**2. MRI of the mouse: Challenges for cardiac image-based mouse phenotyping**

#### **2.1. The mouse as a research model**

344 Practical Applications in Biomedical Engineering

science and translational research.

frequency reserve), for comparative studies.

image-based phenotyping.

balancing anesthetic, Nitrous Oxide.

Despite the existence of a plethora of cardiac functional techniques for characterization of mechanical structure, function and dysfunction, a parallel need exists for development of invasive and non-invasive tools and techniques to describe the left ventricular (LV) tissue material properties as these relate to the: (a) mechanical pumping function of the LV; (b) myocardial oxygen demand defining myocyte metabolic status; (c) coronary blood flow and its auto-regulation; (d) arhythmogenic risk; (e) cell-signaling pathways responsible for growth and remodeling during development and disease. Reinforcing basic physiology work, invasive catheterization experiments [Georgakopoulos 1998] have also allowed determination of inotropic and lusitropic cardiac status, while Magnetic Resonance Imaging (MRI) experimentation, methodologies and technology advances have facilitated migration of such work to a non-invasive imaging platform, with tremendous potential for future basic

Specifically, advances in MRI techniques (myocardial spin tagging [Zerhouni 1988, Axel 1989], DENSE [Aletras 1999], and harmonic phase imaging [Osman 1999]) have been introduced over the last decades to quantify cardiac function, allowing myocardial tracking, motion and strain quantification in normal and genetically engineered mice [Rockman 1991, Franco 1999, Brede 2001, Yang 2002, Engel 2004, Wilding 2005]. Critical to such work has been the validation of the underlying hypothesis of morphological and functional scaling from mouse to human (through consideration of global cardiac function, circulatory control, blood flow distribution, Ca2+ storage and cycling, myosin light chain distribution, and force

This chapter provides an overview of the major physiological issues and challenges for mouse MR imaging and discusses the most recent and major advances in conventional and new cardiac Magnetic Resonance imaging strategies, that ultimately allow quantification of motion, global, and regional cardiac function, strain, and elasticity, characterizing inotropic and lusitropic contractile function and dysfunction in humans and transgenic mice for

Specifically, this chapter attempts a detailed reference to the mouse as a research model, focusing on its genetic background and homology with the human genome and to the developmental and morphological differences between mouse and man, thus addressing cellular and global organ similarities and differences. As a basic determinant of structure, cardiac functional differences are associated, justified by carefully-controlled indices that determine integrative physiological control and functional activities, including metabolism, perfusion, angiogenetic, collateral flow, and coronary reserve. The importance and impact of anesthesia for image-based phenotyping in patho-physiological status is addressed with brief references to the possible mechanisms and cellular and sub-cellular target sites of anesthesia action. The section is complemented with recent findings on heart rate variability (as a result of the widely used inhalational anesthesia use) under optimal anesthesia conditions using isoflurane and long term physiological stability elicited from the use of the

A historical overview of the evolution of mouse cardiac MRI is also attempted (in direct correlation with the evolution and progress of human clinical cardiac MRI, radiofrequency The mouse emerged as an attractive animal research model following the rapid advances in experimental molecular biology techniques that allowed targeted mutagenesis in single genes [Capecchi 1989], in addition to the tremendous success for extraction, manipulation, and use of embryonic stem cells [Koller 1989]. Practical and ethical advantages were also associated with mice, such as their stable genetic lines, immune system, short gestation periods, low cost, and ease of use. Initial research strides were supported by US National initiatives including the Human and Mouse Genome projects administered through the National Institutes of Health (NIH) for cloning and mapping the entire human and mouse genomes [Collins 2003, Gregory 2002], efforts that were successfully completed in 2003.

Todate, multiple thousands of different knockout mice have been constructed (by individual Institutions, Laboratories, or National or International Consortia), most of which affect cardiovascular function. As attempts to match genotype to phenotype continue, an increasing number of knockout or loss-of-function mice is expected to be generated.

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

[Webb 1996]. While availability of data on the cardiomyofiber structure is scarce, a prior publication [McLean 1992] indicates some fiber orientation differences, primarily on the

**Figure 2.** Relative size scales of the human, adult, and mouse embryo [Courtesy of the Duke Center for In Vivo Microscopy, an NIH/NIBIB National Biomedical Technology Resource Center (P41 EB015897)].

Prominent differences also exist for the conduction system and coronary anatomy. In contrast to the human where the sino-atrial (SA) node is distinctly located on the right atrium, in the mouse it lies on the superior vena cava (SVC), at the juncture of the SVC with the right atrium. Coronary anatomy exhibits branching of a septal coronary artery from the left coronary system supplying the left anterior ventricular wall, while the right coronary artery branches into a right coronary and a circumflex vessel supplying the posterior

Cardiomyocytes are reported to have lengths of approximately 80-100 μm along their major axis, and lengths of 10-20 μm along their minor axis [Doevendans 1998, Smaill 1991). Eightyfive percent (85%) of the total number of cardiac cells (amounting to 7-10 million approximately) are interstitial while the remaining 15% that represent cardiomyocytes occupy 90% of the available total ventricular tissue volume [Doevendans 1998]. Also important is an increased capillary to fiber ratio (of the order of 1.4) in the mouse, a direct necessity for the higher energetic and metabolic demands compared to rats, canine, or humans, (with a fiber ratio of the order of 1-1.1) [Sabbah 1995, Przylenk 1983, Olivetti 1989,

middle layer of the myocardium.

ventricular wall [Doevendans 1998].

*Cellular size and content* 

The Current Status, Challenges, and Future Perspectives 347

However, despite the usefulness, practicality, and low costs associated with the study of the mouse, important genetic, developmental, morphological, and physiological differences exist between mouse and man [Doevendans 1998, Schaper 1998]. The section that follows addresses succinctly common and distinctly differing features and functional differences in mouse and man and discusses the mechanisms of anesthesia effects on the mouse physiology and cardiac contractile machinery.

#### *2.1.1. Genetic background and gene homology*

Through various structural genomics attempts to map and compare human and mouse genomes [Schaper 1998, Gregory 2002], an overall homology of more than 85% and an identity of more than 80% were recorded [Schaper 1998], suggestive of the increased conservation during differentiation of the two species. Based on such findings, it is shown that the homeobox (hox) genes (transcription factors responsible for development), other transcription factors that bind to promoters encoding acute phase genes, and heat shock proteins, exhibit increased identity with humans. Based on Schaper et al. [Schaper 1998], however, structural genomics shows that (except primates) all other species (including mouse, rat, horse, bovine, etc) are equidistant from man, and evidence is thus inconclusive for the choice of the best species (based only on genetic homology and protein identity) for assessment of cardiovascular function and dysfunction.

#### *2.1.2. Developmental and morphological differences between mouse and human*

#### *Cardiac dimensions*

Indicatively, the human heart weighs around 250-300 g (left and right ventricular [RV, LV] and atrial chambers), and has an intrinsic rhythm of around 60-70 beats per minute (bpm), while the murine heart has a weight of 0.2-0.3 g and beats at a rate of 600-700 bpm. Additionally, the most evident external morphological difference between the adult mouse heart and the human heart is its shape and size (Figure 2). Developmental differences also exist; human development starts during the 3rd week of gestation and lasts approximately 7 months to complete, while mouse cardiac development spans a little more than 2 weeks. At human birth, most of the cardiac organ development has been completed while in the infant mouse cardiac development may still be in progress.

#### *Cardiac anatomy*

No major differences exist in the outer morphology, ventricular structure, or valves. Noted differences exist, however, in atrial and venous parts. Specifically, the left superior caval vein drains directly into the right atrium in the mouse, whereas the pulmonary vein has an opening in the left atrium [Doevendans 1998], and a secondary atrial septum is lacking [Webb 1996]. While availability of data on the cardiomyofiber structure is scarce, a prior publication [McLean 1992] indicates some fiber orientation differences, primarily on the middle layer of the myocardium.

**Figure 2.** Relative size scales of the human, adult, and mouse embryo [Courtesy of the Duke Center for In Vivo Microscopy, an NIH/NIBIB National Biomedical Technology Resource Center (P41 EB015897)].

Prominent differences also exist for the conduction system and coronary anatomy. In contrast to the human where the sino-atrial (SA) node is distinctly located on the right atrium, in the mouse it lies on the superior vena cava (SVC), at the juncture of the SVC with the right atrium. Coronary anatomy exhibits branching of a septal coronary artery from the left coronary system supplying the left anterior ventricular wall, while the right coronary artery branches into a right coronary and a circumflex vessel supplying the posterior ventricular wall [Doevendans 1998].

#### *Cellular size and content*

346 Practical Applications in Biomedical Engineering

physiology and cardiac contractile machinery.

*2.1.1. Genetic background and gene homology* 

assessment of cardiovascular function and dysfunction.

mouse cardiac development may still be in progress.

*Cardiac dimensions* 

*Cardiac anatomy* 

*2.1.2. Developmental and morphological differences between mouse and human* 

Indicatively, the human heart weighs around 250-300 g (left and right ventricular [RV, LV] and atrial chambers), and has an intrinsic rhythm of around 60-70 beats per minute (bpm), while the murine heart has a weight of 0.2-0.3 g and beats at a rate of 600-700 bpm. Additionally, the most evident external morphological difference between the adult mouse heart and the human heart is its shape and size (Figure 2). Developmental differences also exist; human development starts during the 3rd week of gestation and lasts approximately 7 months to complete, while mouse cardiac development spans a little more than 2 weeks. At human birth, most of the cardiac organ development has been completed while in the infant

No major differences exist in the outer morphology, ventricular structure, or valves. Noted differences exist, however, in atrial and venous parts. Specifically, the left superior caval vein drains directly into the right atrium in the mouse, whereas the pulmonary vein has an opening in the left atrium [Doevendans 1998], and a secondary atrial septum is lacking

Todate, multiple thousands of different knockout mice have been constructed (by individual Institutions, Laboratories, or National or International Consortia), most of which affect cardiovascular function. As attempts to match genotype to phenotype continue, an

However, despite the usefulness, practicality, and low costs associated with the study of the mouse, important genetic, developmental, morphological, and physiological differences exist between mouse and man [Doevendans 1998, Schaper 1998]. The section that follows addresses succinctly common and distinctly differing features and functional differences in mouse and man and discusses the mechanisms of anesthesia effects on the mouse

Through various structural genomics attempts to map and compare human and mouse genomes [Schaper 1998, Gregory 2002], an overall homology of more than 85% and an identity of more than 80% were recorded [Schaper 1998], suggestive of the increased conservation during differentiation of the two species. Based on such findings, it is shown that the homeobox (hox) genes (transcription factors responsible for development), other transcription factors that bind to promoters encoding acute phase genes, and heat shock proteins, exhibit increased identity with humans. Based on Schaper et al. [Schaper 1998], however, structural genomics shows that (except primates) all other species (including mouse, rat, horse, bovine, etc) are equidistant from man, and evidence is thus inconclusive for the choice of the best species (based only on genetic homology and protein identity) for

increasing number of knockout or loss-of-function mice is expected to be generated.

Cardiomyocytes are reported to have lengths of approximately 80-100 μm along their major axis, and lengths of 10-20 μm along their minor axis [Doevendans 1998, Smaill 1991). Eightyfive percent (85%) of the total number of cardiac cells (amounting to 7-10 million approximately) are interstitial while the remaining 15% that represent cardiomyocytes occupy 90% of the available total ventricular tissue volume [Doevendans 1998]. Also important is an increased capillary to fiber ratio (of the order of 1.4) in the mouse, a direct necessity for the higher energetic and metabolic demands compared to rats, canine, or humans, (with a fiber ratio of the order of 1-1.1) [Sabbah 1995, Przylenk 1983, Olivetti 1989, Doevendans 1998]. The duration of the cell cycle (once G1 is entered) is reported to be almost the same in mouse and man [Schaper 1998].

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

elastance (ES), and the end-systolic pressure volume relationship) and lusitropic (minimum developed ventricular pressure rates - dP/dtmin, Weiss and Glantz relaxation constants, and the end-diastolic pressure-volume relationship), left ventricular contractile indices [Constantinides\_ABME 2011, Doevendans 1998]. Recent MRI findings also support similar torsional patterns, as reported by normalized (to left ventricular lengths) twist and torsional angles [Henson 2000, Zhou 2003, Liu 2006, Zhong 2010], however, limited and less steep responses in the force-frequency relationship-response of the mouse compared to humans [Stull 2002] refers to altered calcium kinetics [Stuyvers 1997] and handling under stress.

In consideration of the various morphological and anatomical comparisons listed, the murine cardiovascular system (under normal physiological conditions) resembles relatively closely evoked responses in larger mammals. As mentioned, numerous cardiovascular indices in mice match corresponding values in rats and humans, predicted by allometric

At the integrative physiological level, the major blood pressure regulating systems, namely the baroreceptor reflex [Ma 2002] and the renin-angiotensin system (regulating electrolyte

However differences do exist, including the increased sensitivity of blood pressure to anesthesia and temperature, especially under conditions of stress [Barbee 1992, Rosenblum 1997, Sarin 1990, Janssen 2004, Constantinides\_ILAR 2011], higher resting cardiac sympathetic

Overall, under physiological conditions, evidence supports that bioenergetically and hemodynamically the mouse scales linearly with larger mammals and humans [Dobson 1995, Nielsen 1958, Phillips 2012], exhibiting a similar maximal aerobic capacity across species [Phillips 2012]. Preliminary evidence for allometric scaling to heart size in mechanical kinematic performance has also been presented [Popovic 2005, Zhong 2010]

Given all considerations, extrapolations of inferences from mice to man is appropriate under physiological conditions, however, similar attempts in pathological models or states may be

Even if advances in telemetric techniques have allowed the pursuit of mouse studies under conscious conditions to a large extent, most of the research (terminal, invasive, or noninvasive imaging) studies utilize anesthetics. Anesthetics, however, are known to cause severe cardio-depression [Hart 2001] with adverse physiological effects on hormonal release, centrally to the heart and peripherally to the vasculature [Price 1980, Ohnishi 1974] at the cellular level, affecting calcium entry through L-type Ca2+ channels, the calcium

balance) [Cholewa 2001] seem to resemble closely those in larger mammals.

*Allometric scaling of function, energetics, and metabolism* 

**2.2. The impact of anesthesia for mouse studies** 

tone [Janssen 2000, Janssen 2002], neurotransmitter release [Ehmke 2003], and others.

supported by preliminary comparative mouse-human results in this present work.

*Integrative physiological control* 

scaling laws.

indeed risky.

The Current Status, Challenges, and Future Perspectives 349

#### *2.1.3. Functional physiological differences*

#### *Cardiovascular metabolism*

Energetic requirements in biological organisms are often assessed by their oxidative capacity, myosin ATPase and SERCA activities [Blank 1989] with a commensurate increased demand in small animals, such as rodents. In compensation of such increased energetic demand, an increased volume density of mitochondria is reported by Doevendans et al. [Doevendans 1998] in mice (37.9%) compared to humans (25.3%), with similar myofibril density (approximately 50%). However, recent findings by Phillips et al. [Phillips 2012] argue in favor of a constant density of mitochondria across species (approximately 21%), yet an increased enzymatic activity of mitochondrial enzymes in mice and a smaller dynamic range of metabolism is reported. Also of interest are the relative patterns of cardiac myosin chain isoforms (αα, ββ, αβ), a mouse αα-isoform compared to a human-ββ isoform dominance [Sheuer 1979], and a developmental switch in mice compared to humans, resulting in higher myosin ATPase activity. Noteworthy and more important is the increased basal activity of metabolic enzymes (Complex V) [Phillips 2012] and the smaller energetic reserve in mice compared to humans [Blank 1989, Phillips 2012], raising concerns for the appropriateness of the mouse as a proper model for comparative pharmacologic (doputamine, dipyridamole) stress studies in ischemia-infarction-reperfusion models, compared to human disease [van Rugge 1992, Wiesmann\_Circ\_Res 2001, Williams 2001].

#### *Perfusion, angiogenesis-collateral flow, coronary reserve*

Based on the extensive number of literature publications that focused on the elucidation of the basic principles of cardiovascular physiology (during the early and latter half of the 20th century), including the autoregulation of blood flow in rats and canine [Feng 2001, Sandgaard 2002], important differences exist in mouse and man with regard to perfusion, capillary density [Stoker 1982, Rakusan 1994], total blood volume (2.5 ml in the mouse compared to 400-500 ml in humans), relative distribution of blood flow in the capillary bed and redistribution capability subject to stimuli (temperature, anesthesia) [Barbee 1992, Rosenblum 1997, Sarin 1990, Constantinides 2011], angiogenetic capacity and formation of collateral vessels, as well as the resting coronary reserve, as factors that primarily project to ischemia-reperfusion studies (or other cardiovascular pathology models) and comparative analyses between pigs, canine, and man.

#### *In vivo cardiac function*

While direct comparative studies between mouse and man are still few, nevertheless published results favor similar hemodynamic and global cardiac functional indices, including blood pressure [Janssen 2002, Constantinides, 2011], inotropic (ejection fraction [EF], maximum developed ventricular pressure rates - dP/dtmax, stroke work (SW), preload adjusted maximum power (PAMP), Preload recruitable stroke work (PRSW), end-systolic elastance (ES), and the end-systolic pressure volume relationship) and lusitropic (minimum developed ventricular pressure rates - dP/dtmin, Weiss and Glantz relaxation constants, and the end-diastolic pressure-volume relationship), left ventricular contractile indices [Constantinides\_ABME 2011, Doevendans 1998]. Recent MRI findings also support similar torsional patterns, as reported by normalized (to left ventricular lengths) twist and torsional angles [Henson 2000, Zhou 2003, Liu 2006, Zhong 2010], however, limited and less steep responses in the force-frequency relationship-response of the mouse compared to humans [Stull 2002] refers to altered calcium kinetics [Stuyvers 1997] and handling under stress.

#### *Integrative physiological control*

348 Practical Applications in Biomedical Engineering

*Cardiovascular metabolism* 

almost the same in mouse and man [Schaper 1998].

*Perfusion, angiogenesis-collateral flow, coronary reserve* 

analyses between pigs, canine, and man.

*In vivo cardiac function* 

*2.1.3. Functional physiological differences* 

Doevendans 1998]. The duration of the cell cycle (once G1 is entered) is reported to be

Energetic requirements in biological organisms are often assessed by their oxidative capacity, myosin ATPase and SERCA activities [Blank 1989] with a commensurate increased demand in small animals, such as rodents. In compensation of such increased energetic demand, an increased volume density of mitochondria is reported by Doevendans et al. [Doevendans 1998] in mice (37.9%) compared to humans (25.3%), with similar myofibril density (approximately 50%). However, recent findings by Phillips et al. [Phillips 2012] argue in favor of a constant density of mitochondria across species (approximately 21%), yet an increased enzymatic activity of mitochondrial enzymes in mice and a smaller dynamic range of metabolism is reported. Also of interest are the relative patterns of cardiac myosin chain isoforms (αα, ββ, αβ), a mouse αα-isoform compared to a human-ββ isoform dominance [Sheuer 1979], and a developmental switch in mice compared to humans, resulting in higher myosin ATPase activity. Noteworthy and more important is the increased basal activity of metabolic enzymes (Complex V) [Phillips 2012] and the smaller energetic reserve in mice compared to humans [Blank 1989, Phillips 2012], raising concerns for the appropriateness of the mouse as a proper model for comparative pharmacologic (doputamine, dipyridamole) stress studies in ischemia-infarction-reperfusion models, compared to human disease [van Rugge 1992, Wiesmann\_Circ\_Res 2001, Williams 2001].

Based on the extensive number of literature publications that focused on the elucidation of the basic principles of cardiovascular physiology (during the early and latter half of the 20th century), including the autoregulation of blood flow in rats and canine [Feng 2001, Sandgaard 2002], important differences exist in mouse and man with regard to perfusion, capillary density [Stoker 1982, Rakusan 1994], total blood volume (2.5 ml in the mouse compared to 400-500 ml in humans), relative distribution of blood flow in the capillary bed and redistribution capability subject to stimuli (temperature, anesthesia) [Barbee 1992, Rosenblum 1997, Sarin 1990, Constantinides 2011], angiogenetic capacity and formation of collateral vessels, as well as the resting coronary reserve, as factors that primarily project to ischemia-reperfusion studies (or other cardiovascular pathology models) and comparative

While direct comparative studies between mouse and man are still few, nevertheless published results favor similar hemodynamic and global cardiac functional indices, including blood pressure [Janssen 2002, Constantinides, 2011], inotropic (ejection fraction [EF], maximum developed ventricular pressure rates - dP/dtmax, stroke work (SW), preload adjusted maximum power (PAMP), Preload recruitable stroke work (PRSW), end-systolic In consideration of the various morphological and anatomical comparisons listed, the murine cardiovascular system (under normal physiological conditions) resembles relatively closely evoked responses in larger mammals. As mentioned, numerous cardiovascular indices in mice match corresponding values in rats and humans, predicted by allometric scaling laws.

At the integrative physiological level, the major blood pressure regulating systems, namely the baroreceptor reflex [Ma 2002] and the renin-angiotensin system (regulating electrolyte balance) [Cholewa 2001] seem to resemble closely those in larger mammals.

However differences do exist, including the increased sensitivity of blood pressure to anesthesia and temperature, especially under conditions of stress [Barbee 1992, Rosenblum 1997, Sarin 1990, Janssen 2004, Constantinides\_ILAR 2011], higher resting cardiac sympathetic tone [Janssen 2000, Janssen 2002], neurotransmitter release [Ehmke 2003], and others.

#### *Allometric scaling of function, energetics, and metabolism*

Overall, under physiological conditions, evidence supports that bioenergetically and hemodynamically the mouse scales linearly with larger mammals and humans [Dobson 1995, Nielsen 1958, Phillips 2012], exhibiting a similar maximal aerobic capacity across species [Phillips 2012]. Preliminary evidence for allometric scaling to heart size in mechanical kinematic performance has also been presented [Popovic 2005, Zhong 2010] supported by preliminary comparative mouse-human results in this present work.

Given all considerations, extrapolations of inferences from mice to man is appropriate under physiological conditions, however, similar attempts in pathological models or states may be indeed risky.

#### **2.2. The impact of anesthesia for mouse studies**

Even if advances in telemetric techniques have allowed the pursuit of mouse studies under conscious conditions to a large extent, most of the research (terminal, invasive, or noninvasive imaging) studies utilize anesthetics. Anesthetics, however, are known to cause severe cardio-depression [Hart 2001] with adverse physiological effects on hormonal release, centrally to the heart and peripherally to the vasculature [Price 1980, Ohnishi 1974] at the cellular level, affecting calcium entry through L-type Ca2+ channels, the calcium

binding sensitivity of the contractile proteins to calcium, on conduction and excitability, and possibly on other sarcoplasmic reticular sites [Price 1980]. Also prominent are effects on the central and peripheral nervous system, but most notable are effects on the metabolism (through mitochondrial vasomotor changes in coronary circulation and perfusion, vasodilation, and blood flow changes in the microvasculature [Kober 2005], possibly synergistic to Nitric Oxide), and the decoupling of oxidative phosphorylation manifested through myocardial oxygen consumption changes.

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

**Figure 3.** Schematic representation of a typical myocyte with indicative (highlighted) target sites of

effects, especially for prolonged imaging studies often exceeding 45 minutes in duration. Apart from physiological indices of hormonal release, respiration and metabolism, carefully controlled cardiac indices (such as heart rate, ejection fraction, arterial pressure, and heart variability) ensure proper conditions of study of the cardiovascular system avoiding detrimental hypotension-induced blood volume changes, metabolic and contractile downregulation, and arrhythmogenicity. Heart variability (HRV) analyses have also been applied for phenotypic screening of transgenic mice, study of heart rhythm mediators through signaling pathways as well as the effects of pharmacologic intervention on intrinsic heart rhythm and arrythmogenesis [Thireau 1997, Bernston 1997, Gehrmann 2000]. Standardization of HRV analyses tools for mice have, however, been limited due to the numerous data acquisition types and analysis techniques employed. Most importantly, variability of HRV is dependent on the type of anesthesia being used, that may also be further influenced by anesthetic balancing agents such as nitrous oxide (N2O), medical air, and oxygen. Interpretation of HRV results has also been difficult [Hoit 2004], due to their dependence on a number of factors including aging, posture, circadian variability, and the duration of the sampling periods used, with recent evidence supporting the fact that the primary contribution of autonomic activity in the mouse is due to the sympathetic tone, in contrast to the parasympathetic vagus contribution [Janssen 2002, Janssen 2000,

inhalational anesthesia.

Constantinides\_IEEE 2010].

The Current Status, Challenges, and Future Perspectives 351
