**3.2. Advances in specialized hardware for high field animal MRI**

#### *3.2.1. High field magnets*

While gradient and RF coil technology spans numerous decades of history (as early as Paul Lauterbur's ground-breaking inception of zeugmatography [Lauterbur 1973]), the stringent technical requirements imposed by the mouse physiology for cardiac imaging have led to optimization of prior technology or to recent introductions of new technologies, as reviewed and discussed briefly in this section. Interestingly, large-bore high-field systems emerged during the past decade at field strengths spanning 4.7, 7.1, 9.4, and 11.7T maintaining, however, the same design technology as originally developed for human superconducting systems in the late 1980's and 1990's.


**Table 1.** Summary of various cardiovascular and biochemical physiological indices of anesthetized mice using Isoflurane from bench studies. Stable mouse responses were obtained under both bench and scanner experimental conditions for periods spanning 1-1.5 hrs post-induction. [This table originally appeared in Constantinides C et al., Effects of Isoflurane Anesthesia on the Cardiovascular Function of the C57BL/6 Mouse, published in ILAR e-Journal 52(2), available online at *http://delsold.nas.edu/ilar\_n/ilarjournal/52\_2/PDFs/v5202e-Constantinides.pdf*].

#### *3.2.2. Gradient coil technology*

356 Practical Applications in Biomedical Engineering

other critical physiological parameters (including, but not limited to HR, coefficient of variation [CV], mean arterial pressure [MAP], blood pH, glucose, and insulin) maintaining

homeostasis throughout the study as shown in Table 1 [Constantinides\_ILAR 2011].

**Figure 6.** Physiological monitoring of mice during MRI using a dedicated recording system and the Labview software; (b) specially designed mouse positioning cradle, and (c) RF coil-mouse cradle assembly for a small-bore, high field animal scanner.[Photographs acquired from the Center for In Vivo Microscopy at Duke University Medical Center, an NIH/NIBIB National Biomedical Technology Resource Center].

While gradient and RF coil technology spans numerous decades of history (as early as Paul Lauterbur's ground-breaking inception of zeugmatography [Lauterbur 1973]), the stringent technical requirements imposed by the mouse physiology for cardiac imaging have led to optimization of prior technology or to recent introductions of new technologies, as reviewed and discussed briefly in this section. Interestingly, large-bore high-field systems emerged during the past decade at field strengths spanning 4.7, 7.1, 9.4, and 11.7T maintaining, however, the same design technology as originally developed for human superconducting

**3.2. Advances in specialized hardware for high field animal MRI** 

*3.2.1. High field magnets* 

systems in the late 1980's and 1990's.

The requirements for increased spatial resolution acquisitions (of orders matching cellular size of approximately 100-200 μm3) led to fast switching, high-amplitude (up to 1000 mT/m), high-slew rate (up to 11250 mT/s) gradients (exhibiting linearity of better than ±3-5%/mm) for ex-vivo constructions of atlases [Johnson 2002] or in-vivo high-resolution isotropic cardiac imaging [Bucholz 2008, Perperidis 2011]. Often associated with extra inserts (of approximately 6-20 cm in inner diameter), such gradients impose further spatial restrictions in animal placement and monitoring during cardiac imaging. However modern, commercially available systems are often equipped with actuator-controlled manual or computerized cradle positioning systems.

## *3.2.3. RF coils*

In a similar fashion to the evolution and migration of magnet and gradient technology from human to mouse applications, advances were noted in RF coils. Of all three hardware components (magnets, gradients, transmit/receive coils), RF coils received most of the attention of the research community for mouse cardiac MRI [Doty 2007]. Despite the use of simple resonant coil loops for cardiac studies early on, image-based phenotyping attempts [Wiesmann 2000, Wiesmann\_Card\_Magn\_Reson 2001] redirected attention to highthroughput imaging and to the deployment and construction of multiple-mouse phased arrays [Bock 2003, Ramirez 2010]. Availability of multiple receivers, in combination with parallel imaging techniques (SENSE, GRAPPA), have reinforced attempts for highthroughput mouse acquisitions at high fields [Sosnovick 2007, Schneider 2010]. More recent developments of scaled-down birdcage coils have provided optimal imaging solutions for single mouse MRI as exemplified by direct comparative studies with simple surface [Markiewicz 2006], planar, and cylindrical spiral multi-turn coils (Figure 7) [Constantinides\_Concepts 2011].

Study of the Murine Cardiac Mechanical Function Using Magnetic Resonance Imaging:

loading effects between the RF and gradient coil inserts, imposing the need for use of specially constructed isolation shields. The inhomogeneous response of such coils, obviously translates to issues associated with artifacts, imperfect application of magnetization preparation pulses, or biased quantitative measures in cardiac functional and perfusion imaging. Furthermore, flip-angle sensitive techniques, such as fat saturation in cardiac MRI cannot be applied, due to the inherent variations in the flip angle spatial

**3.3. Conventional and new imaging methodologies for mouse cardiac imaging** 

non-uniform data regridding-Fourier transformations) is attempted below.

technology in clinical scanners that eventually allowed ultra-short TR imaging.

well as other prep-pulses including inversion, DENSE, spin-tagging and others.

For mouse cardiac imaging, segmented k-space acquisitions in an interleaved fashion are executed (to allow reconstruction of multi-frame images at spatial resolutions reaching 87 μm isotropic) based on a trigger pulse (prospective [Cassidy 2004, Bovens 2011], retrospective [Bishop 2006, Bovens 2011], or self-gated [Hiba 2007]), or a combination of the ECG and the respiratory [of order of 50-120 breaths/minute approximately] signal (thereby minimizing cardiac motion and respiratory artifacts). Navigator echoes may also be used prior to the gating pulse with a number of preparation pulses (pre-pulses) often executed immediately after the trigger pulse (Figure 8). Such optional pulses may include fat suppression and spectral-spatial pulses for black-blood imaging and/or fat suppression, as

Similar to hardware developments numerous pulse acquisition schemes migrated from human cardiac applications to the mouse. Nevertheless, a number of new techniques emerged over the years that include cartesian (rectilinear) and radial (spiral, twisted projection, other) data acquisition schemes for fast mouse cardiac imaging. A brief review and analysis of the most-successful imaging and reconstruction practices (fast Fourier and

Two were the major types of Cartesian imaging acquisition techniques that migrated from prior human MRI efforts, extensively used for mouse cardiac imaging. They are both gradient echo sequences and adhere to lexicographic k-space or rectilinear sampling and are based on spin-warp imaging. The first, a steady state incoherent pulse sequence, was developed by Haase [Haase 1986] and became known as Fast Low Angle Shot and was dubbed with the acronym FLASH (equivalent to Spoiled Gradient Recalled Imaging, later developed by General Electric). The second type, a steady state coherent sequence known as Steady State Free Precession (SSFP) was developed early on by Carr [Carr 1958], employing ultra-short repetition times (TR) for imaging later adopted in an equivalent sequence with gradients by Hinshaw [Hinshaw 1976] and Frahm et al. [Frahm 1987], a sequence known as Fast Imaging with Steady Precession (FISP) [with the equivalent acronyms of Gradient Refocussed Acquisition at Steady State (GRASS) and balanced Fast Field Echo (B-FFE), as the GE's and Phillip's equivalent versions]. While SSFP was neglected for many years it regained tremendous interest over the last decade following advances in gradient

distribution, thereby limiting their potential applications.

*3.3.1. Cartesian k-space pulse sequence acquisitions* 

The Current Status, Challenges, and Future Perspectives 359

Traditionally, RF surface coils reduce the spatial coverage (and hence the field-of-view) compared to volume coils, while maintaining higher local signal-to-ratio (SNR). The limitation for expansion of the spatial region of surface coil coverage (while maintaining or improving SNR), yielded to phased array designs, often in combination with circularlypolarized volume transmit coils. Introduction of alternative coils to phased array designs, such as the one shown presents one of the first attempts to ameliorate such a surface coil limitation, in consideration of the small spatial scales, and the complexity and cost of phased arrays. Additionally, the use of such a surface coil is not prohibitive in terms of its concurrent use with dedicated and specially designed transmit (or receive coil arrays).

Specifically, imaging results showed improved performance of the cylindrical spiral coil in comparison to the flat counterpart. The cylindrical coil has increased field of penetration that allows visualization of the entire lateral and inferior myocardial walls with adequate relative SNR (rSNR). Its performance compares well and outperforms its flat equivalent in septal, inferior, anterior, and lateral myocardial areas (rSNR improvement between 27 and 167%), despite its non-optimal placement and positioning on the mouse (anteriorly and laterally), in this particular case. Its design and response, that clearly exploits the additive effect of the transverse B1-field component, also compares favorably with a commercially available birdcage coil within the region where the mouse heart resides. However, the birdcage coil exhibits the best performance associated with three to five times higher rSNR values over the entire left ventricular myocardial regions. Promising are anticipated uses and applications of new hyperpolarized cryostat volume (birdcage) coils that have recently become commercially available [Kovacs 2005].

The fact that the constructed coils are transceiver surface resonators, certainly presents a conceptual and practical limitation for their wide-spread use, in the context of inhomogeneous excitation and their ability to allow quantitative cardiac mouse imaging. Direct use of such coils for high field mouse cardiac imaging is likely to result in increased loading effects between the RF and gradient coil inserts, imposing the need for use of specially constructed isolation shields. The inhomogeneous response of such coils, obviously translates to issues associated with artifacts, imperfect application of magnetization preparation pulses, or biased quantitative measures in cardiac functional and perfusion imaging. Furthermore, flip-angle sensitive techniques, such as fat saturation in cardiac MRI cannot be applied, due to the inherent variations in the flip angle spatial distribution, thereby limiting their potential applications.

#### **3.3. Conventional and new imaging methodologies for mouse cardiac imaging**

Similar to hardware developments numerous pulse acquisition schemes migrated from human cardiac applications to the mouse. Nevertheless, a number of new techniques emerged over the years that include cartesian (rectilinear) and radial (spiral, twisted projection, other) data acquisition schemes for fast mouse cardiac imaging. A brief review and analysis of the most-successful imaging and reconstruction practices (fast Fourier and non-uniform data regridding-Fourier transformations) is attempted below.

#### *3.3.1. Cartesian k-space pulse sequence acquisitions*

358 Practical Applications in Biomedical Engineering

[Constantinides\_Concepts 2011].

become commercially available [Kovacs 2005].

In a similar fashion to the evolution and migration of magnet and gradient technology from human to mouse applications, advances were noted in RF coils. Of all three hardware components (magnets, gradients, transmit/receive coils), RF coils received most of the attention of the research community for mouse cardiac MRI [Doty 2007]. Despite the use of simple resonant coil loops for cardiac studies early on, image-based phenotyping attempts [Wiesmann 2000, Wiesmann\_Card\_Magn\_Reson 2001] redirected attention to highthroughput imaging and to the deployment and construction of multiple-mouse phased arrays [Bock 2003, Ramirez 2010]. Availability of multiple receivers, in combination with parallel imaging techniques (SENSE, GRAPPA), have reinforced attempts for highthroughput mouse acquisitions at high fields [Sosnovick 2007, Schneider 2010]. More recent developments of scaled-down birdcage coils have provided optimal imaging solutions for single mouse MRI as exemplified by direct comparative studies with simple surface [Markiewicz 2006], planar, and cylindrical spiral multi-turn coils (Figure 7)

Traditionally, RF surface coils reduce the spatial coverage (and hence the field-of-view) compared to volume coils, while maintaining higher local signal-to-ratio (SNR). The limitation for expansion of the spatial region of surface coil coverage (while maintaining or improving SNR), yielded to phased array designs, often in combination with circularlypolarized volume transmit coils. Introduction of alternative coils to phased array designs, such as the one shown presents one of the first attempts to ameliorate such a surface coil limitation, in consideration of the small spatial scales, and the complexity and cost of phased arrays. Additionally, the use of such a surface coil is not prohibitive in terms of its concurrent use with dedicated and specially designed transmit (or receive coil arrays).

Specifically, imaging results showed improved performance of the cylindrical spiral coil in comparison to the flat counterpart. The cylindrical coil has increased field of penetration that allows visualization of the entire lateral and inferior myocardial walls with adequate relative SNR (rSNR). Its performance compares well and outperforms its flat equivalent in septal, inferior, anterior, and lateral myocardial areas (rSNR improvement between 27 and 167%), despite its non-optimal placement and positioning on the mouse (anteriorly and laterally), in this particular case. Its design and response, that clearly exploits the additive effect of the transverse B1-field component, also compares favorably with a commercially available birdcage coil within the region where the mouse heart resides. However, the birdcage coil exhibits the best performance associated with three to five times higher rSNR values over the entire left ventricular myocardial regions. Promising are anticipated uses and applications of new hyperpolarized cryostat volume (birdcage) coils that have recently

The fact that the constructed coils are transceiver surface resonators, certainly presents a conceptual and practical limitation for their wide-spread use, in the context of inhomogeneous excitation and their ability to allow quantitative cardiac mouse imaging. Direct use of such coils for high field mouse cardiac imaging is likely to result in increased

*3.2.3. RF coils* 

Two were the major types of Cartesian imaging acquisition techniques that migrated from prior human MRI efforts, extensively used for mouse cardiac imaging. They are both gradient echo sequences and adhere to lexicographic k-space or rectilinear sampling and are based on spin-warp imaging. The first, a steady state incoherent pulse sequence, was developed by Haase [Haase 1986] and became known as Fast Low Angle Shot and was dubbed with the acronym FLASH (equivalent to Spoiled Gradient Recalled Imaging, later developed by General Electric). The second type, a steady state coherent sequence known as Steady State Free Precession (SSFP) was developed early on by Carr [Carr 1958], employing ultra-short repetition times (TR) for imaging later adopted in an equivalent sequence with gradients by Hinshaw [Hinshaw 1976] and Frahm et al. [Frahm 1987], a sequence known as Fast Imaging with Steady Precession (FISP) [with the equivalent acronyms of Gradient Refocussed Acquisition at Steady State (GRASS) and balanced Fast Field Echo (B-FFE), as the GE's and Phillip's equivalent versions]. While SSFP was neglected for many years it regained tremendous interest over the last decade following advances in gradient technology in clinical scanners that eventually allowed ultra-short TR imaging.

For mouse cardiac imaging, segmented k-space acquisitions in an interleaved fashion are executed (to allow reconstruction of multi-frame images at spatial resolutions reaching 87 μm isotropic) based on a trigger pulse (prospective [Cassidy 2004, Bovens 2011], retrospective [Bishop 2006, Bovens 2011], or self-gated [Hiba 2007]), or a combination of the ECG and the respiratory [of order of 50-120 breaths/minute approximately] signal (thereby minimizing cardiac motion and respiratory artifacts). Navigator echoes may also be used prior to the gating pulse with a number of preparation pulses (pre-pulses) often executed immediately after the trigger pulse (Figure 8). Such optional pulses may include fat suppression and spectral-spatial pulses for black-blood imaging and/or fat suppression, as well as other prep-pulses including inversion, DENSE, spin-tagging and others.

Study of the Murine Cardiac Mechanical Function Using Magnetic Resonance Imaging:

**Figure 8.** Rectilinear k-space sampling schemes employed for mouse cardiac imaging. (Left) FLASH imaging, and (right) SSFP. Shown on top are optional preparatory pulses and the inherent capability to perform multi-phase CINE acquisition based on the ECG (or a combined cardiac-respiratory) trigger pulse. [Gx, Gr=readout gradient, Gy, Gpe=phase encoding gradient, Gz, Gss=slice selection gradient, θ=flip

**Figure 9.** Non-rectilinear k-space sampling for cardiac imaging. Shown is a typical spiral readout scheme used in association with the DENSE (STEAM-based) preparatory pulses. (Left) ke represents the encoding pulse. The same pulse (un-encoding) is executed during the spiral readout (right). [Gx, Gr=readout gradient, Gy, Gpe=phase encoding gradient, Gz, Gss=slice selection gradient, θ=flip angle,

angle, α=RF pulse angle].

α=RF pulse angle].

The Current Status, Challenges, and Future Perspectives 361

**Figure 7.** (Top, Left) Mouse XFdtd B1-field simulation of a four-spiral cylindrical coil under loaded conditions with a mouse computational model (top, right) [color bar varies between 0 – -40 dB];(Bottom) Flat, cylindrical spiral, and birdcage axial phantom response using conventional SPGR pulse sequence acquisitions and corresponding cardiac MRI from each of the three coils in mice postmortem showing improved performance by the birdcage coil. [This figure originally appeared in Constantinides C et al., Intercomparison of performance of RF Coil Geometries for High Field Mouse Cardiac MRI. Concepts of Magnetic Resonance Part A, 38A(5):236-252, 2011, DOI 10.1002/cmr.a.20225, reproduced with permission from Wiley Publications].

#### *3.3.2. Spiral or non-cartesian k-space pulse sequences*

The ultrafast mouse heart rates, even under anesthesia, necessitate fast imaging acquisitions (within tRR=100-120 ms). While Cartesian sampling schemes cover k-space adequately nevertheless, non-rectilinear and spiral k-space acquisitions have gained tremendous interest recently as efficient and fast sampling schemes. Radial acquisition variants by Bucholz et al. [Bucholz 2008], and STEAM-based DENSE encoding followed by interleaved spiral acquisitions proposed by Zhong [Zhong 2010] (Figure 9) are also adopted for mouse cardiac functional imaging.

Study of the Murine Cardiac Mechanical Function Using Magnetic Resonance Imaging: The Current Status, Challenges, and Future Perspectives 361

360 Practical Applications in Biomedical Engineering

**Figure 7.** (Top, Left) Mouse XFdtd B1-field simulation of a four-spiral cylindrical coil under loaded conditions with a mouse computational model (top, right) [color bar varies between 0 – -40

dB];(Bottom) Flat, cylindrical spiral, and birdcage axial phantom response using conventional SPGR pulse sequence acquisitions and corresponding cardiac MRI from each of the three coils in mice postmortem showing improved performance by the birdcage coil. [This figure originally appeared in Constantinides C et al., Intercomparison of performance of RF Coil Geometries for High Field Mouse Cardiac MRI. Concepts of Magnetic Resonance Part A, 38A(5):236-252, 2011, DOI 10.1002/cmr.a.20225,

The ultrafast mouse heart rates, even under anesthesia, necessitate fast imaging acquisitions (within tRR=100-120 ms). While Cartesian sampling schemes cover k-space adequately nevertheless, non-rectilinear and spiral k-space acquisitions have gained tremendous interest recently as efficient and fast sampling schemes. Radial acquisition variants by Bucholz et al. [Bucholz 2008], and STEAM-based DENSE encoding followed by interleaved spiral acquisitions proposed by Zhong [Zhong 2010] (Figure 9) are also adopted for mouse

reproduced with permission from Wiley Publications].

cardiac functional imaging.

*3.3.2. Spiral or non-cartesian k-space pulse sequences* 

**Figure 8.** Rectilinear k-space sampling schemes employed for mouse cardiac imaging. (Left) FLASH imaging, and (right) SSFP. Shown on top are optional preparatory pulses and the inherent capability to perform multi-phase CINE acquisition based on the ECG (or a combined cardiac-respiratory) trigger pulse. [Gx, Gr=readout gradient, Gy, Gpe=phase encoding gradient, Gz, Gss=slice selection gradient, θ=flip angle, α=RF pulse angle].

**Figure 9.** Non-rectilinear k-space sampling for cardiac imaging. Shown is a typical spiral readout scheme used in association with the DENSE (STEAM-based) preparatory pulses. (Left) ke represents the encoding pulse. The same pulse (un-encoding) is executed during the spiral readout (right). [Gx, Gr=readout gradient, Gy, Gpe=phase encoding gradient, Gz, Gss=slice selection gradient, θ=flip angle, α=RF pulse angle].

In summary, numerous practical benefits are associated with mouse cardiac MR imaging, including the non-invasive nature of the technique, the inherent capability to map cardiac morphology and function, for both LV and RV chambers, and their motional patterns. High spatial and temporal resolution imaging can thus be achieved, through execution of highthroughput protocols, yielding direct, accurate estimates of global and regional indices of cardiac function, avoiding any assumptions whatsoever or model-based derivation approaches endorsed by other imaging techniques such as ultrasound.

Study of the Murine Cardiac Mechanical Function Using Magnetic Resonance Imaging:

emerging techniques and applications develop to investigate in-vivo sarcomeric force generation with contrast agents [Constantinides\_ABME 2011], the current scientific focus is

**Figure 10.** (Top) Typical bright blood true-short axis cardiac imaging at an apical, middle, and basal levels of the C57BL/6 murine heart using a liposomal contrast agent; (Bot) Black-blood spin-echo images

**Figure 11.** (A) ECG-gated four chamber view of a healthy mouse heart at 78 x 78 μm2 in-plane resolution, and (B) corresponding self-gated (retrospective) view of a separate mouse using a navigator echo at the same spatial resolution, at 11.7 T using FLASH-MRI [Images courtesy of Bruker Biospin

Correspondingly, estimated cardiac volumes can easily be computed using standard image processing tools and converted to absolute volume units using the voxel dimension and the myocardial tissue density. Hemodynamic indices such as end-diastolic (EDV) and end-

A B

still on image-based calculations of global indices of function.

of the murine heart at basal, mid, and apical levels**.** 

MRI, Ettlingen, Germany].

The Current Status, Challenges, and Future Perspectives 363

Despite the extensive use of cartesian imaging with adequate SNR performance and spatial resolution (using FLASH, SSFP, or FISP), 3D acquisition studies maybe more efficiently completed using radial or spiral imaging sequences, especially for dynamic cardiac imaging (with pharmacologic interventions or contrast agent infusions). Nevertheless important and critical drawbacks are associated with such sequences, including the necessity to maintain data density as sampling extends to outer k-space regions, the convoluted and complex reconstructions (often associated with data re-gridding, kernel de-convolution, filtering, and inverse fourier transformation), and inherently lower SNR performance than rectilinear imaging. Thus, the choice between cartesian and radial imaging reduces to a tradeoff between SNR, spatial resolution, and efficiency of data sampling for 3D coverage.
