Biomedical Imaging in Experimental Models of Cardiovascular Disease

Abstract

Major advances in biomedical imaging have occurred over the last two decades and now allow many physiological, cellular and molecular processes to be imaged non-invasively in small animal models of cardiovascular disease. Many of these techniques can be also used in humans, providing pathophysiological context and helping to define the clinical relevance of the model. Ultrasound remains the most widely used approach and dedicated high frequency systems can obtain extremely detailed images in mice. Likewise, dedicated small animal tomographic systems have been developed for magnetic resonance (MR), positron-emission tomography (PET), fluorescence imaging and computed tomography in mice. In this article we review the use of ultrasound and PET in small animal models as well as emerging contrast mechanisms in magnetic resonance such as diffusion tensor imaging, hyperpolarized MR, chemical exchange saturation transfer imaging, MR-elastography and strain, arterial spin labeling and molecular imaging.

Small animal models play a crucial role in cardiovascular investigation. The last two decades have witnessed major advances in the breadth and sophistication of molecular biology techniques to better understand and characterize small animal models of cardiovascular disease. Likewise, major developments have occurred in the techniques available to image these models noninvasively. We focus in this review on advances in ultrasound, magnetic resonance and positron emission tomography (PET) to image the left ventricle (LV) in small animal models. However, many of these techniques can also be used to image the right ventricle (RV) and the vascular system in mice, including models of arterial disease, valvular disease, aneurysm and thrombosis (see supplement).^1–3 In addition to providing insights into disease pathophysiology, biomedical imaging in small animal models is playing an increasingly important role in the drug development process.⁴

Assessment of Cardiac Function

The evaluation of systolic and diastolic function in experimental models of cardiovascular disease has relied mainly on echocardiography, both transthoracic and transesophageal, and more recently on MRI. The ability to perform echocardiography in mice was first enabled by advances in the design of ultrasound transducers for clinical use.⁵ Due to the small size and the rapid heart rate of rodent’s hearts (LV long axis of approximately 5–8mm in an adult mouse, heart rate of 500–700bpm depending on the strain), echocardiography in these species requires high frequency probes (at least 7MHz in rats, 10–12MHz in mice) and high frame rates (at least 120Hz), to ensure sufficient spatial and temporal resolution. Technological advances in ultrasound have allowed the development of very high frequency probes (20–50MHz), with significantly better spatial resolution. The frame rates using very high frequency probes are acceptable; a very high frame rate (close to 1000Hz) can be obtained using ECG gating. This approach, however, requires the mouse to be perfectly still and have a stable EKG tracing, and cannot be done in awake or lightly sedated animals. While the use of ultra-high frequency systems is appealing,⁶ accurate insights into murine pathophysiology can be obtained on standard clinical ultrasound systems.⁷

One particularly crucial aspect in physiological studies, especially in rodents, is the role of sedation or anesthesia. All sedatives or anesthetics used affect cardiovascular function, generally with bradycardia and negative inotropy. Measures of systolic function are steeply and positively correlated with heart rate in mice,⁸ whereas LV volumes are negatively correlated. Additionally, most sedatives and anesthetics do not induce a stable hemodynamic state; although investigators should strive to acquire images at a similar heart rate, such a measure may not be sufficient to eliminate the variability of measurements. Finally, different genetic murine backgrounds may display different sensitivities to anesthesia; for example, mice on a FVB background tend to become less bradycardic with isoflurane than mice on a C57BL6 background.⁹

Ideally, echocardiography should be performed in awake mice; this requires training the mice and is not extensively performed. Ketamine, contrary to most sedatives, has a tachycardic effect, at least at low doses, but induces a very light sedation. 2,2,2-tribromoethanol has little cardiovascular effect, however, it may induce peritonitis when injected serially. Isoflurane is extensively used by investigators, and in spite of some individual and temporal variability, it may be an acceptable compromise. The impact of anesthesia in mice is discussed in more detail elsewhere.⁹ Since hemodynamic conditions are crucial to cardiac interpretation, the type and dose of sedation/anesthesia used, and the heart rate at which the images were obtained should ideally be reported.

Early papers took advantage of the high spatial resolution and sample rate of M Mode to validate the calculation of LV mass with necropsy measurements,⁵ and pharmacological manipulations to document changes in pressure-dimensions in mice.^{10, 11} The morphological and functional measurements derived from M Mode are extensively used to this day (Figure 1), and are often sufficient to report differences in LV dimensions, wall thickness or fractional shortening produced by a genetic modification or a model affecting global LV function. However, M Mode is not appropriate in any model involving regional wall motion abnormalities, such as ischemic models or myocarditis. Calculating the LV ejection fraction (EF) from the M-Mode is extensively done and reported, however, makes an assumption that the LV is a sphere (cubed formula). This assumption is not only incorrect, but the magnitude of the error also increases in large and remodeled LVs, making comparison between groups potentially fraught with error. This error can be avoided by using two-dimensional images and a modified Simpson’s rule approach to measuring EF.

Figure 1: Functional and hemodynamic imaging in mice.

(A) M-Mode obtained from a parasternal short axis view. Vertical scale: each bar 0.1mm, horizontal scale: each bar 0.1sec. (B) Parasternal Long axis view and (C) Parasternal short axis view, midpapillary level. LV: left ventricle, RV: right ventricle, AW: anterior wall (anteroseptum), PW: posterior wall (inferolateral). (D). Pulsed Doppler of the mitral inflow velocities obtained from an apical 4 chamber view. Vertical scale: each bar 30cm/s, horizontal scale: each bar 0.1sec. (E). Pulsed Doppler of the lateral mitral annular velocities obtained from an apical 4 chamber view. Vertical scale: each bar 6 cm/s, horizontal scale: each bar 0.1sec. (F) Two-dimensional view of a parasternal short axis view at the level of the aortic and pulmonary valves. Ao: aortic valve, PA: pulmonary artery, RA: right atrium, LA: left atrium, RVOT: right ventricular outflow tract. (G) correct positioning of the Doppler sample. (H) Doppler tracing of the flow at the pulmonary valve level. PAT: pulmonary acceleration time, PET: pulmonary ejection time.

The spatial resolution of two-dimensional imaging has improved using very high frequency probes (Figure 1) and allows precise measurement of 2D-derived LVEF. A single plane measurement in the parasternal long axis view using a single method of disks, an area-length method, or speckle tracking, correlates reasonably with MRI volumes and EF.^{12, 13} Acquiring and measuring three (base, mid, apex) or multiple short axis views, separated by a known distance may be more precise but is more time consuming.^13–15 The 3D reconstruction of the LV from short axis views has been applied both to transthoracic and transesophageal echocardiography and can allow measurement of LV and RV volumes,^14–18 however, they are time consuming, and rarely used. This method can also be useful in measuring a wall motion score index, or a percentage of akinetic myocardium, which reflects MI scar in non reperfused myocardial infarction models.¹⁹

Cine MRI of the murine heart can be performed on dedicated small animal scanners. It provides tomographic images with high spatial and temporal resolution, allowing the anatomy and function of both the LV and RV to be assessed.^{20, 21} The high flow of blood in the murine heart creates excellent contrast between the blood pool and myocardium via a spin-refreshment (inflow) effect. However, due to its cost, limited availability, and need for moderate to deep anesthesia, cardiac MRI is less frequently used to measure LV volumes and EF in mice than echocardiography. The strengths and limitations of several imaging modalities for the assessment of cardiovascular disease in mice are provided in Table 1.

Table 1:

Selection of Modality for Imaging Small Animal Models of Cardiovascular Disease.

	Ultrasound	MRI	SPECT/PET	Optical (In Vivo)
Hardware	Point of care	Best performed with dedicated small animal scanners	Most systems are integrated with micro-CT scanners	Intravital microscopy, fluorescence tomography, optoacoustic systems
Complexity	Low complexity, no danger, no barriers to use	Requires significant expertise	Requires controlled environment and radio-pharmacy	Noninvasive tomographic imaging possible in NIR range.
Anesthesia/Sedation	Can be avoided	Required	Required	Required
Anatomy	Good	Excellent. No contrast agent needed. LV, RV, atria, aorta and PA all well visualized	Excellent if integrated with CT	Emerging
Function	Excellent, can be performed in awake animals (physiology not perturbed by sedation)	Excellent but requires anesthesia. LV and RV well visualized and quantified	ECG-Gated SPECT and PET feasible in mice	Volumetric optoacoustic techniques emerging
Strain	Excellent	Excellent, several mechanisms/techniques	-	Emerging
Perfusion	Excellent	Excellent	Gold Standard	Excellent
Viability	Not routine	Gold Standard - LGE	Gold Standard −18FDG	Emerging
Hemodynamics	Excellent	Excellent	-	Emerging
Elastic Properties	Emerging	Emerging	-	Emerging
Metabolism	-	Gold Standard	Gold Standard	Emerging
Microstructure	Not routine	Excellent	-	Excellent
Molecular Imaging (intra-vascular)	Excellent	Excellent	Excellent	Excellent
Molecular Imaging (extra-vascular)	-	Excellent	Excellent	Excellent

CT = computed tomography, NIR = near infrared, PA = pulmonary artery, LGE = Late Gadolinium Enhancement, FDG = Fluorodeoxyglucose.

T1, T2 and Late Gadolinium Enhancement

The development of late gadolinium enhancement (LGE) has allowed infarct size in mice and rats to be precisely quantified.^{22, 23} Gadolinium chelates accumulate passively and transiently in the extracellular space and their detection is context specific. In an acute infarct LGE results from the disruption of the cardiomyocyte membrane,²⁴ while in a chronic infarct LGE is produced by expansion of the extracellular space due to fibrosis.²⁵ LGE in the murine heart correlates extremely well with TTC and histological stains for fibrosis.^{22, 23} The acquisition of LGE images in mice can be performed in one of two ways.²³ The first involves the use of a 180° inversion prepulse followed by a delay (inversion time) for the magnetization in healthy myocardium to null. The rapid heart rate and flow of blood in the murine heart also allow LGE images to be obtained by acquiring cine images with a very high (T1-weighted) flip angle.²³

T2-weighted MR imaging is sensitive to the presence of edema and has been used to measure the area-at-risk (AAR),²⁶ although there has been some debate in the clinical literature about the exact pathophysiological significance of elevated T2 (transverse magnetic relaxation time).²⁷ In the murine heart conventional T2-weighted imaging is challenging to perform. T2-preparation prepulses, however, have been used in mice to detect areas of edema,^{28, 29} which correlate well with the AAR by histology or microsphere injection.^{28, 29} The longitudinal relaxation time (T1) in the myocardium is also increased by edema and can be used to detect acute injury.³⁰ In addition to measuring the AAR, T1 mapping in mice may also be able to provide a measure of microvascular function.³¹

While measuring the T2 of the myocardium in the murine heart is more challenging than in humans, T1 mapping is in many ways simpler. If the mouse’s heart rate is maintained above 500 beats per minute then the exponential T1 recovery curve can be sampled at least every 120ms to derive an accurate T1 value.^{24, 32} More sophisticated approaches, such as the modified Look-Locker Imaging or MOLLI sequence can be used in mice,³³ but are vital in humans because of their slower heart rate. In clinical practice a normal reference range for T1 values must be acquired on each individual scanner.³⁴ This is fortunately not necessary in small animal imaging provided the active and control groups are imaged with the same sequence and under the same experimental conditions.

The development of T1 mapping has been particularly useful in the characterization of murine models of pressure overload and cardiomyopathy.^35–37 If T1 maps of the myocardium are acquired before the injection of gadolinium and at several time points after its injection, the extracellular volume fraction (ECV) of the myocardium can be derived.³⁴ The ECV value is derived from the slope of R1 (1/T1) in the myocardium/R1 in the blood with a correction factor for hematocrit.³⁴ The ECV of the murine heart is similar to humans (24–26%),^{35, 36} and is increased in the presence of LV fibrosis.^35–37 It should be stressed, however, that an increase in ECV is not specific for fibrosis since ECV can also be increased by cell death and inflammation in acute injury and by chronic process such as amyloid.³⁴

Imaging the Viscoelastic Properties of the Myocardium

In mice the assessment of diastole is hampered by the high heart rate and frequent fusion of the E and A wave of mitral inflow. There are no optimal ways to avoid this fusion without decreasing the heart rate below physiological values. Careful titration of isoflurane decreases the heart rate, however, inter-mice variability still exists and fusion can be observed in some mice with a heart rate as low as 400 bpm. Furthermore, the simultaneous negative inotropic effect of increasing isoflurane is itself variable. Ivabradine, an inhibitor of the I_f current, increases the duration of the spontaneous depolarization of the sinus node thus selectively decreases the heart rate without negative inotropic effects. In mice, ivabradine decreases the heart rate by as much as 20%.^38–40 Echocardiography performed 6 hours after an intraperitoneal injection of 3μg/g body weight ivabradine (total of 2 injections in 24 hours) in a model of sepsis, reported a decrease in the heart rate from 601 (547–612) bpm to 447 (430–496) bpm, without changes in LV ejection fraction or anterior wall strain rate.³⁸ However, intraperitoneal injection of ivabradine a few minutes prior to echocardiography is not recommended as this results in great heart rate variability, both between the mice and over time.

Hemodynamics in the murine heart can be readily assessed with Doppler imaging. Most investigators report the mitral inflow pattern (E/A), the amplitude of the medial mitral annular early diastolic velocity (e’), and the ratio of E/e’ as a parameter of left atrial pressure (Figure 1) to assess diastolic function by echo. Although e’ and E/e’ are important parameters of diastole, the analysis of diastolic dysfunction in humans also includes measurement of the left atrial volume index and of the tricuspid regurgitation maximum velocity.⁴¹ Rodents very rarely develop tricuspid regurgitation even in models of pulmonary hypertension. Investigators have reported left atrial volume index using the antero-posterior dimension in the parasternal long axis view and the mediolateral dimension in the parasternal short axis view, then calculating the LA volume using a prolate ellipse formula, and normalizing by body weight.^{42, 43} The LA volume index increased with age and displayed fair correlations with invasive measures of diastolic function.⁴³ Other investigators have reported LA area measured on the 4-chamber view and demonstrated increases in this parameter in models of chronic volume overload (AAV fistula), DOCA-salt,⁴⁴ and TAC.⁴⁵

The measurement of hemodynamic indices in the murine heart is feasible with phase-contrast MRI,^{46, 47} however, MRI has been used more extensively to assess the material properties of the myocardium. MRI-based elastography of the heart is highly analogous to ultrasound-based shear wave elastography.⁴⁸ An acoustic wave from an external device is directed towards the heart and its propagation through the myocardium can be used to calculate myocardial stiffness.⁴⁹ In brief motion sensitizing magnetic gradients (small subsidiary magnetic fields) are applied along each Cartesian axis and phase difference images are acquired.^{49, 50} The term phase here refers to the phase (rotational angle) of the magnetization and not to the physiological phase of the cardiac cycle. A Fourier transform of the phase-difference image produces a complex wave image in each direction from which myocardial stiffness can be calculated. While highly accurate and very promising, the technique in mice is challenging and still fairly time consuming.^{50, 51}

The stiffness of the myocardium can also be estimated using ultrashort TE (UTE) MRI,⁵² and a contrast mechanism known as T1-rho,⁵³ without the need to inject an exogenous contrast agent. The echo-time (TE) in MRI refers to the time taken to refocus the magnetization and is not related to echocardiography. To detect the magnetization it must be flipped out of the longitudinal plane and into the transverse plane (orthogonal to the main magnetic field). Once in the transverse plane the magnetization decays exponentially with a time constant (T2) determined by the properties of the tissue. In stiff materials such as bone and tendon the T2 is extremely short and the signal cannot be detected with a conventional MR approach. UTE sequences have, therefore, been developed to image these tissues and can also be applied to detect fibrosis in the heart.⁵² In fibrotic/stiffer myocardium signal will be present in the UTE image but not in an image acquired with a conventional TE. In contrast, in normal myocardium substantial signal will be present in both images. This dual echo UTE (DUTE) approach has been implemented in mice with myocardial infarction with encouraging results.⁵²

T1-rho is an advanced MR contrast mechanism that has been shown to be sensitive to cardiac fibrosis.⁵³ Unlike conventional MRI, where a single short radiofrequency pulse is used to tip the magnetization from the longitudinal plane into the transverse plane, with T1-rho imaging a continuous radiofrequency pulse is applied parallel to the magnetization in the transverse plane.⁵³ This technique, known as spin-locking, prevents the loss of magnetization via T2 decay. Instead, the magnetization undergoes precession in the transverse plane with a frequency in the kHz range. Molecular moieties with vibrational frequencies in the kHz range exchange energy with these protons, resulting in the decay of the T1-rho signal. This energy exchange mechanism occurs at a very different range of frequencies to those involved in T1 relaxation (recovery of longitudinal magnetization). In fibrotic myocardium T1-rho is prolonged and the technique may be able to detect both local and diffuse fibrosis.⁵³ T1-rho imaging of the heat has been successfully performed in large animals and humans,^{54, 55} and its implementation in mice should be feasible.

Imaging of Myocardial Metabolism

Magnetic resonance can be performed on several atoms including protons (¹H), fluorine-19 (¹⁹F), carbon-13 (¹³C) and phosphorus-31 (³¹P). Spectroscopy of (³¹P) can be used to quantify high-energy phosphates such as ATP and phosphocreatine,^{56, 57} and (¹³C) to characterize key intermediates in cardiac metabolism.^58–62 The rate of ATP synthesis can even be imaged in the beating mouse heart.⁶³ The abundance of (³¹P) and (¹³C) in in the body is far lower than protons, which limits their detection. However, this can be addressed using hyperpolarization,⁵⁸ which dramatically increases the signal or polarization of these atoms. In brief, by cooling a suitable molecule such as pyruvate down to 4° Kelvin, while applying energy in the microwave range, the magnetic polarization can be increased by >10,000-fold.⁵⁸ At this level of polarization (¹³C) images or spectra can be obtained rapidly and repeated longitudinally to provide a dynamic picture of metabolism (Figure 2). The principal limitation of the approach is that once the specimen is heated for injection the degree of hyperpolarization rapidly dissipates. Imaging must, therefore, be performed immediately after sample heating, which creates technical challenges and significant logistical complexity. Nevertheless, valuable insights into metabolism in the heart have been obtained with hyperpolarized MR in small animal models and humans.^58–62

Figure 2: Imaging of myocardial metabolism.

(A) MRI of rat hearts after the injection of hyperpolarized (¹³C)-pyruvate. Maps of pyruvate and its downstream metabolites (bicarbonate, lactate, pyruvate hydrate, alanine) are shown with no suppression (NS) of the blood pool signal or with the suppression of flow (FS) in the blood pool. Fed rats injected with the pyruvate dehydrogenase kinase inhibitor (dichloroacetate, DCA) show an increase in myocardial bicarbonate. Reproduced with permission.⁵⁹ (B) (¹³C)-lactate signal produced by macrophages infiltrating a healing infarct in a rat heart. When macrophages and monocytes are depleted a substantial reduction in signal is seen. Reproduced with permission.⁶⁰ (C) Injection of hyper-polarized (¹³C)-fumarate in control and infarcted rats. Conversion to (¹³C)-malate is seen in the infarct. Reproduced with permission.⁶¹ (D) Chemical exchange saturation transfer (CEST) imaging of myocardial creatine content. In mice fed a high-fat diet the CEST creatine signal is decreased. Reproduced with permission.⁶⁵

The proton pool can be divided into those protons that spin at the resonance frequency of water (on-resonance) and those that belong to chemical moieties such as amides that are shifted slightly off-resonance. There has been significant interest in exploiting the transfer of energy between these two pools using a technique known as chemical exchange saturation transfer (CEST) imaging.⁶⁴ CEST detects the transfer of energy from the off-resonance to the on-resonance proton pool, which reduces the on-resonance signal. Off-resonance species of interest in the heart include creatine,^{64, 65} which could provide a proton-based assay of cardiac energetics and avoid the cost and complexity of (³¹P) and (¹³C) imaging. However, several technical challenges must be addressed when performing CEST imaging of the heart. Meticulous attention must be paid to achieving as uniform a main magnetic (B0) field as possible by positioning the heart at the isocenter of the magnet and paying close attention to field shimming. Notwithstanding the technical challenges, in vivo CEST imaging of the heart has been performed in small animal models,⁶⁵ (Figure 2) and holds significant promise.

The imaging of myocardial metabolism with (¹¹C)-based PET has been performed in mice,⁶⁶ and is analogous to hyper-polarized MRI of (¹³C) with several important distinctions. The superb sensitivity of PET allows very low doses of a (¹¹C)-labeled metabolite to be injected, which may better reflect normal physiology. However, kinetic analysis of the (¹¹C) PET signal provides limited information on the flux of the molecule through various intermediates in a metabolic pathway. The cost and complexity of PET (cyclotron, radiopharmacy) also limit its broader application in mice and optical techniques, including fluorescence and Cerenkov luminescence, may allow some aspects of myocardial energetics to be assayed with significantly less complexity.⁶⁷

Myocardial Perfusion

Anatomical imaging of the coronary arteries to detect congenital abnormalities or stenoses is feasible in mice using micro-CT,⁶⁸ and perfusion imaging, using contrast echocardiography or MRI, can delineate the area-at-risk or infarction size in myocardial ischemia and infarction models.^69–71 Investigators have used pulsed Doppler with a sample placed on the left coronary artery in the parasternal short axis or long axis view to measure coronary flow velocities, at rest and with vasodilators (adenosine or isoflurane), and calculate the coronary flow reserve.^72–74 Semi-quantitative perfusion imaging has also been validated using myocardial contrast echocardiography.^{75, 76} The coronary reserve in healthy young wild-type mice is approximately 2.5, and is decreased in aged or obese mice.^{72, 76, 77} Myocardial perfusion imaging in mice using echocardiography has given excellent signal using probes at frequencies of 10–15 MHz, as these frequencies are close to the resonance frequency of commercially available microbubbles (Figure 3). Microbubbles can also be decorated with ligands to adhesion molecules, selectins and von Willebrand factor to detect endothelial abnormalities during reperfusion.⁷⁸ Stressing the mouse heart can be challenging as cardiac sympathetic tone predominates at rest at room temperature.^{79, 80} Nevertheless, investigators have been able to unmask subtle functional cardiac abnormalities using stress echocardiography. Pharmacological stress, such as dobutamine has been used,^{76, 81} but recently echocardiography has also been performed immediately after peak exercise on a treadmill, which may provide more clinically relevant results.^{82, 83}

Figure 3: Imaging of perfusion and strain in the murine heart.

(A, B) Measurement of perfusion in a healthy mouse with microbubbles. (A) Image acquired immediately after high energy pulses and (B) after replenishment of the myocardium with contrast. LV: left ventricle, RV: right ventricle, AW: anterior wall, PW: posterior wall. (C) Arterial spin labeling in mice fed a standard diet (SD) and a high fat high sucrose diet (HFHSD), which decreases myocardial perfusion. Reproduced with permission.⁹¹ (D) Arterial spin labeling in a mouse with a myocardial infarction showing a dramatic reduction of myocardial blood flow in the infarct zone. Reproduced with permission.³² (E) Echo microbubbles targeted to Von-Willebrand factor are retained in injured myocardium after ischemia-reperfusion. Reproduced with permission.⁷⁸ Radial strain tracings obtained from the parasternal long axis view. The upper panel shows the endocardial and epicardial tracing, the lower panel the strain-time curve during one cardiac cycle. Horizontal scale: each bar 0.1sec. LV: left ventricle, RV: right ventricle, AW: anterior wall, PW: posterior wall. (G-I) DENSE imaging of mouse with myocardial infarct. (G) Late gadolinium enhancement of infarct zone. (H) Displacement map of each voxel in systole. Remote zone – black arrows, infarct zone red arrows. (I) The primary eigenvector (direction) of strain in the remote zone is radial, as indicated by the orientation of the black lines, but is highly perturbed in the infarct zone. (G-I) Reproduced with permission.¹⁰⁵ (J) Circumferential strain by DENSE in an infarcted mouse is highly reduced in the infarct zone (arrows). Reproduced with permission.¹⁰⁸

PET imaging provides several approaches to image myocardial perfusion including [¹³N]-ammonia, [⁸²Rb]-Rubidium and [¹⁵O]-water. The radioactive half-life of [¹⁵O] is extremely short, which makes it suited only to highly specialized PET centers. [⁸²Rb] is produced by a generator and does not require a cyclotron for its production, providing a useful option to sites without one. [⁸²Rb] perfusion imaging has been successfully performed in rats with myocardial infarction and amiodarone therapy.^{84, 85} Likewise, [¹³N]-ammonia has been used to image perfusion in mice.⁸⁶ [¹⁸F]-Flurpiridaz is a novel PET tracer that binds to mitochondrial complex 1 in a flow dependent manner, but its use in rodents has been limited.

The evaluation of myocardial perfusion with MRI can be performed using endogenous and exogenous contrast mechanisms. The two most widely used endogenous approaches involve arterial spin labeling (ASL) and BOLD (blood oxygen level dependent) imaging. The physical basis of BOLD lies in the paramagnetic properties of deoxyhemoglobin (deoxyHb), which serves as the endogenous tracer and affects T2 and T2*.⁸⁷ An increase in myocardial work increases the production of deoxyHb but this is offset by an increase in myocardial blood flow/volume, which decrease deoxyHb concentration. The principal limitation of BOLD imaging lies in the relative (stress/rest ratio) nature of the measurement. The technique can detect an increase in myocardial blood flow in humans/large animals from a breath-hold induced rise in myocardial CO₂,^{88, 89} but the feasibility of this in rodents is unclear.

ASL classically relies on the upstream inversion of protons, which affect T1 in a downstream tissue in a flow dependent manner.⁹⁰ In the heart, however, the approach most widely used is to measure T1 in a section of the myocardium after applying slice-selective and non-selective inversion prepulses.⁹⁰ In the slice-selective case only the spins in a thin slice of interest (~1mm thick) are inverted, whereas in the non-selective case the spins in the entire heart are. In the presence of robust myocardial blood flow a significant difference in T1 in the slice of interest will be seen with the selective vs. non-selective prepulses, allowing myocardial blood flow (ml/g/min) to be calculated (Figure 3).^{32, 90, 91} ASL in mice has shown that their resting myocardial blood flow is greater than 5ml/g/min (Figure 3),^{32, 91} which is substantially higher than humans. Fist pass perfusion with Gd in the rodent heart can be performed but requires the images to be acquired extremely rapidly. Standard techniques, such as Fermi deconvolution, can be used to derive a quantitative value of perfusion (ml/g/min).⁹²

Myocardial Mechanics

Myocardial deformation indices (strain and strain rate) can be imaged by echocardiography (Figure 3) and have the ability to detect subtle and early abnormalities of systolic function and to predict subsequent major adverse cardiac events in a variety of cardiovascular pathologies. Tissue Doppler imaging (TDI) allows the high temporal sampling of tissue velocities in a direction parallel to the beam (radial direction on the parasternal views). Strain rate can then be integrated from the velocities.⁹³ Radial strain rate obtained from TDI was the first parameter to be validated in mice using microsonometry,⁹⁴ and was shown to predict mortality in a mouse model of doxorubicin cardiotoxicity.⁹⁵ Speckle tracking follows the pixels on 2 dimensional images and allows the detection of myocardial motion independent of its direction. Strain can be integrated from myocardial motion.⁹⁶ The temporal resolution of speckle tracking depends on the frame rate and the algorithm applied. Speckle tracking derived strain was also shown to decrease early and predict outcome in models of myocardial infarction, pressure overload and sepsis.^97–100 The strain values reported in the literature for healthy 2 months old C57BL6 mice are quite comparable to those of humans (approximately −15 to −20% for longitudinal strain, 32–38% for radial strain, 29–30 for circumferential strain).^{97, 100} The strain rate reported using tissue Doppler imaging however is higher than that reported by speckle tracking (20–27/sec,^{94, 95, 101} vs. 8–9/sec,^{97, 100}), raising the issue of an insufficient frame rate on the 2 dimensional images.¹⁰²

Myocardial strain in rodents can be imaged using several MRI-based approaches. These include feature tracking, velocity-encoding, tagged cine imaging and a technique known as displacement encoded stimulated echo (DENSE) imaging (Figure 3). While feature tracking is extremely convenient,¹⁰³ the most rigorous 3D approaches involve the use of tagging and DENSE.^104–106 MR tagging involves magnetic labeling of protons in the transverse plane to eliminate their signal by a process known as saturation. The tags persist for several hundred milliseconds, which in a mouse is more than sufficient to measure strain across the entire cardiac cycle. The principal limitation of tagging is that the spatial resolution of the tag lines is substantially lower than the resolution routinely obtainable with MRI.¹⁰⁷ The post-processing of tagged cines is also not straightforward and dedicated analysis software is required to do this.

DENSE imaging overcomes many of the limitations of MR tagging. It produces strain images of the heart at high spatial resolution and has been extensively validated in mice.^{105, 106} DENSE uses the stimulated echo refocusing mechanism (three successive 90° excitation pulses) and incorporates motion-sensitizing gradients after the first and third pulses. Of note, the polarity (direction) of the two motion-sensitizing gradients is opposite. If a proton remains stationary the effect of the first motion-sensitizing gradient on its phase (rotational angle of magnetization) will be completely reversed by the second motion-sensitizing gradient. However, if the proton undergoes translation between the two motion-sensitizing gradients the effects on its phase are not equal and a net phase difference results. The motion of the protons in each voxel can be calculated from the phase difference that accrues during the DENSE sequence in that voxel. 3D DENSE imaging has shown that the primary eigenvector (predominant direction) of myocardial strain is radial and that the secondary and tertiary eigenvectors of strain both have significant circumferential and longitudinal components (Figure 3).¹⁰⁶ DENSE maps of myocardial strain can be easily co-registered with corresponding images of LGE, perfusion, metabolism and inflammation to provide a detailed phenotypic picture of the heart.¹⁰⁸

Myocardial Microstructure

The diffusion of water in a tissue can be used to derive several useful metrics of its microstructure.¹⁰⁹ Water diffuses most readily along the long axis of the cardiomyocytes and least freely orthogonal to them.¹¹⁰ Diffusion tensor imaging (DTI) can be used to resolve the preferred directions (eigenvectors) of diffusion in the heart and hence its microstructure.^{109, 110} Most DTI studies of the heart in mice and rats have been performed ex vivo,^111–114 but with specialized high-performing gradient systems and sequences, the technique can be performed in vivo.²⁹

Several useful scalar terms can be derived from the amount (eigenvalue) of diffusion along each eigenvector. The mean diffusivity (MD) reflects the average of the three eigenvalues and increases in the myocardium when there is an increase in its ECV and water content due to edema and/or cardiomyocyte death.^{29, 111} In mice serial in vivo DTI has shown that MD remains elevated for approximately 3 weeks after ischemia-reperfusion.²⁹ The fractional anisotropy (FA) of the myocardium is determined from the ratios of the diffusion eigenvalues and reflects the degree of tissue anisotropy and coherence. FA is reduced in acute injury, often corresponding with an increase in MD,^{29, 111} but it is also perturbed in conditions associated with cardiomyocyte disarray such as hypertrophic cardiomyopathy (HCM).¹¹⁵

The orientation of cardiomyocytes in the heart is reflected by the primary eigenvector of the diffusion tensor and is often described in terms of its helix angle (HA).¹¹⁰ This is defined as the angle the primary eigenvector makes with the local radial plane of the heart (Figure 4).¹¹⁶ HA evolves from roughly 60° in the subendocardium to −60° in the subepicardium and is highly conserved across mammalian species.^{29, 113, 116} In all mammals the cardiomyocytes in the midmyocardial are circumferential while those in the subendocardium and subepicardium are oriented obliquely (Figure 4).^{110, 114, 116} The term HA is in some ways a misnomer since the cardiomyocytes in the heart are arranged in a 3D syncytium and do not form mechanically continuous helices, such as DNA. Nevertheless it is extremely useful to integrate the primary eigenvectors in each voxel into streamlines or tracts that provide a comprehensive 3D representation of cardiac microstructure.^{29, 113, 116, 117} These tracts can be classified by the HA in each segment or by the tractographic propagation angle (PA). The PA describes the angle between adjacent segments in a tract and is generally < 4°/voxel (Figure 4).¹¹⁷ Microstructural disorder causes PA to increase and the metric has been used to detect areas of myocardial injury and electrical instability with a high degree of accuracy.¹¹⁷

Figure 4: Microstructure of the heart.

The microstructure of the rodent heart can be characterized with diffusion tensor MRI. (A) The arrangement of cardiomyocytes in the heart can be described by the angle, termed the helix angle (HA), they make with the local radial plane of the ventricle. The HA in the heart ranges from 60° in the subendocardium to −60° in the subepicardium. Reproduced with permission.¹¹⁶ (B) The orientation of myocytes in the heart is highly ordered and allows them to be integrated into virtual streamlines or tracts. The propagation angle (PA) describes the angle between adjacent segments in these tracts and is <4°/voxel in most of the LV. Reproduced with permission.¹¹⁷ (C, D) Cardiomyocyte orientation tracts in the lateral wall of the rat and human heart, respectively, color-coded by HA. Myocardial microstructure is highly conserved across species and is extremely similar in rodent and human hearts. Reproduced with permission.¹¹⁶ (E) Cardiomyocyte orientation tracts in the mouse heart color-coded by PA. The PA in the LV is increased at the right ventricular insertion points but is otherwise fairly homogeneous. (F) Late gadolinium enhancement image of a mouse with a myocardial infarct. (G) PA in the infarct increases substantially and a PA threshold of 4° allows the infarct to be accurately detected. (E-G) Reproduced with permission.¹¹⁷ (H-K) Rat heart imaged during diastole and systole. The HA maps during diastole and systole are similar but the myocardial sheetlets, defined by the plane of the secondary and tertiary eigenvectors, undergo a radial re-orientation in systole. Reproduced with permission.¹¹²

Cardiomyocyte contraction results in radial thickening of the myocardium, which is the primary eigenvector (major direction) of myocardial strain.¹⁰⁶ However, there are no radially oriented cardiomyocytes in the heart.^{110, 113} Rather, the radial yield of cardiomyocyte contraction is mediated via cardiomyocyte sheetlets, which consist of groups of 5–10 cardiomyocytes.¹¹⁸ The reorientation, extension and shearing of these sheets allows the heart to undergo radial strain during systole and diastole.¹¹⁸ The orientation of these sheets is given by the secondary eigenvector of the diffusion tensor and changes dynamically through the cardiac cycle.^{112, 119, 120} During diastole the sheets are oriented tangential to the surface of the heart and undergo a radial reorientation in systole (Figure 4).^{112, 119, 120} This can be quantified by following the orientation of the secondary eigenvector across the cardiac cycle. Because two mirror-image populations of sheets exist in the subendocardium and subepicardium the absolute value of the sheet angle (E2A) is used. In humans with dilated cardiomyopathy and HCM abnormalities in E2A have been detected in systole and diastole, respectively.¹²⁰ DTI of the heart in small animals provides a wealth of information but remains a highly specialized technique and requires dedicated sequences and gradient systems to be performed in vivo.²⁹

Imaging of Cell Death

In addition to LGE, manganese (Mn) based imaging has also been used to image cardiomyocyte viability.^{121, 122} Mn is paramagnetic and its uptake can be detected with MRI much like Gd. In the case of Mn, however, areas of viable myocardium take up the probe. Studies in rodents and large animals have shown that the addition of Mn imaging to LGE may allow areas of viable myocardium to be distinguished more accurately than LGE alone.^{121, 122} Molecular imaging approaches to image several forms of cell death including apoptosis, necrosis and autophagy have been developed and tested in the murine heart. Much of the initial work in apoptosis focused on the use of annexin-V to detect apoptosis with SPECT.¹²³ Small animal imaging of apoptosis has now been performed by attaching radiotracers, iron-oxide nanoparticles and Gd-labeled liposomes to annexin-V (Figure 5).^124–127

Figure 5: Molecular imaging of cell death.

(A) Detection of apoptosis in the murine heart in vivo with an annexin-decorated iron-oxide nanoparticle. Within a few hours of ischemia-reperfusion any unbound nanoparticles have washed out while phosphatidylserine-bound nanoparticles are retained and produce signal hypointensity on MRI. Reproduced with permission.¹²⁷ (B) Detection of necrosis in a murine infarct with Gd-TO, a DNA-binding gadolinium chelate. The probe cannot cross an intact cell membrane but in the presence of necrotic cardiomyocytes can access and bind to nuclear/cell free DNA. (C) The TO moiety on the probe both binds to nucleic acids and provides a fluorescent readout, allowing the co-localization of Gd-TO with DAPI to be assessed by microscopy. (B-C) Reproduced with permission.²⁴ (D) TO can be conjugated to a 40kD dextran to create an optical probe capable of detecting necrosis ex vivo in the murine heart. (E) Simultaneous injection of Dex-TO and a near-infrared annexin (AV-750) in mice with ischemia-reperfusion shows that areas of myocardium with low-moderate levels of annexin uptake do not lose cell integrity and may not fully execute the cell death cascade. (D-E) Reproduced with permission.¹²⁸ (F) The injection of rapamycin (RAP) either before or after the onset of reperfusion significantly reduces the portion of the area-at-risk that becomes positive for annexin-uptake. Reproduced with permission.¹³¹ (G) Detection of transplant rejection (arrow) with ⁹⁹Tc-duramycin. Reproduced with permission.¹²⁹

The pathophysiological significance of annexin-V uptake has been closely studied in small animal models and is nuanced. The vast majority of the area-at-risk (AAR) following ischemia-reperfusion injury shows some uptake of annexin-V (Figure 5).¹²⁸ However, only portions of the AAR with high levels of annexin uptake seem to fully execute the cell death cascade.¹²⁸ Low levels of annexin-V uptake may, therefore, be a marker of ischemic injury rather than cell death. In addition, rupture of the cardiomyocyte membrane due to necrosis will also result in the uptake of annexin-V by phosphatidylserine on the inner side of the cell membrane. A dual contrast approach in which annexin is used with a marker of cell membrane rupture, such as LGE or antimyosin imaging, has been implemented to address this.^{126, 127} Several other ligands (C2-synaptotagmin, duramycin) have been developed to bind to phosphatidylserine on the surface of apoptotic cells and have been imaged in small animal models (Figure 5).^{129, 130} Many of the concerns and limitations of annexin-based imaging apply equally to these ligands.

Vital fluorochromes, most notably thiazole orange (TO), have also been modified to support in vivo imaging of necrosis in small animals (Figure 5).²⁴ Gd-TO cannot cross an intact cell membrane and in the absence of necrosis is rapidly washed out of healthy myocardium. Access to nuclear DNA in ruptured cardiomycytes, or to cell-free DNA secreted into the interstitial space, results in Gd-TO binding and retention of the probe in the myocardium.²⁴ The imaging of autophagy in vivo is challenging although a cathepsin-activatable near infrared fluorochrome has been used to image autophagy noninvasively in the heart.¹³¹ The near infrared nature of the probe allows fluorescence tomography of the heart to be performed non-invasively in mice.^{131, 132} However, the sensitivity of this fluorochrome for autophagy is moderate. More recently, a second-generation autophagy detecting nanoparticle, with extremely high sensitivity and specificity, has been generated to address this limitation.

Myocardial Inflammation

The imaging of myocardial inflammation is frequently performed by detecting the uptake of [¹⁸F]-fluorodeoxyglocose (FDG) by inflammatory cells in the heart. However, this approach relies on the complete suppression of glucose uptake by cardiomyocytes, which is challenging. Consequently, numerous probes have been developed to image myocardial inflammation with greater specificity. Iron-oxide nanoparticles are avidly taken up by macrophages and other cells in the reticulo-endothelial system (Figure 6).¹³² In the preclinical context a fluorochrome can be easily conjugated to the surface of the iron-oxide nanoparticle, which allows its uptake to be imaged at the whole organ level by MRI and fluorescence tomography, and at the cellular level with flow cytometry and microscopy.¹³² Several other MRI-based approaches have been used to image the infiltration of inflammatory cells into the heart including the use of Gd and [¹⁹F]-containing liposomes,^{108, 133–135} and nanoparticles labeled with both [¹⁹F] and [⁸⁹Zr] (zirconium) (Figure 6).^{108, 133–136}

Figure 6: Imaging of myocardial inflammation and remodeling.

Macrophages infiltrating healing myocardial infarcts (arrows) in the murine heart can be imaged with a variety of nanoparticles, which they avidly take up. (A) Iron-oxide nanoparticles, which produce signal hypointensity on the MR images. Reproduced with permission.¹³² (B) Gadolinium-loaded liposomes, which increase the longitudinal relaxation rate (R1) in the vicinity of their uptake. Reproduced with permission.¹⁰⁸ (C) The signal from (¹⁹F)-loaded liposomes can be distinguished from the proton MRI signal allowing simultaneous (¹⁹F) and (¹H) MRI to be performed. Reproduced with permission.^{133, 134} (D) [⁶⁴Cu]-labeled macrin can be detected with micro-PET. (w = surgical wound in chest wall). Reproduced with permission.¹⁴¹ (E) Pre-labeling of myeloid cells with [⁸⁹Zr] and (¹⁹F) nanoparticles. After myocardial infarction (MI) there is a rapid egress of myeloid cells from the liver (li) and spleen (sp). [⁸⁹Zr] autoradiography shows a corresponding infiltration of these cells into the myocardium after MI or ischemia-reperfusion (IR). Reproduced with permission.¹³⁶ (F-G) Imaging of macrophage infarct infiltration with a myeloperoxidase (MPO) activatable gadolinium-chelate. Activation of the probe in wildtype mice produces signal hyperintensity in the infarct, which is absent in the homozygous knockout mouse. Reproduced with permission.¹³⁸ (H) MMP activation in mice after MI. The SPECT perfusion images (green) delineate the infarct zone, which shows significant uptake of the MMP-targeted probe (red). Reproduced with permission.¹⁴⁷ (I) In vivo microPET of CXCR4 distribution in a murine infarct with corresponding ex vivo autoradiography below. Reproduced with permission.¹⁴⁸ (J) Gadolinium chelate targeted to elastin is taken up in the anterior wall and apex of a healing murine infarct. Reproduced with permission.¹⁴⁹ (K-L) Gadolinium chelate targeted to type 1 collagen accurately detects areas of fibrosis in a healed murine infarct. Reproduced with permission.²⁵

The degradative enzymes released by inflammatory cells can also be directly imaged. Cathepsin-activatable fluorochromes, for instance, are robustly activated by neutrophils and macrophages.¹³⁷ Another approach involves the detection of myeloperoxidase (MPO) secreted by inflammatory cells with a MPO-activatable Gd chelate (Figure 6).¹³⁸ Several small molecule PET probes have been developed to image myocardial inflammation but have not proven superior to [¹⁸F]-FDG.¹³⁹ However, a novel PET tracer, [¹⁸F-DHMT], was able to image the generation of reactive oxygen species in rodents treated with doxorubicin.¹⁴⁰ In addition, the radiolabeled nanoparticle, [⁶⁴Cu]-Macrin, appears to be highly promising. Preclinical use of the probe in mice showed that it could accurately detect myocardial inflammation (Figure 6).¹⁴¹ Ischemic injury also results in endothelial inflammation, which can be detected in mice with PET, MRI and targeted microbubbles.^{3, 78}

Myocardial Healing, Angiogenesis and Fibrosis

Molecular imaging of angiogenesis using ligands to the α_vβ₃-integrin has been successfully performed in small animal models.^{142, 143} The uptake of these agents, however, is context specific. The RGD peptide, for instance, has also been reported to bind to α_vβ₃-integrin on the surface of myofibroblasts.¹⁴⁴ The detection of fibroblast/myofibroblast activation can also be performed using radiolabeled fibroblast activation protein inhibitors (FAPI).¹⁴⁵ Optical and SPECT-based approaches have been developed to image matrix metalloproteinases (MMPs) in the remodeling heart. The rate constant of MMP activation in the heart, however, may not be sufficient to support robust detection of the optical construct in vivo.¹⁴⁶ In contrast, the SPECT based approach has been used with significant success (Figure 6).¹⁴⁷ The probe consists of a chemical inhibitor to MMPs (2 and 9) conjugated to either [⁹⁹Tc] or [¹¹¹In] and has been used in models of myocardial infarction and anthracycline cardiotoxicity.^{140, 147} The role of CXCR4 in the remodeling heart is multifaceted and its imaging with PET in mice has provided valuable insights (Figure 6).¹⁴⁸

Molecular imaging agents have been developed to directly image collagen and elastin during ventricular remodeling (Figure 6).^{25, 149} A peptide targeting type 1 collagen has been conjugated to Gd and used to specifically image fibrosis in infarcted mice.²⁵ The highly reactive allysine groups on collagen, which cross link it and make it more resistant to breakdown, can be detected with an aldehyde (allysine) binding Gd chelate.¹⁵⁰ This probe may be a marker of active fibrogenesis and provide a complementary readout to other collagen-targeted probes.

Myocardial Regeneration

Human studies involving the injection of bone-marrow derived and mesenchymal stem cells have been disappointing. Interestingly studies in small animal models using advanced imaging techniques did not show encouraging results. In one such study, serial in vivo diffusion tensor MRI tractography (DTI-tractography) was used to evaluate the impact of bone marrow derived mononuclear cells injected directly into mice three weeks after ischemia-reperfusion injury (Figure 7).²⁹ Serial imaging before and after cell injection showed no restoration of myocardial anisotropy and microstructure in >90% of injected mice. Moreover, in some of the mice cell injection actually resulted in a loss of coherent microstructure (Figure 7).²⁹

Figure 7. Diffusion tensor MRI and reporter imaging of cell therapy.

(A) Islets of surviving cardiomyocytes in a rat infarct create a network of residual tracts, which must be differentiated from regenerated myocardial tracts. Reproduced with permission.¹¹³ (B-C) DTI-tractography of the murine heart in vivo. Tracts intersecting a region-of-interest (red rectangle, inset) in the lateral wall are shown from a lateral and basal perspective, respectively. (D-I) Serial in vivo DTI tractography of mice with ischemic injury injected with bone marrow mononuclear cells. (D) In the majority of mice no response to cell injection was seen and a severe loss of tracts in the apical half of the ventricle persisted after injection. (E-G) However, in some mice cell injection was associated with the loss of tracts after injection. (E) Before injection coherent tracts (arrows) are seen in the anterior and inferolateral walls. (F) Imaging of the same mouse 7 days after injection reveals the loss of these tracts. (G) The transmural slope of helix angle in the inferolateral wall is normal before injection (blue curve) and severely perturbed (black curve) post-injection. (H) Cell injection produced an increase in mean diffusivity and a decrease in fractional anisotropy, which are signatures of injury or inflammation, rather than regeneration. (I) In >90% of the mice studied cell injection produced a negative (neutral or deleterious) response. (B-I) Reproduced with permission.²⁹ (J) Serial bioluminescence imaging shows that bone marrow mononuclear cells injected into infarcted mice do not survive, with a vast reduction in signal from day 4 to day 21. Reproduced with permission.¹⁵³ (K) Conversely embryonic stem cells, shown at weeks 3 and 4 by bioluminescence and PET reporter imaging, survive and grow (arrows) but distribute beyond the heart and form teratomas. Reproduced with permission.¹⁵⁴

The labeling of stem cells with iron-oxide probes can be extremely useful in guiding their injection.^{151, 152} However, the persistence of the iron-oxide label in the myocardium does not equate with cell survival.¹⁵² Reporter techniques, in which the imaging readout relies on the active synthesis of a protein/enzyme by the cell, address this limitation (Figure 7).^{153, 154} Bioluminescent and PET reporter probes have been used to show that bone marrow mononuclear cells injected into infarcted myocardium survive extremely poorly,¹⁵³ and conversely that embryonic cells survive extremely well but form teratomas (Figure 7).¹⁵⁴

The Right Ventricle and Pulmonary Circulation

Right ventricular anatomy and function is difficult to assess using murine echocardiography due to its complex geometrical shape and small size. Although investigators have attempted to measure RV size, EF, and wall thickness using echocardiographic parasternal views,^155–157 these parameters are obtained with more precision using MRI. The tricuspid annular velocity (S’) and tricuspid annular plane systolic excursion (TAPSE) can also be measured from the apical 4 chamber view and were decreased in a model of hypoxia and vascular endothelial growth factor inhibitor, and in heart failure.^{155, 156} Due to the paucity of tricuspid regurgitation in rodents, the measurement of pulmonary artery or right ventricular pressure using the maximum velocity of the tricuspid regurgitation jet is not feasible. Pulsed wave Doppler, however, has shown inverse correlations between systolic pulmonary artery pressure, measured by a pressure catheter in mice, and both the pulmonary acceleration time (PAT) and PAT/ejection time.¹⁵⁸ These echocardiographic indices have been used widely in models of pulmonary hypertension.^{155, 159, 160} While most MRI based approaches in mice have focused on the imaging of the LV, many of the advanced MRI techniques, described above, can also be performed in the murine RV. However, others are limited by the thin nature of the RV wall. At the present time MRI of the murine RV is limited in most centers to measurements of RV volumes and function, including RV strain by feature tracking, with cine MRI.¹⁶¹

Future Prospects and Conclusion

Advances in biomedical imaging continue to occur at a rapid pace. The development of integrated multimodality platforms, such a PET-MR and optoacoustic systems, is opening up new possibilities in both the clinical and research domains. Many of the recently developed molecular imaging agents also incorporate a therapeutic capability (theranostic agents), which is ideally suited to testing in small animal models. Machine learning is playing an increasing role in biomedical imaging and will likely impact all areas of small animal imaging from acquisition to reconstruction and analysis. New approaches will also need to be developed to better integrate biomedical imaging data with genomic/metabolomic/proteomic data. Major opportunities for innovation will continue to exist at the interface of molecular biology and advanced biomedical imaging and will require a high degree of collaboration between the two communities to be fully realized.

Non-Standard Abbreviations and Acronyms

T1

Longitudinal Magnetic Relaxation Time

T1-Rho

Magnetic Relaxation Time with Spin-Locking

UTE

Ultrashort Echo Time

CEST

Chemical Exchange Saturation Transfer

BOLD

Blood Oxygen Level Dependent

ASL

Arterial Spin Labeling

DENSE

Displacement Encoded Stimulated Echo

DTI

Diffusion Tensor Imaging

MD

Mean Diffusivity

FA

Fractional Anisotropy

HA

Helix Angle

PA

Propagation Angle

E2A

Absolute Value of Sheet Angle

TO

Thiazole Orange

MPO

Myeloperoxidase

MMP

Matrix Metalloproteinase

FAPI

Fibroblast Activation Protein Inhibitor