency selective pre-pulses are then followed by a crusher gradient. After this magnetization preparation, a segmented RF-spoiled gradient echo pulse train is performed for image data collection (‘Imaging’ in Fig 1). In this example, imaging starts with (six) dummy RF excitations, and ramped RF excitations are used for imaging.

Fig 1
Schematic view of one segmented k-space gradient echo shot of the sequence: first the frequency selective on-resonance saturation pulse (3-50 ms), (optionally) followed by a frequency selective fat saturation pulse (10 ms). This is followed by a crusher ...

By using a frequency selective narrow band saturation pre-pulse, schematically shown in Fig 2, the blood that is off-resonance will not be saturated, and will be enhanced relative to the on-resonant background that appears signal suppressed (i.e. the IRON method [6]). While this describes part of the proposed angiographic method, in practice the T2* broadened spectral peak of the blood-pool and the attenuation band of the pre-saturation pulse overlap. This causes a duality in the origin of the signal from the blood-pool. Off-resonant spins in the blood pool will have high signal since they do not experience the saturation pulse. The on-resonant frequency components in the blood-pool will exhibit high signal despite the saturation pulse because of rapid T1-recovery of their magnetization. Disregarding potential inflow-effects, the relative contribution from the blood-pool (from components in- and outside of the pre-pulse suppression bandwidth) is thus determined by the contrast agent concentration (i.e. Δν0, T1 and T2*) and the vessel geometry (i.e. θ) as well as the spectral width (BWIRON) of the saturation pre-pulse.

Fig 2
Schematic effect of the IRON preparation pulse on the longitudinal magnetization in the frequency domain (as derived from the numerical RF pulse shape). The absolute magnitude is proportional to the recovered magnetization after the previous pulse train. ...

The dual origin of the signal from the blood-pool in IRON-MRA (i.e. off-resonance and T1) is investigated, quantified, and discussed below (see Fig 3).

Fig 3
Calculated signal intensities for simulations with several combinations of T1 and T2* as a function of the imaging delay during which T1-relaxation occurs. Off-resonant tissue would not be affected by the saturation pulse, which is modeled by setting ...

Signal from blood outside the spectral band of the pre-pulse

Broadband RF excitation pulses of the segmented k-space gradient echo imaging sequence will affect all spins present in the imaging volume. The spins outside the frequency band of the IRON pre-pulse (centered around ωIRON with bandwidth BWIRON) will not be saturated before imaging and the imaging characteristics of a regular segmented k-space gradient echo imaging sequence can be expected for those frequency components in the blood-pool. Apart from the T1 relaxation in-between shots, the magnetization experiences both excitation and T1-relaxation during the acquisition train. This will influence the measured signal strengths of the different k-space profiles. With a low-high or centric k-space profile order, the available magnetization and therefore the resulting signal intensity of the center profiles will be higher than that of the more peripheral k-space profiles where magnetization will be depleted by the preceding RF excitation pulses. The magnitude of this effect depends on T1 and the applied flip-angles. An incremental flip angle up to the nominal angle in the imaging part of the sequence will (partly) correct for this signal strength modulation and minimize the resulting artifacts such as image blurring and ghosting.

Signal from blood inside the spectral band of the pre-pulse

The magnetization of on-resonant structures (i.e. within the frequency band of the saturation pulse) experiences the saturation pre-pulse prior to imaging. During the time period from the center of the on-resonant suppression pulse to the start of the RF excitation pulses used for imaging, T1-relaxation occurs. This period will be denoted ‘imaging delay’ (see Fig 1). If fat saturation is applied, the imaging delay is further increased by about 10 ms at 3T. During the dummy RF excitations and those used for imaging there is concomitant T1-recovery and saturation.

This sequence of events leads to a contrast formation that depends on T1. For long T1's of regular tissue, the signal recovery during these recovery times will be small, resulting in signal suppression in the image. However, post contrast, the blood will have a T1 that is close to the duration of the imaging pulse train length. Apart from signal suppression, this will result in a regrowth of the magnetization during the pulse train and thus to an increase in signal as a function of the acquired k-space profiles. With a low-high profile order in k-space, this leads to a modulation of the signal intensity with amplification of the signal of the higher k-space profiles similar to a high-pass filter [7].

Composite signal from blood

The signal from the blood-pool can be calculated by integrating the Bloch equations [5], taking into account the relative contributions of both on-resonant and off-resonant components as described above. In contrast to T1, the relative contribution of the off-resonant component depends on the orientation of the vascular structure with respect to the main magnetic field (see Eq. 1). Thus, in case of unfavorable orientations of the vessel the on-resonant T1 saturation recovery contrast may act as a backup enhancement.

The signal fraction β from outside the saturation band (with bandwidth BWIRON) of the total signal of the blood-pool, can be investigated by changing the spectral width of the suppression pulse. With broadband suppression pulses, all of the signal will experience saturation recovery (β=0), while smaller widths will add off resonant signal components, provided the frequency shift induced by the contrast agent is large enough (0< β<1). A practical issue is that the imaging delay in Fig 1 depends on the duration of the pre-pulse and thus, on the bandwidth BWIRON. To study this, a series of images with gradually increasing BWIRON can be acquired, yielding image intensities Snarrow,i with i running over a range of different bandwidths BWIRON. For comparison a baseline scan can be acquired in which a short broadband saturation pulse (BWIRON larger than largest expected off-resonance signal) is used. By increasing the time delay between this short pulse and the imaging part of the sequence, the period during which T1 recovery occurs can be matched to that of a longer pulse with a narrower bandwidth BWIRON (see Fig 1). The signal intensity in the images obtained with the short suppression pulse is denoted Sbroad,i. From the ratio of the signals of the matching pairs Snarrow,i/Sbroad,i and with a reasonable T1 estimate, the off resonance fraction βi at a certain spectral width can be calculated as


where ξ01 is the ratio of relative signal strengths from outside (ξO) and inside (ξI) the suppression band as obtained from the integration of the Bloch equations for the T1 at hand.


Numerical simulations

The contrast between the blood-pool and the background tissue has a dual origin as described above: One part is attributable to off-resonance and the other to T1. To examine these relative contributions, a numerical integration of the Bloch equations was performed using matrix calculations [5,11]. RF pulses were considered instantaneous, read-out direction effects were ignored and crusher gradients were considered perfect. The simulations included dummy excitations and flip-angle sweep and the pulse train was repeated to obtain pseudo-steady state conditions. The magnetization started in its equilibrium along the z-axis with a value of 100%. The simulations were performed using the scan parameters of the in vivo acquisition described below: TR/TE/α=3.5ms/1.4ms/15°, 19 k-space profiles per imaging pulse train with 6 preceding dummy excitations. The simulations were performed with and without the influence of the pre-saturation pulse to show the effect of off-resonance. The case without saturation (i.e. off resonance) was implemented by setting the flip-angle of the pre-pulse to zero.

The computed relative transverse magnetization was used to weight the profiles in k-space of a digital phantom. The temporal order of the calculated k-space profiles was matched to that of the scanner. The digital phantom consisted of an oval homogeneous region with pre-defined T1 and T2 on a blank (no signal) background. The numerical simulations were implemented in C++ on the ImageXplorer software platform (Image Sciences Institute, Utrecht, The Netherlands). After the Fourier transform, an ROI was placed centrally inside the area of the digital phantom to determine the resulting relative signal strength ξ. The signal ξ was calculated for T1 and T2* combinations of blood with MION-47 contrast (T1=150 ms, T2*=5 ms), for fat (T1=290 ms, T2*=40 ms) and for soft tissue as a representatives of the background in vivo (T1=1100 ms, T2*=25 ms) [11,12]. The numerical simulations were also performed for conventional first pass contrast enhanced MRA parameters (T1=10 ms, T2*=5 ms). While the value of ξ reflects the influence of T1 recovery during the imaging delay, ghosting and edge effects in the simulated images represent T1 effects during an imaging pulse train. These effects were not quantified.

Phantom studies

The imaging agent used in the experiments was MION-47 (200 mM), a monocrystalline iron oxide nanoparticle [8]. The intravascular half-time of MION-47 is over 10 hours in rodents [9,10].

To investigate the dependency of the signal on the angle of the object with respect to the main magnetic field (see Eq. 1), a syringe filled with 6 ml of rabbit blood and 0.05ml MION-47 with a length to diameter ratio of 3.5 was suspended in gelatin (Jell-O, Kraft Foods, Northfield, IL). This was placed on a revolving platform with an angular scale and put inside the 8-channel headcoil of the 3T scanner (Achieva, Philips Healthcare, Best, the Netherlands) at room temperature. The syringe was imaged in the coronal-plane and the phase of the signal was measured on 3D T1-weighted segmented k-space gradient echo images. These measurements were performed for a syringe orientation in parallel with the main magnetic field and at 18 angular increments of 10° about the anterior-posterior axis. Imaging parameters were TR/TE/α=3.7ms/1.9ms/15°, voxel size 0.6×0.6×2mm, matrix 256×256×10, full echo, 19 k-space profiles per imaging pulse train, BWIRON=340 Hz. For quantification an ROI was positioned in the center of the syringe. The measured phase Δφ in the ROI as a function of the rotation angle was converted to a frequency shift via Δν= Δφ/2πTE after phase unwrapping. This frequency shift and the standard deviation in the ROI were plotted for each angular position of the syringe and subsequently fitted to Eq 1 (with an offset in the starting angle to the main magnetic field and an absolute phase offset as additional degrees of freedom).

In a second phantom experiment, the center-frequency ωIRON of the saturation band was varied from - 500 to 500 Hz in increments of 50 Hz (using BWIRON =102 Hz). Imaging parameters were used as described above. Imaging as a function of ωIRON was repeated for several orientations of the syringe relative to the main magnetic field. The signal magnitude was measured in a centrally positioned ROI for each frequency increment. The center frequency of the suppression was estimated by fitting a Gaussian on a constant background.

In vivo experiment

The study was approved by the Johns Hopkins Institutional Animal Care and Use Committee. Imaging was performed in a New Zealand White rabbit (3 kg). After sedation with acepromazine (1mg/kg i.m.) and ketamine (40mg/kg i.m) free breathing anesthesia was maintained with intravenous thiopental. MION-47 nanoparticles (see Phantom studies) were given as a slow bolus infusion of 80 μM Fe/kg. Imaging was performed within 2 hours of injection of the contrast agent.

The influence of the center-frequency ωIRON in vivo was investigated analogously to the phantom experiment with variation of the saturation band from -500 to 500 Hz in increments of 50 Hz (using BWIRON =102 Hz). Implemented as a dynamic scan, the segmented k-space gradient echo imaging parameters were: TR/TE/α=3.5ms/1.4ms/15°, 19 k-space profiles per k-space segment, fractional echo, 512×245×5 acquisition matrix, reconstructed to 512×512×10 with a FOV of 140×140×10 mm3, 4 signal averages, 100° RF excitation angle of the saturation pulse at 0 Hz offset using a 4 -channel carotid surface coil (Pathway MRI, Seattle, WA, USA) for signal reception.

The ratio β between the signal attributable to T1 recovery (within the saturation bandwidth) and that originating from off resonance (outside of the saturation bandwidth) was quantified in vivo by measuring the signal in two images. The first set of images was acquired with a long pre-pulse (BWIRON ranging from 102 to 1020 Hz) and the second set with the short pre-pulse (BWIRON=1700 Hz). To allow for equal T1 recovery, an additional time delay was added following the short pre-pulse, such that both images have the same imaging delay (see Fig 1). Equation 2 was used to estimate β for these in vivo images. The imaging parameters were as above, except 512×206×8 acquisition matrix, reconstructed to 512×512×15, 12 signal averages with a FOV of 140×140×15 mm3.

The T1 recovery in-between RF excitations used for imaging will lead to relative signal changes during the acquisition of a k-space segment and therefore to ghosting and changes in edge definition. These effects were qualitatively investigated by swapping fold-over and readout direction.


Numerical simulations

The relative weight of the transverse magnetization of the different phase encoding profiles was calculated using the numerical simulations described above. In Fig 3 the resultant changes in the image intensities in the ROI of the digital phantom are shown. Scenarios with and without pre-saturation depict the effect of off-resonance. After saturation, the different combinations of T1 and T2* show the dependence of the signal on T1 relaxation during the time delay between pre-pulse and imaging. The influence of the saturation pre-pulse on the magnetization is largest for longer T1 values, and almost disappears for T1=10ms, i.e. when T1 is shorter than the imaging delay.

The influence of T1 recovery during the imaging pulse train on the relative strengths of the k-space profiles is shown in Fig 4. Here, the increase of magnetization for the more peripheral profiles in the case of blood with MION-47 is shown. The resulting effect on edge-enhancement was investigated in the in-vivo experiment.

Fig 4
Relative strength of the transverse magnetization (‘signal’) as a function of time. Simulation parameters are TR/TE/α=3.5ms/1.4ms/15°, TFE factor 19, T1=120 ms, T2*=8 ms, 6 dummy profiles (not plotted), imaging delay 22 ...

Phantom studies

In Fig 5 the frequency shift of the blood in the syringe is plotted as a function of its angle with respect to the main magnetic field. The fit of Eq.1 to the data yields a maximum frequency shift Δν0 of 219 Hz (with an offset angle of 5°, see fitted curve in Fig 5). The signal intensity in the syringe as a function of the rotation angle is plotted in Fig 6 when applying the IRON MRA sequence with BWIRON=340 Hz. The intensity in the background remains constant at 15 (not shown).

Fig 5
Frequency shift of the blood as function of the rotation angle of the phantom: (○) measured values and (solid line) a fit of Eq. 1 with a fitted offset angle of 5 degrees for the syringe with respect to the main magnetic field (at 0 degrees on ...
Fig 6
Signal intensity of the blood in the syringe as function of the rotation angle of the phantom (B0 in vertical direction in the images) using a BWIRON of 340 Hz: (○) average values in ROI (box) with standard deviations. At five positions images ...

In Fig 7 the center-frequency ωIRON of the pre-pulse was gradually changed to show the spectrally selective behavior of the saturation pulse with BWIRON=102 Hz. From Eq. 1, the maximum positive frequency shift is expected at 0° (syringe parallel to main magnetic field), no shift at 55° and a (half as large) negative shift at 90°. The locations of the signal minima in Fig 7 are at 121 Hz, 0 Hz and -59 Hz respectively. According to Eq. 1 this corresponds to a Δν0 of 180 Hz. For reference, a background peak from gelatin is shown with a center frequency at 4 Hz. Note that this was measured in the center of the phantom and that this is shim dependent. Frequency offsets of up to 50 Hz can easily occur at the periphery of the phantom. As expected, the background signal does not change as a function of the phantom orientation (data not shown).

Fig 7
Average signal in an ROI with the phantom as a function of the shift of the saturation pre-pulse (BWiron 102 Hz) relative to the water frequency. Three orientations are shown: (Δ) oriented parallel to the main magnetic field (○) at 55 ...

In vivo experiment

The in vivo signal changes as a function of the frequency ωIRON is shown in Fig 8 for the blood in the aorta (parallel to the main magnetic field), the renal artery (where perpendicular to the field), the background fat and the background muscle. The minimum of the signal intensity depends both on the tissue and on the orientation relative to the main magnetic field.

Fig 8
Influence of the IRON pulse on the signal from different tissues. As a function of the off-set frequency ωIRON, the signal is shown from the aorta (●), the renal artery (○), the perirenal fat (□) and background tissue in ...

The change of the signal after narrow vs. wide band saturation as a function of the influence of the imaging delay is illustrated in Fig 9. Shown are MIPs with a window level covering the whole pixel value range. The top row shows the IRON-MRA for different BWIRON. The bottom row shows the use of a wide band suppression pulse, with an imaging delay for saturation recovery equal to that in the top row. Quantitative values measured in the aorta and in the psoas muscle of this rabbit are displayed in Fig 10a. At very small imaging delays and large bandwidths, the curves converge, as the bandwidth of the pre-pulse is larger than the MION-induced frequency shift of the blood-pool. For smaller bandwidths, the signal components from spins that are resonating beyond the cutoff frequency of the suppression pre-pulse are increasingly contributing to the overall signal. For very long pulses and thus narrow bandwidths, the difference between the two curves in Fig 10a is constant as almost the entire blood-pool is unaffected by the pre-pulse. By using the simulated signal strengths shown in Fig 4, the fraction β of the signal from outside the saturation band is estimated, see Fig 10b. For small bandwidths of the pre-pulse, values in the aorta are close to ∼1 (90% off-resonant spins), while in muscle only a modest fraction (10%) is off-resonant. In contrast, for shorter pre-pulses with larger bandwidths, all spins are resonating within the suppression bandwidth (β ≈ 0).

Fig 9
In vivo MIP images of the abdominal aorta and caval vein with different bandwidths of the saturation pre-pulse (a) 1020 Hz (b) 510 Hz (c) 340 Hz (d) 255 Hz (e) 170 Hz (f) 128 Hz (g) 102 Hz showing both T1 and off resonance signal. Images with a high bandwidth ...
Fig 10
(a) In vivo signal intensity in the abdominal aorta (circles) and in the psoas muscle (triangles) after administration of MION-47 (from source images of Fig 9). In the scans the imaging delay was changed, (solid markers) by using a longer pre-pulse (bandwidth ...

The numerical simulations predicted a T1-dependent change in signal of consecutively acquired profiles in k-space. In the image, these signal variations lead to a modulation of the signal strength in the fold-over direction, as shown in Fig 4. Consistent with these theoretical findings, the resultant edge enhancement effect can be observed in the in vivo images in Fig 11 (arrows). By changing the fold-over direction, this edge-enhancement changes its orientation accordingly. The effect is best visible in the region of the infra-renal abdominal aorta in Fig 11. The edge enhancement appears more pronounced with broadband saturation as compared to saturation with a narrower bandwidth (Fig 11c and d). Note that in these images additional fat suppression was applied and the window-level was adjusted to increase conspicuity of the effect.

Fig 11
In vivo example of MRA of the abdominal aorta windowed to exaggerate the desired effect: (a) Foldover direction right-left and (b) feet-head with broadband suppression(BW 1700Hz) (c) Foldover right-left and (d) feet-head with narrow bandwidth suppression ...


In this paper it is shown that IRON-MRA supports positive contrast generation between the blood-pool and the background after injection of MION-47 superparamagnetic nanoparticles. The performance of this method was investigated in silico, in vitro and in vivo. Positive contrast is obtained in part by the shifted frequency of the blood-pool post contrast if a spectrally selective saturation pulse precedes the imaging part of the sequence. The off-resonant blood-pool magnetization outside the saturation band is not affected by this pre-pulse, while background tissue that remains on-resonant, appears signal-attentuated in the images. The second component of this positive contrast is created by the T1-shortening after MION-47 administration.

With a T1 value that is typical for iron-oxide blood pool agents in the steady state [13], IRON-MRA provides a robust technique, and the relatively low T1 supports vessel enhancement even at a 55° angle where the off-resonance effect only minimally contributes to the contrast generation. In Fig 3, the in vivo findings using the IRON technique are corroborated by the results from the numerical simulations. The contrast in IRON-MRA can be deduced from the figure by examining the curves for T1=150 ms and T2*=5 ms, which is typical for the MION enhanced blood-pool. For a certain bandwidth, the difference in the relative signal intensity with (open circles) versus the intensities without experiencing the saturation (closed circles) signifies the off-resonant enhancement of the blood-pool in IRON. The difference between the closed circles (the off-resonant blood-pool) and the open markers in the other curves (i.e. on-resonant background tissue) explains the total contrast with IRON, being partly off-resonance and partly T1 mediated. Note that the calculated signal intensities are always larger than zero even for zero imaging delay due to T1 recovery during the dummy RF excitations.

When T1 values are extremely low such as during first pass imaging, frequency selective saturation recovery and non-selective saturation recovery may lead to very similar results in the blood-pool. Straightforward T1-weighted imaging with larger flip angles may then be considered.

Recently, an alternative off-resonance contrast angiography method was published by Edelman et al. [1]. In their implementation, a frequency selective excitation was used in conjunction with a frequency selective saturation pulse as described here. With this implementation the positive side of the frequency spectrum is exclusively visualized. This means, that the negative frequency components do not contribute to the signal and the directional dependency of the technique is therefore amplified. The added value of this paper relative to [1] may also include the explicit inclusion of the T1-effect for maximizing angiographic contrast generation in clinical applications, which was minimized in [1] to single out the off resonance effect.

The experiments in this paper employed an iron oxide based contrast agent, which induces a large frequency shift at low concentrations (here Δν0 ≈180 Hz for 1.7 mM MION). With gadolinium compounds, a higher concentration would be needed to achieve the same frequency shift, proportional to the main field strength and the susceptibility Δχ (see also [1]). For example a Δν0 ≈ 180 Hz would require a concentration of around 13 mM of gadolinium [14]. Therefore, and consistent with the findings in [1], IRON-MRA in conjunction with gadolinium would only be possible during the first pass of the contrast agent [1]. However, during the first pass, T1 with gadolinium would be very short and alternative contrast enhancement mechanisms may be preferred. Additionally, first pass imaging may circumvent venous over projection, which is a potential drawback of the steady state acquisition of IRON-MRA.

A major limitation of off-resonance angiography is that the frequency shift depends on the geometry of the vessel. The vessel has to resemble a cylinder that is preferably aligned in parallel with the main magnetic field in order to maximize contrast generation induced by the frequency shift. For practical purposes, a cylindrical segment that is at least four times longer than its diameter suffices [15]. While this is a weak constraint as demonstrated by the utility of the same effect in cerebral perfusion arterial input function determination [16], the orientation to the main magnetic field remains important. While the directional dependency of the contrast enhancement mechanism would direct the application toward imaging of vessels in parallel with or perpendicular to the main magnetic field, we did not observe an unequivocal signal attenuation in the in vivo images. Significant line broadening due to T2* shortening, thereby shifting parts of the blood signal away from resonance, may partly explain this (see the slightly wider attenuation peak of blood versus fat or muscle in Fig 8). Other in vivo effects such as in-flow, B1 inhomogeneity and partial volume effects, may also play a role. Still, an estimated signal loss of 20-30% can be expected at certain angles based on the phantom and in vivo experiments (see Figs 7 and and8).8). Furthermore, a limited shimming may cause inhomogeneous background suppression in vivo. These issues will have to be addressed with more in-depth in vivo experiments, see e.g. [17].

In the phantom experiment that was conducted to demonstrate the angular dependency of the frequency shift, a slightly higher concentration of MION was used than in vivo (1.7 mM vs. 1 mM). This leads to accelerated T1 recovery but simultaneously also to a larger frequency shift. In vivo, and at the lower dose, the frequency shift is expected to be about 50% of that from the phantom studies. Note that in Fig 9 a significant in vivo off-resonance effect for BWIRON up to 300 Hz can be appreciated.

The accuracy of the in vivo calculation of the relative contribution of the signal from inside and outside the suppression band depends on the knowledge of the T1-value. However, this is generally unknown in practice. For example, the values of the aorta in Fig 10b are calculated using an estimated T1 value of 150 ms. If the T1 changes by 20%, the fraction β will change by 30%, and is therefore rather sensitive to T1 variations. Therefore, this ratio should be interpreted as indicative of the order of magnitude rather than as an exact number.

In the two in vitro experiments, there was a discrepancy between the frequency shift as measured from the phase shift (Δν0 =219 Hz) as compared to the location of the peaks of maximum saturation (Δν0 =180 Hz). Apart from statistical differences (e.g. the peak location fit was reproducible within ±15 Hz), this difference may be explained by the convolution of the resonance frequency spectrum with the saturation profile and the subsequent fitting to a Gaussian model. The choice of the ROI also has influence, as the syringe is only a limited approximation of an infinite cylinder. Since the maximum frequency shift in parallel to the main magnetic field is only 26 Hz different between these experiments, this was not further explored.

The numerical simulations of the Bloch equations have some limitations. First, the T1 of the blood with MION is estimated and so are the T2* values. These are, to our knowledge, not currently available for 3T in the literature. However, with the short echo-times used (a few milliseconds), the inaccuracy in T2* is not likely to affect the results of the numerical simulations. Only for very short values as found during first pass imaging, there may be some influence, which will, however, not change the global findings in a major way. Second, in the numerical simulations, only the global signal intensity changes were taken into account. Neither distortion due to off-resonance signal nor the related chemical shift in the readout direction have been addressed. The chemical shift of the blood-pool with MION is expected to be rather small relative to the acquisition bandwidth typically used for MRA.


The vascular signal formation of a new MRA method was investigated and is based in part on the susceptibility induced spectral shift of the blood-pool after the administration of an iron oxide nanoparticle. The relative contribution of this off-resonance effect in relation to T1-enhancement has been characterized both in vitro and in vivo. The method is particularly well suited for use in a steady-state acquisition, and supports enhanced contrast between the blood-pool and background tissue. The technique is orientation dependent, but the dual contrast enhancement mechanism (off-resonance and T1) may alleviate this effect to a large degree.


This work is supported by the NIH grant RO1 HL084186 and in part by a grant from the Donald W. Reynolds Foundation. Dr. Stuber is compensated as a consultant by Philips Medical Systems NL, the manufacturer of equipment described in this presentation. The terms of this arrangement have been approved by the Johns Hopkins University in accordance with its conflict of interest policies.


List of symbols

°degrees (superscript small circle)
αGreek small alpha
open square
solid triangle
open triangle
solid circle
open circle
solid square
μGreek small mu
ΔGreek capital Delta
ξGreek small xi
θGreek small theta
βGreek small beta
of the order of (centered tilde)
ωGreek small omega


1. Edelman RR, Storey P, Dunkle E, Li W, Carillo A, VU A, Carrol TJ. Gadolinium-enhanced off-resonance contrast angiography. Magn Reson Med. 2007;57:475–84. [PubMed]
2. Prince MR, Meaney JF. Expanding role of MR angiography in clinical practice. Eur Radiol. 2006;16 2:B3–8. [PubMed]
3. Fain SB, Riederer SJ, Bernstein MA, Huston J., 3rd Theoretical limits of spatial resolution in elliptical-centric contrast-enhanced 3D-MRA. Magn Reson Med. 1999;42:1106–16. [PubMed]
4. Tombach B, Reimer P, Bremer C, Allkemper T, Engelhardt M, Mahler M, Ebert W, Heindel W. First-pass and equilibrium-MRA of the aortoiliac region with a superparamagnetic iron oxide blood pool MR contrast agent (SH U 555 C): results of a human pilot study. NMR Biomed. 2004;17:500–6. [PubMed]
5. Haacke EM, Brown RW, Venkatesan R, Thompson MR. Magnetic Resonance Imaging: Physical Principles and Sequence Design. Wiley-Liss; 1999.
6. Stuber M, Gilson WD, Schar M, Kedziorek DA, Hofmann LV, Shah S, Vonken E, Bulte JW, Kraitchman DL. Positive contrast visualization or iron oxide-labeled stem cells using inversion recovery with on-resonant water suppression. Magn Reson Med. 2007;58:1072–7. [PubMed]
7. Epstein FH, Mugler JP, 3rd, Brookeman JR. Spoiling of transverse magnetization in gradient-echo (GRE) imaging during the approach to steady state. Magn Reson Med. 1996;35:237–45. [PubMed]
8. Shen T, Weissleder R, Papisov M, Bogdanov A, Jr, Brady TJ. Monocrystalline iron oxide nanocompounds (MION): physicochemical properties. Magn Reson Med. 1993;29:599–604. [PubMed]
9. Wunderbaldinger P, Josephson L, Weissleder R. Tat peptide directs enhanced clearance and hepatic permeability of magnetic nanoparticles. Bioconjug Chem. 2002;13:264–8. [PubMed]
10. McLachlan SJ, Morris MR, Lucas MA, Fisco RA, Eakins MN, Fowler DR, Scheetz RB, Olukotun AY. Phase I Clinical Evaluation of a New Iron Oxide MR Contrast Agent. J Magn Reson Imaging. 1994;4:301–7. [PubMed]
11. Bernstein MA, King KF, Zhou XJ. Handbook of MRI Pulse Sequences: A Guide for Scientists, Engineers, Radiologists, Technologists. Academic Press; 2004.
12. Stanisz GJ, Odrobina EE, Pun J, Escaravage M, Graham SJ, Bronskill MJ, Henkelman RM. T1, T2 relaxation and magnetization transfer in tissue at 3T. Magn Reson Med. 2005;54:507–12. [PubMed]
13. Hartmann M, Wiethoff AJ, Hentrich HR, Rohrer M. Initial imaging recommendations for Vasovist angiography. Eur Radiol. 2006;16 2:B15–23. [PubMed]
14. Akbudak E, Norberg RE, Conturo TE. Contrast-agent phase effects: an experimental system for analysis of susceptibility, concentration, and bolus input function kinetics. Magn Reson Med. 1997;38:990–1002. [PubMed]
15. van Osch MJ, Vonken EJ, Viergever MA, van der Grond J, Bakker CJ. Measuring the arterial input function with gradient echo sequences. Magn Reson Med. 2003;49:1067–76. [PubMed]
16. Conturo TE, Akbudak E, Kotys MS, Chen ML, Chun SJ, Hsu RM, Sweeney CC, Markham J. Arterial input functions for dynamic susceptibility contrast MRI: requirements and signal options. J Magn Reson Imaging. 2005;22:697–703. [PubMed]
17. Korosoglou G, Shah S, Vonken EJ, Gilson WD, Schär M, Tang L, Kraitchman DL, Boston RC, Sosnovik D, Weiss RG, Weissleder R, Stuber M. Off-Resonance Angiography: A New Method to Visualize Vessels. Phantom and Rabbit Studies. Radiology. 2008;249:501–9. [PMC free article] [PubMed]
PubReader format: click here to try


Save items

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • Compound
    PubChem chemical compound records that cite the current articles. These references are taken from those provided on submitted PubChem chemical substance records. Multiple substance records may contribute to the PubChem compound record.
  • MedGen
    Related information in MedGen
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem chemical substance records that cite the current articles. These references are taken from those provided on submitted PubChem chemical substance records.
  • Taxonomy
    Taxonomy records associated with the current articles through taxonomic information on related molecular database records (Nucleotide, Protein, Gene, SNP, Structure).
  • Taxonomy Tree
    Taxonomy Tree

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...