Diffusion Tensor Cardiovascular Magnetic Resonance Imaging

JACC: CARDIOVASCULAR IMAGING

VOL.

-, NO. -, 2019

ª 2019 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION PUBLISHED BY ELSEVIER

STATE-OF-THE-ART PAPER

Diffusion Tensor Cardiovascular Magnetic Resonance Imaging A Clinical Perspective Zohya Khalique, MD,a,b Pedro F. Ferreira, PHD,a,b Andrew D. Scott, PHD,a,b Sonia Nielles-Vallespin, PHD,a,b David N. Firmin, PHD,a,b Dudley J. Pennell, MDa,b

ABSTRACT Imaging the heart is central to cardiac phenotyping, but in clinical practice, this has been restricted to macroscopic interrogation. Diffusion tensor cardiovascular magnetic resonance (DT-CMR) is a novel, noninvasive technique that is beginning to unlock details of this microstructure in humans in vivo. DT-CMR demonstrates the helical cardiomyocyte arrangement that drives rotation and torsion. Sheetlets (functional units of cardiomyocytes, separated by shear layers) have been shown to reorientate between diastole and systole, revealing how microstructural function facilitates cardiac thickening. Measures of tissue diffusion can also be made: fractional anisotropy (a measure of myocyte organization) and mean diffusivity (a measure of myocyte packing). Abnormal myocyte orientation and sheetlet function has been demonstrated in congenital heart disease, cardiomyopathy, and after myocardial infarction. It is too early to predict the clinical importance of DT-CMR, but such unique in vivo information will likely prove valuable in early diagnosis and risk prediction of cardiac dysfunction and arrhythmias. (J Am Coll Cardiol Img 2019;-:-–-) © 2019 by the American College of Cardiology Foundation.

M

yocardial phenotyping is crucial to under-

derangements in cardiomyopathy, myocardial infarc-

stand the healthy heart and changes in

tion (MI), and congenital heart disease. Hence, there

disease. It offers insight into the cardiac

is increasing interest in its potential clinical useful-

structure
function relationship and is key to diag-

ness and the new insight it brings to the pathophysi-

nosis, prognosis, and determining therapeutic inter-

ology

ventions.

current

describes the cardiac microstructure, the DT-CMR

imaging techniques is limited to a scale of millime-

technique, clinical studies performed so far, and the

ters, and it is not possible to assess the microstructure

potential future role of DT-CMR.

However,

phenotyping

with

of

cardiac

disease.

This

clinical

review

of the myocardium without resorting to biopsy. Diffusion tensor cardiovascular magnetic resonance (DT-

THE CARDIAC MICROSTRUCTURE

CMR) is a novel tool that provides noninvasive in vivo microstructural assessment, offering informa-

CARDIOMYOCYTES. Cardiomyocytes measure approx-

tion at the scale of cardiomyocyte organization

imately 150 by 20 mm and intercalate with several

(Central Illustration). Preliminary studies have shown

other cardiomyocytes (1). Although their branching

DT-CMR

nature means that cardiomyocytes do not strictly

can

identify

unknown

microstructural

From the aCMR Unit, Royal Brompton Hospital, London, United Kingdom; and the bNational Heart and Lung Institute, Imperial College, London, United Kingdom. Dr. Scott has received an equipment grant from NVIDIA. Dr. Firmin and Prof. Pennell have received research support from Siemens. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Manuscript received November 6, 2018; revised manuscript received July 9, 2019, accepted July 11, 2019.

ISSN 1936-878X/$36.00

https://doi.org/10.1016/j.jcmg.2019.07.016

2

Khalique et al.

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019 - 2019:-–-

DT-CMR Clinical Perspective

ABBREVIATIONS

align in fibers, there is a directional grain to

arrangement in systole to facilitate LV wall thick-

AND ACRONYMS

their appearance (Figure 1). This has led to

ening (4,7). However, until the advent of DT-CMR,

variable terminology, when some use “fiber”

there was no way of assessing in vivo microstruc-

to mean cardiomyocytes or groups of car-

tural changes through the cardiac cycle. Recent

“car-

in vivo human and animal work strongly supported

diomyocyte orientation” is used to avoid

the concept of sheetlet reorientation as an important,

cardiovascular magnetic

confusion with the more strictly fiber-like

if not the dominant, myocardial dynamic associated

resonance

skeletal muscle. In the left ventricle (LV),

with radial thickening (8). There is still limited un-

E2A = secondary vector

cardiomyocytes have a left-handed (LH) he-

derstanding of how myocyte shortening translates

angulation

lical course at the epicardium, progressing

into the microstructural rearrangement that causes

EF = ejection fraction

through to a circumferential course in the

wall thickening. However, approaches such as DT-

FA = fractional anisotropy

mesocardium and then a right-handed (RH)

CMR should yield insight into this problem.

HA = helix angle

helical course in the endocardium, as shown

HCM = hypertrophic

in the historical sequential dissection plates

cardiomyopathy

in Figure 1. Recently, histology by Streeter

LH = left-handed

et al. (2) demonstrated the transmural varia-

LV = left ventricular

tion of the fiber angle known as a . This angle,

M2-SE = second-order motion

measured in the plane parallel to the epicar-

compensated spin echo

dium and relative to the LV long axis, is

MD = mean diffusivity

positive at the endocardium and negative at

MI = myocardial infarction

the epicardium, reflecting RH and LH helical

RH = right-handed

orientation, respectively (Figure 2). The near-

STEAM = stimulated echo

zero a in the mesocardium indicates circum-

acquisition mode

ferential cardiomyocytes.

ADC = apparent diffusion coefficient

DCM = dilated cardiomyopathy DT-CMR = diffusion tensor

TA = transverse angle

diomyocytes.

Here,

the

phrase

CMR TECHNIQUE The theory and underlying methods behind diffusion MR were developed in the 1960s by Stejskal and Tanner (9) and advanced in the 1990s (10). The first clinical application of diffusion imaging was in neurology, where it was used for the early diagnosis of stroke (11). Cardiac diffusion imaging is considerably more difficult due to the bulk motion of the heart, but Edelman et al. (12) were able to overcome this by using a stimulated echo acquisition mode (STEAM) technique to produce the first in vivo

SHEETLETS. Cardiomyocytes are aggregated

to form secondary structures, termed “sheetlets” (Figure 3). Each sheetlet is approximately 4 car-

diffusion cardiac image in 1994. DIFFUSION. Diffusion imaging assesses diffusion of

water. Free diffusion, when water molecules may

diomyocytes thick, surrounded and interconnected

diffuse in all directions equally, is known as isotropic

by the collagenous perimysium, allowing shear be-

diffusion. However, in the myocardium where car-

tween neighboring sheetlets during the dynamic

diomyocytes and connective tissues act as barriers,

motion of cardiac contraction (3). It was initially

diffusion is restricted (impermeable barriers) or hin-

believed that sheetlets formed stacked sheets, each

dered (permeable barriers), becoming anisotropic

extending the width of the LV wall (4). It is now

(Figure 4). DT-CMR exploits the effects of the micro-

appreciated that

structure on diffusion to yield information about the

there

are

multiple

local sub-

populations containing counter-sloping sheetlets (5).

underlying myocardial organization.

STRUCTURELFUNCTION RELATIONSHIP. The myocardial

DIFFUSION-WEIGHTED CMR. Diffusion is measured

microstructure is crucial to the dynamics of LV

by assessing the signal loss between 2 images: a

function. The helical arrangement of cardiomyocytes

reference image and a main image with a higher de-

drives myocardial rotation and torsion (6). The

gree of diffusion weighting. The weighting is applied

dominant effect of the larger epicardial radius means

by diffusion encoding gradients and described by the

that cardiomyocyte shortening of the LH epicardial

b-value, which encompasses gradient amplitude,

cardiomyocytes leads to net clockwise rotation

duration, and temporal separation of the gradient

basally and anti-clockwise rotation apically, when

pulses, and is quoted in units of seconds per square

viewed from the apex (6).

millimeter (10). Typically, values of 450 to 600 s/mm 2

However, what is far less well appreciated is the

are used (8,13,14). At higher b-values, the signal in-

contribution of sheetlets to cardiac function. Both

tensity is less easily distinguished from background

Streeter et al. (2) and Spotnitz et al. (7) noted the

noise (15). At lower b-values, other sources of

“wall thickening paradox” in which cardiomyocyte

diffusion-like

thickening of only 8% could not explain wall thick-

including perfusion, confound the signal attenuation

ening of 40%. Histological analysis in rat and dog

(16). The measured diffusion is known as the

myocardium suggested the concept of sheetlet rear-

apparent diffusion coefficient (ADC) to accommodate

rangement from a more LV epicardial wall parallel

factors, such as perfusion and hindered and/or

orientation in diastole to a more wall perpendicular

restricted diffusion that impair accurate measures of

intravoxel

incoherent

motion,

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019

Khalique et al.

- 2019:-–-

DT-CMR Clinical Perspective

C ENTR AL I LL U STRA T I O N DT CMR Noninvasively Identifies Microstructures in the Myocardium

Khalique, Z. et al. J Am Coll Cardiol Img. 2019;-(-):-–-.

(Left) Diffusion tensor cardiovascular magnetic resonance (DT-CMR) demonstrates sheetlets of cardiomyocytes that reorient from wall-parallel in diastole to wallperpendicular in systole, facilitating the left ventricular wall thickening that greatly exceeds that of individual cardiomyocytes. In cardiomyopathy, abnormal sheetlet behavior is identified; in hypertrophic cardiomyopathy (HCM), sheetlets fail to relax to the expected diastolic position, and conversely, in dilated cardiomyopathy (DCM) sheetlets fail to rotate to the expected systolic position. (Right) DT-CMR demonstrates the helical arrangement of cardiomyocytes, with a right-handed helix in the endocardium, circumferential in the mesocardium, and left-handed helix in the epicardium. Contraction of these oppositely oriented myocytes drives clockwise rotation basally and anti-clockwise rotation apically, resulting in left ventricular torsion. In congenital conditions, such as situs inversus, this typical pattern is deranged with impaired torsion.

diffusivity (17). The fact that diffusion yields direc-

orientations uses diffusion gradients applied in at

tional information was developed from the obser-

least 6 directions. Information from the diffusion-

vation that ADC was higher when the diffusion

weighted images is used to determine a diffusion

gradient direction was aligned with the axonal di-

tensor, which is a mathematical extension of a

rection in neurological diffusion imaging (18). The

vector, having a 3-dimensional magnitude and di-

modern approach to determining microstructural

rection (19).

3

4

Khalique et al.

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019 - 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 1 Helical Arrangement of Cardiomyocytes in the Mammalian Heart

Fig . 1 .

Fig . 2 .

Fig . 3 .

Fig . 4 .

Fig . 5 .

Fig . 6 .

Fig . 7 .

Fig . 8 .

Dissection plates showing the transmural evolution of the helical arrangement of cardiomyocytes from (top left) left-handed helix in the epicardium through (bottom left) circumferential in the mesocardium to (bottom right) right-handed helix in the endocardium. Reprinted with permission from Pettigrew et al. (Supplemental Ref. 5).

ACQUIRING DT-CMR DATA. One method to acquire

The key limitation of basic SE is cardiac motion

DT-CMR data is a spin echo (SE) sequence, known as

that occurs during diffusion weighting, which is

the Stejskal-Tanner sequence (9). An excitation pulse

approximately 3 orders of magnitude greater than

is followed by a diffusion gradient that causes spin

the distances diffused by water molecules and leads

dephasing relative to position along the gradient.

to large artefactual signal dropout. This was first

Application of a 180-degree refocusing pulse and a

addressed by bipolar diffusion gradients that offered

second diffusion gradient will realign the spins

first-order motion (constant velocity) compensation

of static tissue to an equal but opposite net magne-

(20). Second- and third- (acceleration and jerk)

tization vector. For spins within a steadily flowing

order motion compensation techniques were then

medium, this pulse sequence will lead to a net phase

developed, although these require high performance

shift. However, in diffusion imaging, the random

gradient systems (Figure 6a) (14,21,22). Third-order

movement of the diffusing spins results in phase

motion compensation addresses the more complex

dispersion within each voxel. The vector sum of the

diastolic motion, but requires longer echo times, so

spins in the presence of phase dispersion leads to

second-order

decreased net magnetization, and thus signal loss

considered optimal. Thus, M2-SE data are typically

(Figure 5). Diffusivity can be calculated from signal

acquired in systole (23,24). Validation using in vivo

loss and diffusion weighting using the Stejskal-

and post-mortem M2-SE DT-CMR in pigs has shown

Tanner equation (9).

good levels of agreement (25). An M2-SE approach is

motion

compensation

(M2-SE)

is

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019

Khalique et al.

- 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 2 Transmural Variation in Cardiomyocyte Orientation

DECILE NO.

ENDOCARDIUM

100 a

1 M d c

80

b

2

α(°)

c

60

da M b

3

c d

40

4 20

M a

b d

5

c

M a bd

MIDWALL 6

0 20 LV WALL THICKNESS -%

ca M b d

0 ac 40 M b d

–20

M acd b

60

80 d M cab

100

d

c M ab

d bc

M

7

c

–40

a b

M d

8

b c

M

–60 a

d

9

100 μ

–80

a

10 –100 EPICARDIUM

(Left) Cardiomyocyte orientation smoothly progresses from a positive angle at the endocardium through to a negative angle at the epicardium. (Right) Helix angle through the myocardial wall from epicardium (negative) to endocardium (positive). a to d represent data from 4 samplings. LV ¼ left ventricular; M ¼ mean. Reprinted with permission from Streeter et al. (2).

attractive and growing in popularity because of the

cycles (Figure 6B). Data acquisition is synchronized so

higher

alternative

that the heart is in the same position when both

methods, its suitability for free-breathing acquisi-

diffusion gradients are applied, thus minimizing the

tion, and its more efficient whole heart coverage. It

effects of cardiac motion. STEAM has been the

has been used in several initial human studies

mainstay of recent clinical DT-CMR studies because it

(14,22,26,27). A variation on M2-SE is a diffusion

does not require specialist hardware or motion

prepared method that allows a flexible choice of

compensation. Since its initial success in producing

image acquisition (22).

diffusion-weighted CMR images, STEAM has been

signal-to-noise

ratio

than

A leading alternative sequence type is STEAM, and

readily applied to DT-CMR in both healthy and

this addresses bulk motion artefact by splitting the

diseased hearts (8,12,28–30). However, the wide

180-degree pulse into 2 90-degree pulses, so that the

applicability of STEAM is offset by several disadvan-

time over which diffusion is measured (D ) is an RR

tages. It has low signal-to-noise ratio, requires 2 reg-

interval, and each image is acquired over 2 cardiac

ular RR intervals per image, which necessitates longer

5

6

Khalique et al.

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019 - 2019:-–-

DT-CMR Clinical Perspective

THE

F I G U R E 3 Cardiomyocytes Aggregated into Sheetlets

CONFOUNDING

EFFECTS

OF

STRAIN.

Strain, a measure of myocardial deformation, has been proposed as a confounder of DT-CMR results (28,32). Myocardial deformation occurs during the diffusion time of both sequences, but the longer diffusion time in STEAM make its effects more significant. Reese et al. (32) originally described this phenomenon, explaining how tissue compression, in the direction of and following the initial STEAM gradient phase encoding, would have the effect of increasing the spatial modulation frequency and signal attenuation, and hence, ADC. An alternative simplified visual representation of how strain affects diffusion measures is shown in Figure 7. Diffusion within a medium that is compressed and then stretched back to its original state results in protons that travel further away from the starting point compared with the stationary case, thus over-

Scanning electron micrograph of the transverse cut surface of a left ventricular mid-wall

estimating diffusivity. Several correction methods for

section shows cardiomyocytes aggregated into sheetlets, separated by peri-mysial

STEAM sequences exist, including bipolar gradients

connective tissue. Adapted with permission from Le Grice et al. (3).

(33) or acquiring data at the “sweet spot,” the timepoint where the effects of the strain history over the

breath-holding or a navigator-led approach, and excludes

patients

with

arrhythmias

(31).

Another

disadvantage of STEAM in measuring diffusion is the effect of strain.

cardiac cycle are nulled (34). A post-processing correction technique uses macroscopic strain data (35). However, although strain has been considered a significant confounder to DT-CMR findings for many years (especially systolic secondary vector angulation [E2A] and E2A mobility), 2 recent studies have shown

F I G U R E 4 Diffusion Is Modulated by Its Environment

TISSUE/CELL LEVEL Hindrance Restriction Barriers

strain has only a limited effect and that uncorrected in vivo DT-CMR data are more representative of ex vivo findings (8,36). Consequently, the dynamic, laminar, and semi-permeable nature of the myocardium means strain has a complex, but limited effect on measured diffusion and requires further study (29,36,37). ANALYZING THE DATA.

The eigensystem. The 3-

dimensional diffusion described by a tensor can be

pictorially

diffusion

in

represented.

an

Although

unrestricted

isotropic

environment

is

a

sphere, the restricted diffusion in the heart is an ellipsoid

(Figure

8).

The

lengths

of

the

3-

dimensional axes of the ellipsoid are labeled l1, l 2

l3 (eigenvalues) in order of magnitude of diffusion, and their directions are E1, E2, and E3 (eigenvectors), respectively. Eigenvectors. The eigenvectors can be projected onto cardiac planes (Figure 9), and several validation studies support the correspondence of E1 with average intravoxel cardiomyocyte orientation, E2 Free diffusion is unrestricted, but in the myocardium can be restricted within cells (green), may traverse a permeable cell membrane (red) or take a hindered path within the extracellular space (blue), rendering it anisotropic. Adapted with permission from Le Bihan (Supplemental Ref. 6).

with the predominant sheetlet orientation, and E3 with the sheetlet normal direction (Supplemental Table 1) (5,8,38–43). Helix angle. The primary eigenvector, the long axis of the

diffusion

ellipsoid,

is

projected

onto

the

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019

Khalique et al.

- 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 5 Diffusion Is Measured by Signal Loss

90°-Pulse

Echo

180°-Pulse G

G

Fixed

A

Flow

B

Diffusion

C

The diffusion tensor cardiovascular magnetic resonance (DT-CMR) sequence uses an excitation pulse followed by a diffusion gradient, a refocusing pulse, and a second diffusion gradient. Static Hþ spins are subject to (A) equal dephasing and rephasing, whereas flowing spins (B) experience phase shift. (C) With random Hþ displacement, there is incomplete rephasing and signal attenuation. Reproduced with permission from Froeling et al. (Supplemental Ref. 7).

circumferential-longitudinal plane and reflects the

change is called E2A mobility and reflects sheetlet

mean intravoxel cardiomyocyte orientation (8,28,39).

function (8).

The angulation created by the projected primary

Tertiary eigenvector angulation. The tertiary eigen-

eigenvector and the circumferential direction is E1A

vector angulation (E3A) is perpendicular to the

but is often described as the helix angle (HA) or a in

sheetlet plane. E3 is projected onto the longitudinal-

some older published data. HA changes only a little

radial plane and is positive when directed apico-

during cardiac contraction (8).

basally (upward).

Transverse angle. The transverse angle (TA) or imbri-

Mean diffusivity. In DT-CMR, mean diffusivity (MD) is

cation angle reflects the tilt of the mean car-

derived from the tensor (19). It is the mean of l 1, l 2,

diomyocyte orientation toward and/or away from

and l3 and measured in units of square millimetres

the central LV axis. It is derived by projecting E1

per second. MD reflects the packing and integrity of

into

then

the myocytes, with low values indicating tight pack-

calculating the angle with the circumferential di-

ing. MD rises with interstitial abnormality (e.g.,

rection (38,43).

fibrosis) because water has more freedom to diffuse

the

Secondary

circumferential-radial

eigenvector

plane

angulation. The

and

secondary

(26,44). MD is similar to the mean ADC from

eigenvector is projected onto the cross-myocyte

diffusion-weighted CMR.

plane (Figure 9). The angulation between E2 pro-

Fractional anisotropy. Fractional anisotropy (FA) is a

jected into this plane and the myocyte perpendicular

scalar value reflecting the degree, but not direction,

direction is termed E2A and is an index of sheetlet

of anisotropy. Isotropic diffusion has an FA value of

orientation (8). Unlike E1A, E2A orientation changes

0, and diffusion restricted to a single direction has a

substantially between systole and diastole. This

value of 1 (Figure 10). FA reflects the degree of

7

Khalique et al.

8

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019 - 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 6 Diffusion Tensor Sequences

R

Diffusion gradient

90°(y) water selective

Diffusion gradient

M2-SE

180°(z)

A

TE/2

EPI readout TE/2

Δ

R

TE/2

Diffusion gradient

90°(z)

90°(y)

R

90°(y) water selective

STEAM

Diffusion gradient

B

EPI readout

TE/2

Ttrigger Δ = RR TM

(A) Second-order motion compensated spin echo sequence (M2-SE). The sequence is triggered by the R wave and has excitation and refocusing pulses. D is the “diffusion time” over which the measured diffusion takes place (typically 20 to 30 ms). (B) Stimulated echo acquisition mode (STEAM) DT-CMR sequence. The sequence runs across 2 cardiac cycles; the diffusion time (D) is 1 RR interval, whereas mixing time (TM) reflects the duration of the magnetization storage in the longitudinal plane between the second and third pulses. Modified with permission from Scott et al. (24).

organization of a tissue, such that a higher value is

cardiac and respiratory motion (24). Scan times can

more organized. Thus, a low FA might be expected

be lengthy due to multiple acquisitions in different

with myocyte disarray, should it occur at an appro-

directions.

priate value relative to pixel size.

DT-CMR is limited in both spatial and angular

VISUALIZING DT-CMR FINDINGS. Tractography provides

3-dimensional visualization, typically tracking E1 across voxels into a smooth trajectory. In reality, this is a simplification because the myocardium does not contain serially aligned cardiomyocytes. Another representation uses superquadric glyphs, which are 3-dimensional bricks derived from the tensor that convey pixelwise information about the shape and orientation of the eigensystem (45). The orientation and size of the 3 axes of the glyphs represent

the

eigenvectors

and

eigenvalues,

respectively. LIMITATIONS OF DT-CMR. Acquiring and interpret-

ing DT-CMR data is hampered by technical issues

resolution. Current in vivo spatial resolution is typically 2.7  2.7 mm 2  6 mm for M2-SE and 2.8  2.8 mm2  8 mm for STEAM (8,14). Each DT-CMR image voxel contains millions of cardiomyocytes and many sheetlets. These are not uniformly aligned, and counter-orientated subpopulations exist (5). DTCMR techniques are not able to resolve multiple populations within a single voxel, so the derived values reflect averages. It is also possible that partial volume effects may be responsible. Techniques such as de-noising and different k-space trajectories may aid reducing voxel size but help preserve the signalto-noise ratio.

DT-CMR OF THE HEALTHY HEART

(46). Success rates for image acquisition vary by sequence type and cardiac phase and can be

Clinical applications of DT-CMR are in relative in-

affected by factors such as mis-triggering and

fancy, but the unique ability to provide in vivo

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019

Khalique et al.

- 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 7 Effects of Strain in an Isotropic Phantom

time 0

time 1 R-R interval

STATIC

diffusion encoding time = 1 cardiac cycle

FITTED TENSOR ELLIPSOID

free isotropic diffusion elastic medium isotropic

STRETCH > 1

stretching

reduction of diffusion along stretching direction

STRETCH < 1

compressing

enhancement of diffusion along compression direction

Deformation of an isotropic substance over the diffusion time affects the measured diffusion result. In the top panel, a systolic measurement is affected by the intervening diastolic stretch, and in the bottom panel a diastolic measurement is affected by the intervening systolic compression. For both, the diffusion pattern is ovoid in shape, rather than the expected isotropic sphere. Reprinted with permission from Ferreira et al. (36).

microstructural information noninvasively has led to

may relate to the laminar structure of the myocar-

some describing it as “virtual histology.” The tech-

dium, because sheetlets are increasingly prevalent

niques are in constant evolution, and although mea-

toward the endocardium (49).

surement of myocyte and sheetlet orientations are

CARDIOMYOCYTE ORIENTATION. Various DT-CMR

reproducible and stable, the more global measures

studies identify the transmural HA gradient in car-

such as FA and MD show variation with technical

diomyocyte orientation (Supplemental Table 1) and

sequence parameters and acquisition, mostly due to

report only minor changes in cardiomyocyte orien-

the varying diffusion times (24,27).

tation toward a more longitudinal systolic orienta-

DIFFUSIVITY

PARAMETERS. MD

and

FA

are

tion, as shown in Figure 12 (8,35,50). The transverse

commonly reported DT-CMR parameters but vary

angle lies in the region of
20 to þ20 degrees, at both

depending upon sequence type, cardiac phase, and

cardiac phases (43).

b-value (24,47,48). In vivo human studies comparing M2-SE and STEAM report higher MD and lower FA using M2-SE (24,27). This result is expected because with a shorter diffusion time, water molecules encounter fewer restrictive barriers (Figure 11). Re-

F I G U R E 8 Pictorial Representation of Diffusion Tensors

λ1, E1

λ1, E1

ported values for MD and FA are provided in Supplemental Table 2. MD and FA tend to be provided

λ3, E3

as global values, usually across a mid-ventricular slice. However, regional differences

have

been

0.04), compared with the endocardium (0.40  0.04)

λ2, E2

and epicardium (0.39  0.004), possibly due to the relative plateau in HA due to the circumferentially oriented mesocardial cardiomyocytes (48). For MD,

λ3, E3

λ2, E2

described: FA peaks in the mesocardium (0.46 

(Left) Isotropic diffusion is shown as a sphere, but restricted diffusion is represented as an (right) ellipsoid shape in DT-CMR. l signifies the eigen-

there is a transmural gradient with an MD of 0.87 

value (magnitude of diffusion) and E denotes the eigenvector (direction of

0.07  10
3 mm 2/s at the epicardium that increases to

diffusion) along the 3 axes.

0.91  0.08  10
3 mm 2/s in the endocardium. This

9

Khalique et al.

-, NO. -, 2019

JACC: CARDIOVASCULAR IMAGING, VOL.

- 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 9 Projections of Eigenvectors on Cardiac Planes

longitudinal di ra al os cr

c yo

m s-

A endo

longitudinal

E2 meso

circumferential epi

E1 B

l

E1 A +T E1

circumferential ra di al

-TA

c yo

circumferential ra di al

yt

e

m s-

os

cr

E2A

l

m s-

ia

E2

ia

Transverse angle

e

circumferential ra d

circumferential ra d

yt

os

cr

longitudinal

-HA

e

E1

+HA circumferential

c yo

yt

E1

cross-myocyte

C Sheetlet planes

Helix angle longitudinal

10

ra d

ia

l

Three myocardial voxels are shown. The primary eigenvector (E1) projected on the circumferential-longitudinal plane gives the helix angle (HA). (A) E1 projected on the circumferential-radial plane gives the transverse angle. (B) The voxels are cut in a cross-myocyte direction, showing the plane perpendicular to E1. (C) Projecting the second eigenvector (E2) onto this sheetlet plane gives E2A. Adapted with permission from Ferreira et al. (29).

SHEETLET ORIENTATION. E2A is used as a DT-CMR

MICROSTRUCTURAL CHANGES IN

index of sheetlet orientation, although E3 is used by

CARDIAC DISEASE

some

investigators.

E2A

changes

substantially

through the cardiac cycle (8). Figure 13 shows how

MI. DT-CMR has been applied in MI, with much in-

sheetlets adopt a wall parallel sheetlet orientation in

terest in acute changes and post-infarct ventricular

diastole (blue), which becomes more wall perpen-

remodeling. Several animal and human studies

dicular in systole (red). This reorientation is inti-

describe increased MD and decreased FA in the

mately linked to the generation of radial strain. A

infarct zone (26,52–55). This accords with myocyte

small movement in the longitudinal direction trans-

swelling and necrosis, extracellular matrix expan-

lates into a larger radial expansion, similar to the

sion, and replacement by less organized collagenous

perpendicular amplification action of a “lazy tong”

fibrosis. Ex vivo porcine DT-CMR 13 weeks after iat-

(Supplemental Ref. 1). This is demonstrated by glyphs

rogenic MI revealed increased mean ADC (1.01 

in Figure 13 and shown in Video 1. Biphasic differ-

0.100  10
3 mm 2/s) compared with control subjects

ences in E2A have been described; a STEAM study of

(0.671  0.106  10
3 mm 2/s; p < 0.001) in the infarct

healthy humans reported diastolic E2A of 26  6 de-

zone (55). FA was significantly reduced in the infarct

grees and systolic E2A of 54  6 degrees, with E2A

from 0.32  0.01 to 0.20  0.03 (p < 0.001). In the

mobility of 27  8 degrees (51). A different STEAM

human study by the same group 26 days post-MI, DT-

study using E3A reported sheet angles of 65  3.6

CMR detected increased mean ADC, decreased FA,

degrees and 41.6  3.7 degrees in diastole and systole,

and decreased RH and increased LH helical structures

respectively (35).

(cardiomyocytes) compared with the remote zone

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019

Khalique et al.

- 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 1 0 Increasing FA

Tensor

Increasing fractional anisotropy

0

1

Increasing microstructural organization

Tissue organization

Fractional anisotropy (FA) in the heart reflects the underlying myocyte organization, such that a value towards zero reflects little organization and a value nearer 1 reflects a more organized structure (packed and anisotropic).

(52). The follow-up human study at a median 191 days

5 patients suggested DT-CMR appeared to detect

after MI described an increased proportion of RH

disarray by reduced FA and lower strain in the

cardiomyocytes in the remote zone, which correlated

septum compared with the free wall of patients with

with increased wall thickness (r ¼ 0.66; p ¼ 0.005)

HCM and 5 healthy control subjects (58). A repro-

and increased wall thickening in the adjacent zone

ducibility study in HCM reported regional differences

(r ¼ 0.61; p ¼ 0.01) (56). The proposed explanation

where FA was lowest and MD highest in the septum,

was that loss of RH cardiomyocytes fit with susceptibility of the endocardium to ischemia and gain of LH cardiomyocytes in the infarct zone and RH cardiomyocytes in the remote zone were both adaptive

F I G U R E 1 1 Expected Diffusion Radii of Free Water

remodeling responses. Recently, a study of MI in mice, sheep, and humans introduced the tractographic propagation angle, defined as the angle between 2 adjacent primary eigenvectors along a given tract and a metric of myofiber curvature, as shown in Figure 14 (57). A propagation angle value of 4 degrees was reported as the

threshold

for

differentiating

normal

from

infarcted myocardium, as correlated to late gadolinium imaging and endocardial voltage mapping. Understanding

microstructural

changes

post-infarct

could offer insight into the pathogenesis and prognostication of negative remodeling and arrhythmias. HCM. A

microscopic

hallmark

of

HCM

is

car-

diomyocyte disarray, detectable currently only by histology. It is hoped that DT-CMR might detect

STEAM has a longer diffusion time (D) than spin echo. The diffusion radii of free water over these timeframes are superimposed upon pig myocardium, cut perpendicular to the

disarray noninvasively, aiding diagnostic differenti-

long axis of the cardiomyocytes. More barriers are encountered during the longer D.

ation from HCM phenocopies and potentially identify

Reproduced with permission from Scott et al (24). Abbreviations as in Figure 6.

those at higher risk of arrhythmias. A study of

11

Khalique et al.

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019 - 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 1 2 Helix Angle in Diastole and Systole

A

End-Diastole

B

End-Systole

Epicardium

60°

C

Base

Apex

D Endocardium

Lateral Wall

12

-60°

Diffusion tensor tractography of the lateral ventricular wall at end-diastole (A and C) and end-systole (B and D). Negative epicardial helix angles are more red and the positive endocardial helix angles are more blue. At end-systole the cardiomyocytes are more longitudinally orientated. Reprinted with permission from Mekkaoui et al. (50).

although only the latter was significant (59). Although

pathophysiology, in which sarcomeric mutations may

this might be related to septal hypertrophy and

result in increased myofilament sensitivity to cal-

expanded extracellular space due to fibrosis and

cium, increasing relative cardiomyocyte tension

disarray, technical factors such as spatial variation in

(Supplemental Ref. 2).

the signal-to-noise ratio might also be responsible;

DILATED CARDIOMYOPATHY. Dilated cardiomyopa-

histological validation was absent. A preliminary

thy DCM has also been studied using DT-CMR. A SE

study suggested FA was focally reduced in patients

study of hamsters with DCM found no change in HA

with HCM compared with control subjects and that

or TA, but elevated MD and reduced FA compared

DT-CMR might be able to detect myocardial disarray

with control hamsters (Supplemental Ref. 3). An

(60).

in vivo STEAM study of 9 patients with nonischemic

Diffusion-weighted CMR can offer another type of

DCM and control subjects found MD was similar be-

myocardial characterization in HCM. In a study of 23

tween groups and cardiac phases (30). FA was lower

patients with HCM, mean ADC was elevated in areas

in patients with DCM compared with control subjects

of fibrosis and correlated with extracellular volume

(diastole 0.56  0.07 vs. 0.63  0.05; p < 0.04, and

(r 2 ¼ 0.72), which suggested that diffusion-weighted

systole 0.58  0.08 vs. 0.62  0.07; p ¼ 0.56). The

CMR might detect the extent of fibrosis without a

transmural HA gradient in diastole was steeper in

contrast agent (Figure 15) (44). However, in vivo DT-

patients with DCM, although this difference seemed

CMR demonstrated an entirely novel abnormality in

to be driven by 2 patients. There was systolic HA

HCM. Diastolic E2A was significantly elevated, and

redistribution in control subjects, with the expected

E2A mobility was reduced (Figure 16) (29). The failure

more longitudinal orientation, but this was not seen in

of sheetlets to return to a more wall parallel orienta-

patients with DCM. E2A findings are shown in

tion in diastole could be described as a failure of

Figure 17. Biphasic E2A changes in control subjects

diastolic relaxation. Interestingly, this microstruc-

(Figures 17A and 17C) showed predominance of low

tural abnormality builds upon our knowledge of HCM

diastolic E2A and a broader systolic distribution.

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019

Khalique et al.

- 2019:-–-

13

DT-CMR Clinical Perspective

F I G U R E 1 3 Secondary Eigenvector E2A Reflects Sheetlet Orientation

A

IN VIVO: DT-CMR AT DIFFERENT TIMES OF THE CARDIAC CYCLE diastole

90 80

|E2A| (Degrees)

80 70 60 50

45

40

systole

30 20 10

10

0 0

0.25 RR

0.5 RR

0.75 RR

RR

Normalized RR Interval

B

HISTOLOGY RELAXED (KCI ARRESTED)

CONTRACTED (BaCI2 ARRESTED)

0 –45

0 +45

–/+90

–45

+/–90

–/+90

–45

+/–90

+45

–45 0

80

10.5 mm

60 1 mm

1 mm

9.9 mm

0

Sheetlet Angle (Degrees)

+45

+45

40 20

ra c

9.0 mm

nt

1 mm

Co

Re l

1 mm

te

ax ed

5.6 mm

d

0

Absolute secondary vector angulation (E2A) changes through the cardiac cycle, rising from diastole (low E2A: blue) to systole (higher E2A: red). (A) Biphasic E2A variation is seen in vivo and corresponds to the (B) histological findings of sheetlet orientation. See Video 1. Adapted with permission from Nielles-Vallespin et al. (8).

However, there was relatively little change in

abnormal in patients with DCM. In control subjects,

the

DCM

median systolic (interquartile range) E2A was 65 (6  ),

(Figures 17B and 17D). The findings of this study must

and in patients with HCM, it was 74 (7  ). In patients

be tempered by its limitations; only 9 patients without

with DCM, systolic E2A was significantly reduced at

LV dilatation and with strain correction, which may

40 degrees (16 degrees), yielding a reduced E2A

E2A

distribution

for

patients

with

have obscured, had altered sheetlet mobility.

mobility and impaired sheetlet mobility (Figure 18).

Nielles-Vallespin et al. (8) studied patients with

Both radial and circumferential strains were signifi-

DCM and HCM, and control subjects and did not use

cantly and similarly reduced in the cardiomyopathy

strain correction. They found similar HA distributions

groups compared with control subjects. However, the

in the 3 cohorts. HA shifted toward more longitudinal

range of E2A (low to low in DCM and high to high in

in systole for all groups. Diastolic E2A was similar

HCM) provided mechanistic insight into these find-

between groups, but systolic E2A findings were

ings. This study was limited by the mean ejection

14

Khalique et al.

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019 - 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 1 4 PA Is Elevated in Areas of Infarction

A

Propagation Angle

B

Normal Human

Base Lateral

C

Infarcted Sheep

C1 Remote Zone PA < 4°

Anterior C2 Infarcted Region PA > 4°

Septum

Apex

0°

Anterior

15°

(A) Propagation angle (PA) is a metric of myofiber curvature, (B) PA in healthy human hearts is uniformly typically <4 , and (C) in a sheep heart with myocardial infarction remote and infarcted areas are delineated by PA #4 , respectively. Reprinted with permission from Mekkaoui et al. (Supplemental Ref. 8).

fraction (EF) in the DCM group of 45  11%, with

difference in the HA histograms among the 3 cohorts,

approximately one-half of the patients having only

but significant findings were present in E2A. Diastolic

mild LV impairment. However, despite this, the

E2A was similar among groups. However systolic E2A

altered sheetlet behavior was statistically significant.

in those with recovered DCM was 59 (14 degrees),

Having established microstructural abnormalities

which was significantly greater than the DCM group

in DCM, the same group studied patients with

(35 degrees [17 degrees]; p < 0.0001), but lower than

recovered DCM and compared them to patients with

the control subjects (65 [8 ]; p ¼ 0.01). When E2A

DCM and healthy control subjects (61). These patients

mobility was plotted against the EF, patients with

were defined by LV size and EF entirely within

recovered DCM and control subjects were indistin-

indexed reference ranges and New York Heart Asso-

guishable by EF but could be discriminated by E2A

ciation functional class I. Again, there was little

(Figure 19).

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019

Khalique et al.

- 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 1 5 Diffusivity Is Elevated in Areas of Fibrosis

ECV

LGE

Diffuse

Patch–like

ADC

0

3 µm2/ms

60%

0

2

30%

In hypertrophic cardiomyopathy, the apparent diffusivity coefficient (ADC) is elevated in regions of fibrosis, as detected by late gadolinium enhancement (LGE) and increased extracellular volume (ECV). Reprinted with permission from Nguyen et al. (44)

Together, these 2 studies suggest that DT-CMR can

epicardium and negative HAs in the endocardium,

detect impaired sheetlet behavior in dilated cardio-

whereas toward the apex, the transmural HA distri-

myopathy that appears to persist even when LV size

bution was more similar to situs solitus. There was a

and EF recover to normal by conventional measures.

mid-ventricular transition zone. These findings were

CONGENITAL HEART DISEASE. There is a paucity of

confirmed in the ex vivo hearts. Mid-LV peak radial

DT-CMR data in congenital heart disease. In an adult

and circumferential strain were reduced, and overall

with transposition of the great arteries corrected by

absolute torsion was significantly reduced in situs

arterial switch, DT-CMR showed predominance of

inversus totalis. This was the first in vivo demon-

longitudinal and oblique fibers in the systemic

stration of substantial departure from the typical

right ventricle, which is believed to be an adaptive

mammalian

response

(Figure

to

the

systemic

pressure

and

load

helical

cardiomyocyte

arrangement

20). In addition to the abnormal car-

(Supplemental Ref. 4). A larger study of 12 patients

diomyocyte arrangement in situs inversus totalis,

with situs inversus totalis, 12 healthy situs solitus

there was also impaired sheetlet mobility with fail-

control subjects, and 2 ex vivo situs inversus totalis

ure to achieve the expected systolic orientation.

hearts showed deranged cardiomyocyte orientation,

This might reflect cardiomyocyte derangement that

sheetlet abnormalities, and functional impairment in

affected the transverse shears that were integral to

situs inversus totalis (62). All situs solitus hearts

sheetlet reorientation (6). These findings reinforced

displayed the expected helical arrangement. By

the connection between the microstructure and

contrast, in situs inversus totalis, HAs were more

cardiac mechanics. The abnormal cardiomyocyte

heterogeneous, both intrasubject and intersubject.

structure in situs inversus totalis raises the question

However, the general pattern was an inverted HA

of whether such patients might have a greater sus-

arrangement at the base, with positive HAs in the

ceptibility to heart failure with a “second hit” to

15

Khalique et al.

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019 - 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 1 6 E2A Abnormalities in HCM

Healthy control

HCM patient 80

45

Diastole

10

80

Systole

45

10

90 80 p = 0.026 70 E2 Angle (Degrees)

16

60 50 40 30 20 p < 0.001 10 Diastole-Systole Controls Controls

Diastole-Systole HCM HCM

Quartiles

Example E2A maps of a mid-ventricle slice. The healthy control subject has a blue (low E2A) map in diastole and a red (higher E2A) map is systole. In hypertrophic cardiomyopathy (HCM), maps are predominantly red, displaying elevated E2A at both cardiac phases. The plot below shows that diastolic E2A is significantly elevated compared with the healthy control subjects, as is systolic E2A. Adapted with permission from Ferreira et al. (29). Abbreviation as in Figure 13.

function,

such

as

hypertension,

infarction,

or

applicability and usefulness of the technique. Two

myocarditis.

main streams of future development are needed:

FUTURE DIRECTIONS OF DT-CMR

technical would

and

address

clinical. spatial

Technical and

developments

angular

resolution,

DT-CMR can identify microstructural changes in

scan acceleration, increased coverage, and refining

disease. Further work is required to broaden the

the

understanding

of

strain

effects.

Clinical

-, NO. -, 2019

JACC: CARDIOVASCULAR IMAGING, VOL.

Khalique et al.

- 2019:-–-

17

DT-CMR Clinical Perspective

F I G U R E 1 7 E2A Findings in Dilated Cardiomyopathy

B

Control

DCM

0.30

0.30

0.25

0.25

0.20

0.20 Frequency

Frequency

A

0.15

0.15

0.10

0.10

0.05

0.05

0

0

–100 –80

–60

–40

–20

0

20

40

60

80

100

–100 –80

–60

–40

E2A Angles [*] Diastole

C

Systole

Diastole

D

0.20

0.20

0.15

0.15

0.10 0.05 0

–0.10

20

80

100

40

60

80

100

60

80

100

Systole

0

–0.10

0

60

0.05

–0.05

–20

40

0.10

–0.05

–40

20

DCM 0.25

Change in Frequency

Change in Frequency

Control

–60

0

E2A Angles [*]

0.25

–100 –80

–20

–100 –80

–60

–40

–20

0

20

E2A Angles [*]

E2A Angles [*]

Data

Data

Model

40

Model

(A and B) E2A histograms for control subjects and patients with dilated cardiomyopathy. Diastolic and systolic distributions are shown by solid and dashed lines, respectively. (C and D) Change from the expected computer model results. Reprinted with permission from von Deuster et al. (30).

development might see DT-CMR playing a role as a

debate on whether patients with DCM whose EF is

diagnostic and prognostic tool. The ability to detect

improving have achieved recovery or remission. In

microstructural changes before manifest disease

HCM, the ability (if possible in the future) to detect

expression may offer a role for earlier diagnosis of

disarray in vivo could assist risk stratification for

cardiomyopathies. It may also help identify pa-

predicting sudden cardiac death. Similarly, identi-

tients at risk of developing heart failure, not only

fying those at risk of potentially life-threatening

in

and

arrhythmias after infarction is another area in

ischemic heart disease. DT-CMR may inform the

which DT-CMR has the potential to offer novel

cardiomyopathy,

but

also

congenital

18

Khalique et al.

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019 - 2019:-–-

DT-CMR Clinical Perspective

F I G U R E 1 8 Abnormal Sheetlet Reorientation in Dilated Cardiomyopathy

SHEETLET PLANE ORIENTATION E2

CONTROLS - DIASTOLE

CONTROLS - SYSTOLE

HCM – DIASTOLE

HCM – SYSTOLE

DCM – DIASTOLE

DCM – SYSTOLE

E1

(Top-right inset) Sheetlet planes are defined by the primary and secondary eigenvectors (E1 and E2). Septal glyphs are shown, color-coded according to absolute E2A (blue towards wall-parallel and red towards wall-perpendicular). Healthy control subjects show a more wallparallel alignment in diastole and more wall-perpendicular alignment in systole. However, in HCM, sheetlets retain the more wallperpendicular arrangement of systole in both cardiac phases. Conversely, in dilated cardiomyopathy (DCM), the more wall-parallel arrangement of diastole persists in both cardiac phases. Replotted from Nielles-Vallespin et al. (8). Abbreviation as in Figure 16.

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019

Khalique et al.

- 2019:-–-

DT-CMR Clinical Perspective

Absolute E2A Mobility (Degrees)

F I G U R E 1 9 E2A mobility and EF Plot in Recovered DCM

60 50 Median p = 0.001

40

Median

30 20 10 0 –10 20

30

40

50

60

70

80

LVEF (%) Controls

Recovered DCM

DCM

Secondary eigenvector (E2A) mobility plotted against left ventricle ejection fraction (LVEF). Patients with DCM patients have low LVEF and low E2A mobility. Both control subjects and recovered DCM groups have normal EF, but the recovered DCM group can be discriminated by their significantly lower E2A mobility. Reprinted with permission from Khalique et al. (61). Abbreviations as in Figures 13 and 18.

F I G U R E 2 0 LV Tractography in SIT

ENDOCARDIUM

SITUS SOLITUS

EPICARDIUM

SIT IN VIVO

Helix angle (degrees) +90 +45 +0 -45

SIT EX VIVO

-90

LV tractography of a situs solitus (SS) and situs inversus (SIT) heart. The contrast between the left-handed helix (blue) in the SS heart, and the right-handed helix (red) in the SIT heart is particularly striking. Modified with permission from Khalique et al. (62).

19

20

Khalique et al.

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019 - 2019:-–-

DT-CMR Clinical Perspective

CONCLUSIONS

HIGHLIGHTS  DT-CMR is a unique technique for noninvasive assessment of myocardial microstructure.

DT-CMR is establishing itself as a technique that can

 DT-CMR has identified novel cardiomyocyte and sheetlet abnormalities in cardiac disease in vivo.

cardiomyocyte and sheetlet structure and micro-

 DT-CMR has the potential to aid clinical diagnosis and risk stratification through microscopic phenotyping.

potential to progress to a clinically useful tool aiding

interrogate the cardiac microstructure and its relation to function. For the first time, it is possible to assess structural dynamics using a noninvasive approach. DT-CMR provides insight into pathological changes in disease, and, with this new understanding, it has the diagnosis and prognosis, as well as guiding patient management.

prognostic information. Expansion to other cardio-

ADDRESS FOR CORRESPONDENCE: Professor Dudley

myopathies and congenital heart disease may offer

Pennell, CMR Unit, Royal Brompton Hospital, London

new

SW3 6NP, United Kingdom. E-mail: Dj.pennell@rbht.

pathophysiological

insight

conditions.

into

these

nhs.uk.

REFERENCES 1. Opie LH, Bers DM. Mechanisms of cardiac contraction and relaxation. In: Mann DL, Zipes DP, Libby P, Bonow RO, editors. Braunwald’s Heart Disease: A Textbook of Cardiovascular Medicine. 10th Edition. Philadelphia, PA: Elsevier Saunders, 2015; p 429–53. 2. Streeter DD, Spotnitz HM, Patel DP, Ross J, Sonnenblick EH. Fiber orientation in the canine left ventricle during diastole and systole. Circ Res 1969;24:339–47. 3. LeGrice IJ, Smaill BH, Chai LZ, Edgar SG, Gavin JB, Hunter PJ. Laminar structure of the heart: ventricular myocyte arrangement and connective tissue architecture in the dog. Am J Physiol Heart Circ Physiol 1995;269:H571–82. 4. LeGrice IJ, Takayama Y, Covell JW. Transverse shear along myocardial cleavage planes provides a mechanism for normal systolic wall thickening. Circ Res 1995;77:182–93. 5. Kung G, Nguyen TC, Itoh A, et al. The presence of 2 local myocardial sheet populations confirmed by diffusion tensor MRI and histological validation. J Magn Reson Imaging 2011;34:1080–91. 6. Young AA, Cowan BR. Evaluation of left ventricular torsion by cardiovascular magnetic resonance. J Cardiovasc Magn Reson 2012;14:49.

gradient sensitization: design, strategy, and significance. J Magn Reson Imaging 1991;1:7–28. 11. Warach S, Chien D, Li W, Ronthal M, Edelman RR. Fast magnetic resonance diffusionweighted imaging of acute human stroke. Neurology 1992;42:1717–23.

22. Nguyen C, Fan Z, Xie Y, et al. In vivo diffusiontensor MRI of the human heart on a 3 tesla clinical scanner: an optimized second order (M2) motion compensated diffusion-preparation approach. Magn Reson Med 2016;76:1354–63.

13. Scott AD, Ferreira PFADC, Nielles-Vallespin S, et al. Optimal diffusion weighting for in vivo cardiac diffusion tensor imaging. Magn Reson Med 2015;74:420–30.

23. Welsh CL, DiBella EVR, Hsu EW. Higher-order motion-compensation for in vivo cardiac diffusion

14. Stoeck CT, von Deuster C, Genet M, Atkinson D, Kozerke S. Second-order motioncompensated spin echo diffusion tensor imaging of the human heart. Magn Reson Med 2016;75: 1669–76. 15. Jones DK, Basser PJ. Squashing peanuts and smashing pumpkins?: How noise distorts diffusionweighted MR data. Magn Reson Med 2004;52: 979–93. 16. Callot V, Bennett E, Decking UKM, Balaban RS, Wen H. In vivo study of microcirculation in canine myocardium using the IVIM method. Magn Reson Med 2003;50:531–40. 17. Le Bihan D, Breton E, Lallemand D, Aubin ML, Vignaud J, Laval-Jeantet M. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 1988;168:497–505.

8. Nielles-Vallespin S, Khalique Z, Ferreira PF, et al. Assessment of myocardial microstructural dynamics by in vivo diffusion tensor cardiac magnetic resonance. J Am Coll Cardiol 2017;69: 661–76.

18. Moseley ME, Cohen Y, Kucharczyk J, et al. Diffusion-weighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology 1990;176:439–45.

10. Le Bihan D, Turner R, Moonen CTW, Pekar J. Imaging of diffusion and microcirculation with

imaging with second-order motion-compensated convex optimized diffusion encoding. Magn Reson Med 2018;80:1074–87.

12. Edelman RR, Gaa J, Wedeen VJ, et al. In vivo measurement of water diffusion in the human heart. Magn Reson Med 1994;32:423–8.

7. Spotnitz HM, Spotnitz WD, Cottrell TS, Spiro D, Sonnenblick EH. Cellular basis for volume related wall thickness changes in the rat left ventricle. J Mol Cell Cardiol 1974;6:317–31.

9. Stejskal EO, Tanner JE. Spin diffusion measurements: spin echoes in the presence of a timedependent field gradient. J Chem Phys 1965;42: 288–92.

21. Aliotta E, Moulin K, Magrath P, Ennis DB. Quantifying precision in cardiac diffusion tensor

19. Basser PJ, Mattiello J, Lebihan D. Estimation of the effective self-diffusion tensor from the NMR

tensor imaging in rats. IEEE Trans Med Imaging 2015;34:1843–53. 24. Scott AD, Nielles-Vallespin S, Ferreira PF, et al. An in-vivo comparison of stimulated-echo and motion compensated spin-echo sequences for 3 T diffusion tensor cardiovascular magnetic resonance at multiple cardiac phases. J Cardiovasc Magn Reson 2018;20:1. 25. Stoeck CT, von Deuster C, Fleischmann T, Lipiski M, Cesarovic N, Kozerke S. Direct comparison of in vivo versus postmortem second-order motion-compensated cardiac diffusion tensor imaging. Magn Reson Med 2018;79:2265–76. 26. Nguyen C, Fan Z, Xie Y, et al. In vivo contrast free chronic myocardial infarction characterization using diffusion-weighted cardiovascular magnetic resonance. J Cardiovasc Magn Reson 2014;16. 27. von Deuster C, Stoeck CT, Genet M, Atkinson D, Kozerke S. Spin echo versus stimulated echo diffusion tensor imaging of the in vivo human heart. Magn Reson Med 2016;76:862–72. 28. Reese TG, Weisskoff RM, Smith RN, Rosen BR, Dinsmore RE, Wedeen VJ. Imaging myocardial fi-

spin echo. J Magn Reson B 1994;103:247–54.

ber architecture in vivo with magnetic resonance. Magn Reson Med 1995;34:786–91.

20. Gamper U, Boesiger P, Kozerke S. Diffusion imaging of the in vivo heart using spin echoes– considerations on bulk motion sensitivity. Magn Reson Med 2007;57:331–7.

29. Ferreira PF, Kilner PJ, McGill L-A, et al. In vivo cardiovascular magnetic resonance diffusion tensor imaging shows evidence of abnormal myocardial laminar orientations and mobility in

JACC: CARDIOVASCULAR IMAGING, VOL.

-, NO. -, 2019

- 2019:-–-

hypertrophic cardiomyopathy. J Cardiovasc Magn Reson 2014;16:87. 30. von Deuster C, Sammut E, Asner L, et al. Studying dynamic myofiber aggregate reorientation in dilated cardiomyopathy using in vivo magnetic resonance diffusion tensor imaging. Circ Cardiovasc Imaging 2016;9:e005018. 31. Nielles-Vallespin S, Mekkaoui C, Gatehouse P, et al. In vivo diffusion tensor MRI of the human heart: reproducibility of breath-hold and navigator-based approaches. Magn Reson Med 2013;70:454–65. 32. Reese TG, Wedeen VJ, Weisskoff RM. Measuring diffusion in the presence of material strain. J Magn Reson B 1996;112:253–8.

42. Tseng W-YI, Wedeen VJ, Reese TG, Smith RN, Halpern EF. Diffusion tensor MRI of myocardial fibers and sheets: correspondence with visible cutface texture. J Magn Reson Imaging 2003;17: 31–42. 43. Chen J, Liu W, Zhang H, et al. Regional ventricular wall thickening reflects changes in cardiac fiber and sheet structure during contraction: quantification with diffusion tensor MRI. Am J Physiol Heart Circ Physiol 2005;289:H1898–907. 44. Nguyen C, Lu M, Fan Z, et al. Contrast-free detection of myocardial fibrosis in hypertrophic cardiomyopathy patients with diffusion-weighted cardiovascular magnetic resonance. J Cardiovasc Magn Reson 2015;17:107.

33. Dou J, Reese TG, Tseng W-YI, Wedeen VJ. Cardiac diffusion MRI without motion effects.

45. Ennis DB, Kindlmann G. Orthogonal tensor invariants and the analysis of diffusion tensor

Magn Reson Med 2002;48:105–14.

magnetic resonance images. Magn Reson Med 2006;55:136–46.

34. Tseng

W-YI,

Khalique et al. DT-CMR Clinical Perspective

Reese

TG,

Weisskoff

RM,

Brady TJ, Wedeen VJ. Cardiac diffusion tensor MRI in vivo without strain correction. Magn Reson Med 1999;216:393–403.

46. Jones DK, Cercignani M. Twenty-five pitfalls in the analysis of diffusion MRI data. NMR Biomed 2010;23:803–20.

35. Stoeck CT, Kalinowska A, von Deuster C, et al. Dual-phase cardiac diffusion tensor imaging with strain correction. PLoS One 2014;9:e107159.

47. Tunnicliffe EM, Scott AD, Ferreira P, et al. Intercentre reproducibility of cardiac apparent diffusion coefficient and fractional anisotropy in healthy volunteers. J Cardiovasc Magn Reson

36. Ferreira PF, Nielles-Vallespin S, Scott AD, et al. Evaluation of the impact of strain correction on the orientation of cardiac diffusion tensors with in vivo and ex vivo porcine hearts. Magn Reson Med 2017;79:2205–15. 37. Axel L, Wedeen VJ, Ennis DB. Probing dynamic myocardial microstructure with cardiac magnetic resonance diffusion tensor imaging. J Cardiovasc Magn Reson 2014;16:89. 38. Hales PW, Schneider JE, Burton RAB, Wright BJ, Bollensdorff C, Kohl P. Histo-anatomical structure of the living isolated rat heart in 2 contraction states assessed by diffusion tensor MRI. Prog Biophys Mol Biol 2012;110:319–30. 39. Scollan DF, Holmes A, Winslow R, Forder J. Histological validation of myocardial microstructure obtained from diffusion tensor cardiac magnetic resonance. Am J Physiol Heart Circ Physiol 1998;275:H2308–18. 40. Hsu EW, Muzikant AL, Matulevicius SA, Penland RC, Henriquez CS. Magnetic resonance myocardial fiber-orientation mapping with direct histological correlation. Am J Physiol Heart Circ Physiol 1998;274:H1627–34. 41. Holmes AA, Scollan Df, Winslow RL. Direct histological validation of diffusion tensor MRI in formaldehyde-fixed myocardium. Magn Reson Med 2000;44:157–61.

2014;16:31. 48. McGill L-A, Scott AD, Ferreira PF, et al. Heterogeneity of fractional anisotropy and mean diffusivity measurements by in vivo diffusion tensor imaging in normal human hearts. PLoS One 2015;10:e0132360. 49. Pope AJ, Sands GB, Smaill BH, LeGrice IJ. Three-dimensional transmural organization of perimysial collagen in the heart. Am J Physiol Heart Circ Physiol 2008;295:H1243–52. 50. Mekkaoui C, Nielles-Vallespin S, Gatehouse PD, Jackowski MP, Firmin DN, Sosnovik DE. Diffusion MRI tractography of the human heart in vivo at end-diastole and end-systole. J Cardiovasc Magn Reson 2012;14:O49. 51. McGill LA, Ferreira PF, Scott AD, et al. Relationship between cardiac diffusion tensor imaging

quantified with diffusion tensor MRI. Am J Physiol Heart Circ Physiol 2003;285:H946–54. 54. Winklhofer S, Stoeck CT, Berger N, et al. Postmortem cardiac diffusion tensor imaging: detection of myocardial infarction and remodeling of myofiber architecture. Eur Radiol 2014;24:2810–8. 55. Wu EX, Wu Y, Nicholls JM, et al. MR diffusion tensor imaging study of postinfarct myocardium structural remodeling in a porcine model. Magn Reson Med 2007;58:687–95. 56. Wu M-T, Su M-YM, Huang Y-L, et al. Sequential changes of myocardial microstructure in patients postmyocardial infarction by diffusiontensor cardiac MR: correlation with left ventricular structure and function. Circ Cardiovasc Imaging 2009;2:32–40. 57. Mekkaoui C, Jackowski MP, Kostis WJ, et al. Myocardial scar delineation using diffusion tensor magnetic resonance tractography. J Am Heart Assoc 2018;7:e007834. 58. Tseng W-YI, Dou J, Reese TG, Wedeen VJ. Imaging myocardial fiber disarray and intramural strain hypokinesis in hypertrophic cardiomyopathy with MRI. J Magn Reson Imaging 2006;23:1–8. 59. McGill L-A, Ismail TF, Nielles-Vallespin S, et al. Reproducibility of in-vivo diffusion tensor cardiovascular magnetic resonance in hypertrophic cardiomyopathy. J Cardiovasc Magn Reson 2012;14: 86. 60. Ariga R, Tunnicliffe EM, Manohar SG, et al. Non-invasive imaging of myocardial disarray associates with ventricular arrhythmia in hypertrophic cardiomyopathy. Eur Heart J 2017;38. 61. Khalique Z, Ferreira PF, Scott AD, et al. Diffusion tensor cardiovascular magnetic resonance of microstructural recovery in dilated cardiomyopathy. J Am Coll Cardiol Img 2018;11: 1548–50. 62. Khalique Z, Ferreira PF, Scott AD, et al. Deranged myocyte microstructure in situs inversus totalis demonstrated by diffusion tensor cardiac magnetic resonance. J Am Coll Cardiol Img 2018; 11:1360–2.

parameters and anthropometrics in healthy volunteers. J Cardiovasc Magn Reson 2016;18. 52. Wu M-T, Tseng W-YI, Su M-YM, et al. Diffusion tensor cardiac magnetic resonance mapping the fiber architecture remodeling in human myocardium after infarction: correlation with viability and wall motion. Circulation 2006;114:1036–45. 53. Chen J, Song S-K, Liu W, et al. Remodeling of cardiac fiber structure after infarction in rats

KEY WORDS cardiomyocytes, diffusion tensor CMR, helix angle, microstructure, sheetlets

A PPE NDI X For supplemental references, a table, and videos, please see the online version of this paper.

21