RESEARCH ARTICLE

Nonivasive quantification of axon radii

using diffusion MRI
Jelle Veraart1,2,3*, Daniel Nunes1, Umesh Rudrapatna4, Els Fieremans2,
Derek K Jones4,5, Dmitry S Novikov2†, Noam Shemesh1†
1
Champalimaud Research, Champalimaud Centre for the Unknown, Lisbon,
Portugal; 2Center for Biomedical Imaging, Department of Radiology, New York
University School of Medicine, New York, United States; 3imec-Vision Lab,
Department of Physics, University of Antwerp, Antwerp, Belgium; 4CUBRIC, School
of Psychology, Cardiff University, Cardiff, United Kingdom; 5Mary MacKillop
Institute for Health Research, Australian Catholic University, Melbourne, Australia

Abstract Axon caliber plays a crucial role in determining conduction velocity and, consequently,
in the timing and synchronization of neural activation. Noninvasive measurement of axon radii could
have significant impact on the understanding of healthy and diseased neural processes. Until now,
accurate axon radius mapping has eluded in vivo neuroimaging, mainly due to a lack of sensitivity
of the MRI signal to micron-sized axons. Here, we show how – when confounding factors such as
extra-axonal water and axonal orientation dispersion are eliminated – heavily diffusion-weighted
MRI signals become sensitive to axon radii. However, diffusion MRI is only capable of estimating a
single metric, the effective radius, representing the entire axon radius distribution within a voxel
that emphasizes the larger axons. Our findings, both in rodents and humans, enable noninvasive
mapping of critical information on axon radii, as well as resolve the long-standing debate on
whether axon radii can be quantified.

*For correspondence:
[email protected]
†
These authors contributed
Introduction
equally to this work Axons facilitate connectivity between distant neurons. Along with myelination, the axon radius deter-
mines the conduction velocity, thereby shaping the timing of neuronal computations and communi-
Competing interests: The
cation (Waxman, 1980). Using a model of action potential neurophysiology (Rushton, 1951), it has
authors declare that no
been shown that the axon radius explains the largest proportion of variance in conduction velocity
competing interests exist.
(Drakesmith et al., 2019). Histological studies demonstrated that axon sizes vary widely within the
Funding: See page 22 human brain, ranging from 0.1 m to more than 3 m (Aboitiz et al., 1992; Innocenti et al., 2015;
Received: 02 July 2019 Liewald et al., 2014), and across species (Olivares et al., 2001; Schüz and Preibl, 1996;
Accepted: 07 January 2020 Liewald et al., 2014). Moreover, axon radii have been shown to be altered in various disease pro-
Published: 12 February 2020 cesses. For example, direct axon counting in post-mortem tissue has suggested that smaller axons
may be preferentially susceptible to axonal injury in multiple sclerosis (Evangelou et al., 2001) due
Reviewing editor: Birte
Forstmann, University of
to inflammation (Campbell et al., 2014). Electron microscopy has revealed a higher percentage of
Amsterdam, Netherlands small-radius axons and a lower percentage of large-radius axons in several anatomically and func-
tionally distinct segments of the corpus callosum in autistic subjects compared to healthy controls
Copyright Veraart et al. This
(Wegiel et al., 2018). From the animal literature, morphometric analysis of adult rat brains showed
article is distributed under the
reduced axonal radii without axonal loss after chronic alcohol feeding (Kjellström and Conradi,
terms of the Creative Commons
Attribution License, which 1993). Such studies indicate that non-invasive metrics capable of reporting on features of the axon
permits unrestricted use and radius distribution could provide important neuroimaging biomarkers for basic research and clinical
redistribution provided that the applications.
original author and source are A particularly relevant neuroimaging modality attuned to the microarchitecture of living brain tis-
credited. sue is diffusion-weighted MRI (dMRI). dMRI is sensitive to the thermal motion of water molecules

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 1 of 27

 Research article Neuroscience

and their interference with microscopic boundaries, such as imparted by cells and subcellular struc-
tures in the brain (Tanner, 1979; Le Bihan, 2003; Le Bihan et al., 1986; Callaghan et al., 1988;
Basser et al., 1994; Jones, 2010; Beaulieu, 2002; Novikov et al., 2019). Applications of dMRI spe-
cialize in revealing macroscopic brain connections (Jbabdi et al., 2015) and in the interpretation of
contrast differences in diffusion-weighted images (Moseley et al., 1990; Baron et al., 2015). How-
ever, reproducible and specific biomarkers for studying disease onset and progression non-invasively
and quantitatively in the entire brain, in particular vis-a-vis axonal properties, would confer clear
advantages. Several studies have used various methods to report on axon radius parameters; still,
despite many attempts, axon radius mapping using dMRI remains highly contested (Assaf et al.,
2013; Horowitz et al., 2015; Alexander et al., 2010; Innocenti et al., 2015; Xu et al., 2014;
Burcaw et al., 2015; Ong et al., 2008; Ong and Wehrli, 2010). Discrepancies between histology
and dMRI-derived axon radii uncovered various confounding factors, for example orientation disper-
sion (Drobnjak et al., 2016; Nilsson et al., 2012), time-dependent extra-axonal diffusion oversha-
dowing the intra-axonal signal at low diffusion weighting (Burcaw et al., 2015; Fieremans et al.,
2016; Lee et al., 2018), weak signal attenuation for typically very narrow axons, especially in the
realistic experimental regime of long diffusion gradient duration (van Gelderen et al., 1994; Neu-
man, 1974), and/or putative shrinkage during tissue preparation (Barazany et al., 2009;
Innocenti et al., 2015; Aboitiz et al., 1992).
Recent advances in biophysical modeling and hardware prompted a revival of MR axon radius
mapping (McNab et al., 2013; Huang et al., 2015; Jones et al., 2018). First, several of the most
crucial confounding factors have been removed using powder-averaging concepts (Callaghan et al.,
1979; Jespersen et al., 2013; Kaden et al., 2016). Averaging diffusion-weighted signals that are
isotropically distributed on a sphere with constant diffusion-weighting strength b has been shown to
factor out the orientation dispersion (Jespersen et al., 2013; Kaden et al., 2016; Mollink et al.,
2017), thereby eliminating one of the most important confounding factors in axon radius mapping
(Nilsson et al., 2012). Second, gradient systems capable of producing relatively strong gradient
pulses have been introduced in human scanners (Jones et al., 2018). Third, it has been shown that
dMRI can be made specific to a particular water population restricted into long, yet micrometer-thin
cylindrical objects by imparting high diffusion-weighting regimes (McKinnon et al., 2017;
Veraart et al., 2019). Often, an axon is too narrow to yield a measurable diffusion-weighted MR sig-
nal decay, hence the popular use of ‘sticks’ (Behrens et al., 2003; Kroenke et al., 2004) when refer-
ring to axons (and possibly glial cell processes) within the context of biophysical modeling of white
matter using dMRI.
The intuition behind promoting specificity to intra-axonal water comes from Callaghan’s
model (Callaghan et al., 1979) of diffusion inside infinitely narrow one-dimensional randomly-ori-
ented cylinders, as applied to intra-neurite diffusion by Kroenke et al. (2004). The spatial Four-
k 2
ier transform e Da ðq^nÞ t of the diffusion propagator (with respect to the diffusion wave vector q)
for a single stick as measured with MRI (Callaghan, 1991), averaged over the orientations n ^ of
k
Da q2 t cos2  a
R
the sticks, yields the asymptotic scale-invariant power law S ¼ d cos  e ~ 1=b as a func-
tion of the diffusion weighting parameter b ¼ q2 t (Le Bihan et al., 1986), with the scaling expo-
nent a ¼ 1=2. Evidently, this power law scaling should be only approximate, for q  1=r, where r
is the cylinder radius. Its observation (McKinnon et al., 2017; Veraart et al., 2019) in the range
6 ms=m2  b  10 ms=m2 is a manifestation of the insensitivity of dMRI to the transverse axonal
dimensions on clinical scanners. However, for sufficiently strong diffusion weighting, the power
law scaling eventually breaks down, and the dMRI measurement becomes sensitive to the axonal
diameter.
Technically, this work addresses the detection and the interpretation of the deviation from the
radius-insensitive a ¼ 1=2 power law signal behavior at the largest possible b (by varying q at fixed
diffusion time t), in rat and human white matter. Indeed, either sensitivity of MR to a finite axonal
radius, or a notable exchange rate between intra- and extra-axonal water at the clinical dMRI time
scales t~100 ms, would alter the very particular power law scaling (Kroenke et al., 2004;
Jensen et al., 2016; McKinnon et al., 2017; Veraart et al., 2019).
Following theoretical considerations, we demonstrate the breaking of the power law scaling at
very high b-values in ex vivo rodent brains, from which metrics associated with the axon radius distri-
bution can be mapped quantitatively. Confocal microscopy of the rat corpus callosum (CC) validated

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 2 of 27

 Research article Neuroscience

that (i) the signal arises mainly from the intra-axonal space, and (ii) the MR-derived axon radius met-
rics are in good quantitative agreement with those derived from histology. We then observe the
same signal signatures in living human brain on the Connectom 3T scanner, that is, a high perfor-
mance research scanner with a maximal gradient amplitude of 300 mT/m – a fourfold increase com-
pared to state-of-the art clinical scanners (Glasser et al., 2016). Our findings both validate the
mechanism with which axon radii are weighted in dMRI (Burcaw et al., 2015), and demonstrate the
accuracy of which properties of the radius distributions can be estimated. After validating and evalu-
ating our methodology in rat and human brain, we further discuss the impact of axon radius meas-
urements in health and disease.

Theory
Power law scaling
In most biophysical models for diffusion in white matter, axons (and possibly glial cell processes) are
represented by zero-radius impermeable ‘sticks’, characterized by locally one-dimensional diffusion,
that is radial intra-axonal diffusivity D?a  0 (Kroenke et al., 2004; Behrens et al., 2003;
Jespersen et al., 2007; Jespersen et al., 2010; Fieremans et al., 2011; Sotiropoulos et al., 2012;
Zhang et al., 2012; Novikov et al., 2014; Novikov et al., 2018; Novikov et al., 2019;
Jensen et al., 2016; Reisert et al., 2017; McKinnon et al., 2017; Veraart et al., 2019). The stick
model then yields an asymptotic intra-axonal orientationally averaged signal decay,

SðbÞ ’ b b a
þ g ; bDka  1 ; (1)

with an intercept g (discussed below), the power law exponent a ¼ 1=2, and the coefficient
b ¼ p=4
f =ðDka Þ1=2 where f is the T2 -weighted axonal water fraction (Veraart et al., 2018;
pffiffiffiffiffiffiffiffiffi

Lampinen et al., 2019) and Dka the parallel intra-axonal diffusivity. This particular signal decay only
holds in the absence of extra-axonal signal, which is assumed to decay exponentially fast and, as
such, to be fully suppressed at sufficiently high b-values (McKinnon et al., 2017; Veraart et al.,
2019). Therefore, we restrict our in vivo analysis to b>6 ms=m2 (Veraart et al., 2019). Our lower
bound on the b-value is significantly higher than previous predictions from Monte Carlo simulations
(Raffelt et al., 2012), thereby minimizing the likelihood of residual extra-axonal signal contributions.
For the ex vivo analysis, we increase this lower bound to b ¼ 20 ms=m2 to compensate for the
reduced diffusivities in fixed tissue (Shepherd et al., 2009).

Breaking of the power law

The following computations always assume that the signal is normalized to Sjb¼0  1. Sensitivity of
MR to either finite axon radius or notable exchange rate between intra- and extra-cellular water
would break the b 1=2 -scaling at large b as follows:
. Finite axon radius: A finite D?
a > 0 results in a truncated power law:

bD? 2
a þOðb Þ 1=2
SðbÞ ’ b e b þ fim ; (2)

with fim  Sjb!¥  0 the signal fraction of a fully restricted immobile water compartment (the so-
called ‘dot’ compartment) (Stanisz et al., 1997; Dhital et al., 2018; Tax et al., 2019;
Veraart et al., 2019). If D? a ¼ 0, the power law offset (found from extrapolating the signal to
b ! ¥), should give the fraction of the dot compartment, g  fim (Veraart et al., 2019). However,
for nonzero D? a and finite b, the Taylor expansion of,

D? 2 pffiffiffi
S ’ b
e a = þ fim ;  ¼ 1= b (3)

around any finite point 0 predicts the  ! 0 intercept g<fim , Figure 1. The always negative differ-
ence  ¼ g fim <0 depends on b, Da , and 0 ; its maximal magnitude
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi p k 2 ?
jmax j ¼ b 2D? a =e ¼ f

?
2e
Da =Da is achieved at the curve’s inflection point  ¼ 2Da . Hence,

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 3 of 27

 Research article Neuroscience

fim
fim

Clinical ǫ
Research (human)
γ
Research (Small animal)

Figure 1. Breakdown of power law scaling: Top left: A nonzero D? a would result in a truncated power-law signal
decay. Although the resulting signal nonlinearity might be too subtle to be discerned within the achievable b-
range, even for (pre-)clinical systems with strong diffusion-weighting gradients, the concavity of the curves plotted
pffiffiffi pffiffiffiffiffiffiffiffiffi ?
as function of  ¼ 1= b for > ¼ 2D? a means that even the smallest Da will result in an extrapolated  ! 0
intercept g<fim when the power law, Equation 1, is used to approximately describe the signal in the delineated b-
ranges. The intercept is maximally negative at the inflection point  (colored dots), beyond which each curve
becomes convex, and the negative intercept g of the linear extrapolation starts to decrease. In all plots here,
diffusivities and b-values are expressed in mm2/ms and ms/mm2, respectively. Top right: One representative curve
(D?a ¼ 0:020) is shown to highlight the differences between the physically plausible dot compartment fim >0, and
the intercept g. The dot compartment is a positive signal fraction of a biophysical compartment, whereas the
intercept is a parameter of the power-law approximation, Equation 1. Their difference  depends on various
parameters, including the axonal signal fraction, diffusivities, the axon radius, and the scan protocol. The predicted
signal decay for the exchange model (dash-dotted; Equation 4) is convex in the entire b-range, where the signal
decay for the finite axon radius model (dotted; Equation 2) is concave until the inflection point. Bottom: The
optimization landscape of Equation 3 shows a shallow valley, relative to the noise floor, for a simulation that
mimics the human component of the study. (Bottom left) The valley is shown in a 2D projection of the landscape
(shown as a function of radius instead of D? a , see Equation 10). (Bottom right) The fit objective function along the
valley is shown (red line) in comparison to the noise floor (dashed line) with an unrealistically high SNR of 250 for
the non-DW signal. The red dot indicates the ground truth value.

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 4 of 27

 Research article Neuroscience

the lower bound fim jmax j for the  ! 0 intercept g may be negative. A negative g is biophysi-
cally implausible if the stick model holds, D? a  0; however, g<0 becomes a natural consequence
of a finite D?
a >0 (and hence, of a finite axonal diameter), Figure 1, in the case when the extrapo-
lated negative intercept overcomes the positive immobile fraction fim . Recently, fim was shown to
be negligible in healthy human white matter (Dhital et al., 2018; Tax et al., 2019; Veraart et al.,
2019). Therefore, a negative intercept is a novel hallmark of MR sensitivity to the inner axon
diameter, even if the signal scaling might appear linear as a function of b 1=2 for b-ranges accessi-
ble on human MR scanners. Importantly, the finite axon radius model, Equation 3, is poorly con-
ditioned as a result of which the simultaneous estimation of fim and D? a is practically impossible,
especially for human MR experiments, see Figure 1. An accurate and precise measurement of D? a
depends on the prior knowledge of fim and requires a dedicated measurement (Dhital et al.,
2018; Tax et al., 2019).

. Exchange: The spherical integration of the two-compartment ‘Kärger’ model (Kärger, 1985)
with a finite exchange rate R>0 yields approximately the following signal decay:
 
SðbÞ » b b 1=2 þ c
b 3=2 þ fim ; c / RTE =D?
e >0 ; (4)

with D?
e the radial diffusivity in the interstitial space, and TE , the echo time, during which
exchange can happen. Importantly, Equation 4 is convex as a function of  ¼ b 1=2 .
The relative fit quality of the models (i.e., Equations 1, 2, and 4) to the dMRI signal decays can
be evaluated qualitatively (convex versus concave shape) or statistically by means of the corrected
Akaike information criterion (AICc) (Burnham and Anderson, 2002).

From diffusivity to effective MR radius

The radial signal attenuation S? c ðrÞ inside the cylinder of radius r in the Gaussian phase approxima-
tion (van Gelderen et al., 1994):
¥ h i
2g2 r 4 P tc 2a2m d=tc 2a2m D=tc 2a2m ðD dÞ=tc 2a2m ðDþdÞ=tc
ln S?
c ðrÞ ¼ D0 6 2
a ða 1Þ
2a 2 d
m tc 2 þ 2e þ 2e e e þ Oðg4 Þ
m¼1 m m (5)
 bD? a ðrÞ þ Oðb 2
Þ;

with b ¼ g2 d2 ðD d=3Þ and tc ¼ r 2 =D0 defines the connection between the intra-axonal radial diffusiv-
ity D?a and the radius r. Here, D0 is the diffusivity of the axoplasm, g the gradient of the Larmor fre-
quency, am is the mth root of dJ1 ðaÞ=da ¼ 0, where J1 ðaÞ is the Bessel function of the first kind, and d
and D are the gradient duration and separation, respectively (Stejskal, 1965).
In the long-pulse limit, that is when d  tc , the dependence on D drops out (Neuman, 1974), and
Equation 5 approaches the Neuman’s limit

7 g2 d
ln S? 4
c ðrÞ ¼ k r ; k ¼ ; d  tc : (6)
48 D0

This limit practically applies to the majority of axons. Importantly, the attenuation is proportional to
the fourth power of the radius r and, as such, it is very weak for narrow axons. Hence the low sensi-
tivity of dMRI to the inner axon diameter.
For an unknown distribution hðrÞ of axons with radii r, the total intra-axonal signal attenuation
becomes a volume-average over the histogram bins ri (Packer and Rees, 1972):

hðri Þr 2 S? ðri Þ hr 2 S?
P
c ðrÞi
S? ½hðrފ ’ iP i c 2 ¼ ; (7)
i hðri Þri hr 2 i

such that the signal contribution of an axon scales quadratically with its radius r. The Taylor expan-
sion of the net signal attenuation S? demonstrates the sensitivity of the dMRI signal to the distribu-
tion’s higher order moments:

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 5 of 27

 Research article Neuroscience

S? ½hðrފ ¼ hr 2 1 kr 4 þ Oðr 8 Þ i=hr 2 i » 1 khr 6 i=hr 2 i

4
(8)
»e kreff
 S?
c ðreff Þ;

where the effective axon radius:

1=4
reff  hr 6 i=hr 2 i (9)

captures the contribution from the whole distribution hðrÞ in a single metric (Burcaw et al., 2015). The
ability to represent the whole distribution by the ratio of its 6th and 2nd moments relies on almost all
bD?
axons falling into the Neuman’s limit, Equation 6. Representing S?
c ðreff Þ  e
a , we can calculate
 1=4
48
rMR ¼ dðD d=3ÞD0 D?
a ; (10)
7

the MRI estimate of reff after estimating D?

a from the orientation-averaged signal using Equation 2.
Note that the effective radius, Equation 9, is heavily weighted by the tail of hðrÞ. Physically, this
happens due to the combination of the weak NMR signal attenuation by small radii, ln S ~ r4 , in
the diffusion-narrowing (Neuman’s) regime (Neuman, 1974), and of the subsequent volume-weight-
ing that emphasizes the thickest axons by an extra factor of r 2 (Packer and Rees, 1972;
Alexander et al., 2010). The error associated with these modeling assumptions is discussed in the
Results section.

Results
Simulations
Accuracy
First, we evaluate the accuracy of axon radius mapping as a function of r for axon radius distributions
extracted from histology; Figure 2 (left and middle panels). We used a simulation framework based
on the matrix formalism for diffusion signal attenuation within fully restricted cylinders (Calla-
ghan, 1997), as implemented in the MISST toolbox (Drobnjak et al., 2010), while mimicking the
entire experimental setup, for both the human and preclinical experiments.
In the case of diffusion restricted in a single cylinder with radius r, the error in the estimated
radius ^r increases with r. Indeed, the missing higher-order Oðg4 Þ corrections to Equations (5)-(6) set
an upper bound on the achievable accuracy for large axons, as estimated recently (Lee et al., 2018).
The combined error in the estimation of reff associated with the approximations made in Equa-
tion 6 and Equation 8 is only 5% for the human set-up when considering the axon radius distribution
provided by Aboitiz et al. (1992); the distributions of Caminiti et al. (2009) result in a subpercent
error. Additionally, we show the errors in the estimation of reff for the axon caliber distributions that
were observed in our different histological sections while considering the scan parameters from our
fixed tissue experiments. The shorter diffusion timings increase the approximation errors, leading to
an underestimation up to 9%.

Feasibility and precision

Figure 2 (right panel) shows a theoretical lower bound on the 95% confidence interval in the voxelwise
estimation of D?a from Equation 2, as predicted using a Cramér-Rao lower bound analysis (Kay, 1993).
Using the dependence D? ?
a » Da ðreff Þ, Equation 8, that approximately identifies reff with the single cylin-
der radius in van Gelderen’s model, can be used to translate the lower bound on D?
a to that on reff .
?
Notably, it follows from Figure 2, that an estimate of Da exceeds zero with a statistical threshold
of p>0:05, if the corresponding reff >1:41 m and reff >0:76 m, when mimicking the diffusion acquisi-
tion and SNR on the Siemens Connectom (Gmax ¼ 300 mT=m) and Bruker Aeon (Gmax ¼ 1500 mT=m)
MR scanners, respectively. In comparison, for a typical acquisition on a modern clinical scanner with
Gmax ¼ 80 mT=m, this lower bound is 3.2 m (Veraart et al., 2019).

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 6 of 27

 Research article Neuroscience

Human Preclinical
4 4 0.04

80 mT/m
0.035
3.5 3.5 300 mT/m
0.03 1500 mT/m
3 3
0.025

2.5 2.5
0.02

Da⊥
2 2

r̂
r̂

0.015

0.01
1.5 1.5

0.005
1 1
0

0.5 0.5
-0.005

0 0 -0.01
0 1 2 3 4 0 1 2 3 4 0.5 1 1.5 2 2.5 3 3.5 4
r r r

Figure 2. Simulations on accuracy and precision of MR-based axon radius mapping. First, the left and middle panel show the difference between the
estimated, ^r, and theoretical, r, effective MR radius associated with various realistic axon caliber distributions (solid dots with different color for different
distributions) for the clinical and preclinical setups, respectively. Axon caliber distributions were adopted from Aboitiz et al. (1992) and
Innocenti et al. (2015) for the clinical simulations (see Figure 7), whereas various axon distributions (see Figure 4) derived from our own histology
were used for the preclinical simulation. The average radii, r , of the axon caliber distribution are shown for comparison (open dots). Additionally, the
accuracy of the framework for a system with single cylinder with radius r is shown (black line). Second (right figure), the feasibility to measure D? a with
statistical significance in case of scan settings and SNR for the Connectom (300 mT/m; blue), Aeon (1500 mT/m; green) protocol, respectively. For
comparison, we also assessed the feasibility for the Prisma protocol as described in Veraart et al. (2019) (80mT/m; red). The shaded areas illustrate
the 95% confidence intervals derived from Cramér-Rao lower bound analysis of model, Equation 2 with fim ¼ 0. The corresponding minimal cylinder
radius r that allows for the observation of significant D? a ðrÞ, r ¼ 0:76 m, 1.41 mm and 3.24 mm for Aeon, Connectom, and Prisma, respectively, is
indicated by the vertical lines. In all plots, diffusivities and radii are expressed in m2 =ms and m, respectively.

Preclinical data
Dot compartment
Because fim has been reported to be significant in fixed tissue by Stanisz et al. (1997), we first have
to estimate the signal fraction of the immobile water compartment fim in the three fixed brain sam-
ples from a dedicated MRI acquisition (see Materials and methods). The estimate ^fim is in the range
of 8 to 17% with a median value of 13%. The range is defined here by the 5th and 95th percentile of
the distribution of the estimated dot fractions in all CC voxels, across the three samples.

Breaking the power law

Figure 3 shows the signal decay, averaged across all CC voxels, based on diffusion measurements in
the three rat brain samples with b up to 100 ms=m2 . We notice that an extrapolation to infinite b,
pffiffiffi
that is 1= b ! 0, yields a small but significant negative offset g, of the order of a few per cent of the
non-attenuated Sj signal, in all three samples after subtracting the ^fim from the diffusion
b¼0
measurements.
We re-evaluate the validity of a perfect stick assumption in the high b-regimes using a AIC analy-
sis. To study fit robustness with respect to the number of degrees of freedom by considering the
full, nested, and extended models to Equation 1. Specifically, we evaluated the following models:
i. fim þ bb a
ii. fim þ bb 1=2 ;
?
iii. fim þ be bDa b 1=2
1=2 3=2


iv. fim þ b b þc
b
v. bb a
vi. bb 1=2
?
vii. be bDa b 1=2


viii. b b 1=2 þ c
b 3=2

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 7 of 27

 Research article Neuroscience

Double-log scale
(a) Preclinical

Double-log scale

(b) Human

Figure 3. Breaking of the power law. The ROI- and spherically averaged signal decay is shows for the different
pffiffiffi
fixed samples of the rat CC (a) and human subjects (b) and as a function of 1= b and on a double logarithmic
scale. The data deviate from the power law scaling with exponent 1/2 that is predicted by the stick model (i.e.
nonlinear signal decay in log-log plot), thereby demonstrating sensitivity of the signal to the radial intra-axonal
signal. The fits of Equation 1 are shown in dashed lines. In all plots, b is expressed in ms=m2 .

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 8 of 27

 Research article Neuroscience

Our analysis shows that a truncated power law (vii), which explicitly accounts for D? a >0 (and hence
does not require a negative intercept parameter), yet sets fim ¼ 0, fits the experimental data signifi-
cantly better than pure power law forms (models (i), (ii), (v), and (vi)) (the difference in AICc<2), with
or without an offset g, if the immobile (dot) compartment is corrected for, that is when using
S* ðbÞ ¼ SðbÞ ^fim . Without dot-correcting the ex vivo data, the power law form (ii) with an intercept
outperforms the other models. In that case, the intercept g is negative, while fim is defined to be
positive. Hence, the intercept encodes both the still water fraction and a negative offset to the inter-
cept associated with the sensitivity to the axon diameter, such that the overall g<0.

Axon radius estimation and histological validation

Axon radii were estimated from the diffusion MRI data for the different CC ROIs (Figure 4) along
with the axon radius distributions extracted histologically, Figure 4. The errors between the associ-
ated tail-weighted effective radii and MR-derived rMR vary between 5 and 21% in the different ROIs.
Notably, a consistent residual overestimation was observed, whereas the previous simulations (Fig-
ure 2) predicted an underestimation.
To further examine the correlations between the MR-derived parameters and underlying micro-
structure, we analyzed 16 patches with the same size as an imaging voxel, that is 100  100 m2
within the genu of the CC of the second sample, CC#2. The confocal light microscopy images of
two of those patches are shown for the various stainings, Figure 5. Two notable biological compo-
nents other than axons were highlighted, namely, astrocytic processes and (neuronal or glial) cell
bodies, which were found to have volume fractions of about 5%. Although the radius of the astro-
cytic processes cannot be measured due to their random orientations w.r.t. the image plane, it is
clear that some of the processes have a large diameter, for example up to 7 m in the first patch.
The average cell body radius is 2.6 m, with an effective radius of 4.3 m. It is worth highlighting
that the T2 -weighted signal fraction of both cell types remains unknown since the corresponding

ROI #3 ROI #4
ROI #2 Splenium
i

ROI #1

Genu

ROI #1 ROI #2 ROI #3 ROI #4

Axon radius distribution (microscopy)

Effective axon radius (microscopy)
Effective axon radius (dMRI)

Figure 4. Histological validation, part I. The axon radius distributions for different ROIs of rat CC#1 are shown
(blue bars).The associated tail-weighted effective radii are shown in the black lines, whereas the corresponding MR
estimates are shown by the red lines. In all plots, r is expressed in m.

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 9 of 27

 Research article Neuroscience

relaxation times are unknown. This unknown difference between compartmental volume (histology)
and signal (MRI) fractions remains the Achilles’s heel of comparisons between MRI measurements
and histological evaluations of tissue microstructure.
Within each of the 16 patches, we extracted the axon radii distribution and derived the average r
and effective radius reff . The box plots of those metrics are shown in Figure 5. The median r and
median effective radius reff across all patches are 0.61 and 1.06 m , respectively. In comparison, the
median rMR , derived from dMRI in 16 voxels within the genu of the CC, is 1.16, 1.10, and
1.19 m for the three rat samples, respectively. The median MR-derived effective axon radius is
between 81 and 97% larger than the median r , whereas the error to the median reff , as derived from
histology, is only 3.4 to 12.8%.

Parameter maps
ROI measurements provided robust estimation, but a remaining question is whether dMRI could be
used to map the effective MR radius in a voxelwise manner. Figure 6 shows the maps of the MR-
derived effective axon radii for all three rat CC’s. The maps appear smooth with very few outlier vox-
els, suggesting that the estimation is robust even when voxelwise data is used. Furthermore, the
qualitative trends are in good agreement with previously reported observations of larger axons in
the body of the CC in comparison to the genu and splenium (Barazany et al., 2009). Inter-subject
variability is not very large and can be attributed to slightly different slice positions. When comput-
ing the effective radius of the CC-averaged signal rMR , the intersubject variability nearly nullifies.
Indeed, we estimate rMR ¼ 1:22; 1:25 and 1.25 mm in the three samples, respectively.

Towards human applicability

Breaking the power law
To assess the applicability of this approach in more realistic settings available for human imaging,
experiments were performed in human subjects on the Connectom scanner, which is capable of pro-
ducing 300 mT/m gradient amplitudes. The dMRI signal decay curves, averaged across all WM
radii [µm]

20µm 20µm
Astrocytes
Cell bodies
Axons

Figure 5. Histological validation, part II. (left) For a second fixed brain sample, CC#2, the confocal microscopy images, stained for neurofilaments (red),
astrocytes (green), and cell bodies (blue), are shown for two representative 100  100 m2 -patches that are positioned within the Genu (microscopy
image of CC shown for ROI positioning). The abundance of astrocytes and cell bodies, both representing 5% of the volume, is clear in both patches.
The astrocytic processes can have a large diameter, up to 7 mm in the first patch. A detailed analysis of the radius distribution of the astrocytic
processes is not possible due to their random orientation w.r.t. the image plane. (middle) Axon radius distributions for all 16 patches of the Genu (each
patch has different color in the bar plot). (right) Boxplots represent the distribution of the average and effective radius of the axon radii distribution that
were extracted from each of the 16 patches within the genu. The effective radius reff is larger than the average r , respective medians are 1.06 and
0.61 mm. The boxplots for the MR-derived axon radius measurements for 16 MR voxels within the genu for the three fixed CC samples are also
shown. In all plots, radii are expressed in m.

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 10 of 27

 Research article Neuroscience

voxels, with b-values up to 25 ms/mm2 for the

CC#1 four human subjects are shown in Figure 3.
Importantly, we find that — in excellent corre-
spondence with the previous preclinical data —
the linear extrapolation of the signal decay as a
pffiffiffi pffiffiffi
CC#2 function of 1= b to 1= b ! 0 produces a signifi-
cant negative offset g in all subjects.
Note that the dot compartment was not mea-
sured directly, because previous dedicated stud-
CC#3 ies revealed a negligible dot compartment,
that is fim ¼ 0, in the healthy white matter
(Dhital et al., 2018; Veraart et al., 2019;
Tax et al., 2019); see Discussion.
0.5 2 µm The AICc analysis of various models demon-
strated that also for the human white matter, the
truncated power law (vii) with D? a >0 and negligi-

Figure 6. Effective radii in the CC. Maps of the ble dot compartment fim ¼ 0 fits the data signifi-
effective radii derived from diffusion MR data, for the 3 cantly better than pure power law forms, with or
samples of the rat CC. without intercept. However, this statistical analy-
sis cannot be interpreted as a data-driven justifi-
cation for fim ¼ 0 because of the degeneracy of
Equation 3, as highlighted in Figure 1.

Comparison with histology

Since direct histological evaluation in volunteers is unfeasible, we turn to validate the MR-derived
metrics in humans with previously reported literature of human corpus callosum microstructure
(Aboitiz et al., 1992; Innocenti et al., 2015). In Figure 7, the MR-derived metrics were directly
compared with axon radius distributions of multiple histological studies (Aboitiz et al., 1992;
Innocenti et al., 2015). Various reports and histological studies show a good correspondence for
the bulk of the distributions, represented by the average radius r , that is the average radius r only
ranges between 0.54 and 0.69 mm. In histological samples, the corresponding effective radius reff
dominated by large axons, shows strong variability. Indeed, compared to r , reff varies more than
three-fold, from 0.91 to 2.9 m.
The four human subjects show good correspondence in terms of rMR . In Figure 7, we show the
individual and combined distributions describing rMR for all voxels in the midbody of the CC for all
four subjects. It is apparent that the combined distribution falls almost entirely within the range
spanned by reff -values as predicted from histology – even without introducing a putative axonal
shrinkage factor (maximally 35% [Aboitiz et al., 1992], and typically within 10% [Tang et al., 1997]).

Parameter distribution and maps

In Figure 8, the distribution and map of D? a for WM voxels in all human subjects, estimated using
the ODF-independent model, Equation 2 with fim ¼ 0, are shown. Considering the statistical bound
from Figure 2, it is to be expected that the estimated D? a is biophysically meaningful for the vast
majority of WM voxels for the Connectom scanner (Figure 8 shows a representative map of D? a and
associated rMR for a single subject of the Connectom cohort), whereas similar measurements on a
modern clinical scanner result in a biophysically implausible negative D?a in approximately 35% of all
WM voxels. Note that the data from a clinical scanner (Siemens Prisma with 80 mT/m gradients) are
adopted from our recent work (Veraart et al., 2019). This suggests that estimating D? a and the asso-
ciated effective axonal radius rMR is only possible on MR systems with ultra-strong gradients
(Jones et al., 2018; Huang et al., 2015). The spatial variability as well as the observed asymmetry
between the hemispheres in, for example, the occipital lobes was noted for all subjects. However,
our cohort is too small and not sufficiently characterized to study the whole brain characterization or
the role of lateralization in large axons of human brain (Eichert et al., 2019; Liewald et al., 2014;
Lebel and Beaulieu, 2009).

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 11 of 27

 Research article Neuroscience

r̄ reff

r̄MR

rMR (MRI)
r̄MR (MRI) rMR
r̄ (Histology)
reff (Histology)

rMR

rMR

rMR rMR

Figure 7. Comparing the effective radius from histology and in vivo dMRI. (top - histology) Axon radius
distributions of multiple histological studies and human CC samples show a good correspondence for the bulk of
the distribution, represented by the average radius r (dashed-dotted lines). Due to mesoscopic fluctuations of the
large axons in histological samples, the corresponding effective radius reff dominated by large axons, shows
strong variability (dotted lines). (bottom - MRI) The four Connectom subjects show good correspondence in terms
of reff . The distribution describing reff for all voxels in the midbody of the CC for all four subjects falls almost
entirely within the range spanned by values predicted by histology, with no need to account for potential
shrinkage (Horowitz et al., 2015) during tissue preparation. In all plots, radii are expressed in m.

Gray matter
It is worth examining the power law scaling also in areas outside the white mater. Therefore, Figure 9
shows the diffusion-weighted signal decay, averaged over all cortical gray matter (GM) voxels as a
function of b in the human subjects. The signal scaling in the WM is shown for qualitative compari-
pffiffiffi
son. The non-linear scaling of the isotropically-averaged signal as a function of 1= b of all human
subjects indicates strong deviations from the ‘stick’ model in the cortical GM, (McKinnon et al.,
2017; Palombo et al., 2019). Accounting for a finite neurite radius, Equation 2, does not describe

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 12 of 27

 Research article Neuroscience

Prisma
0.15 Connectom

0.05 4

0.1

Probability
0.05

0 1

0
-0.05 0 0.05 Da⊥ rMR



Figure 8. Distribution and maps of D? ?

a and rMR . (left) The distribution of Da estimated via Equation 3 for all WM
voxels (all scanner-specific subjects pooled) shown for both scan set-ups. In agreement with Figure 2, Prisma
(80mT/m) data shows a much lower precision for the estimator of D? a . Despite the small yet positive mean value
and the associated negative offset g in Figure 3, a large number of WM voxels yield biophysically implausible
D? ?
a <0 values. Precision drastically improves on the Connectom scanner (300 mT/m). (right) Maps of Da , and of the
effective MR radius heavily weighted by the tail of axonal distribution (Figure 7), for a single subject. Here, rMR is
derived from D? 2
a via Equation 10. In all plots, diffusivities and radii are expressed in m =ms and m, respectively.

pffiffiffi
the data well either. Instead, the convex signal decay as a function of 1= b at high b-values is in
good agreement with the anisotropic exchange model that we derived from the expansion of the
anisotropic Kärger model in the powers of inverse b, Equation 4. Both the finite radius and
exchange model predictions are shown in the absence of an immobile water fraction. The exchange
model fits the data better than all other evaluated models in all subjects according to an AIC analysis
(data not shown). The residence time within the neurites 1=R varies from approximately 10 to 15 ms
or 20 to 30 ms if we assume D? 2 ? 2
e ¼ 1 m =ms or De ¼ 0:5 m =ms, respectively. Dedicated experi-
ments are required for a more precise measurement of the exchange rate.

Discussion
What do we measure with dMRI?
Noninvasively estimating metrics associated with axon radius distributions is a formidable task, yet it
could have a strong impact for numerous areas of research including neuroscience, biomedicine and
even for clinical research and applications. Histological studies have extensively reported axon diam-
eters 2r to be in the range 0.5 2 mm for human WM (Aboitiz et al., 1992; Caminiti et al., 2009;
Liewald et al., 2014; Tang et al., 1997), with only 1% of all axons having a diameter larger than
3 mm (Caminiti et al., 2009). A vigorous debate has emerged in the MRI and neuroanatomy commu-
nities as in vivo, MRI-derived axon diameters are reported to fall within the range 3.5 15 mm
(Alexander et al., 2010; Horowitz et al., 2015; Huang et al., 2015). On the MRI side, the discrep-
ancy has been attributed to the long diffusion pulses that strongly reduce the signal attenuation of
protons restricted in a narrow cylinder (van Gelderen et al., 1994; Burcaw et al., 2015). Therefore,
the time-dependence of extra-axonal diffusion D? e ðtÞ (Burcaw et al., 2015; Fieremans et al., 2016;
Lee et al., 2018), and the undulation or along-axon caliber variation (Nilsson et al., 2012; Bra-
bec, 2019; Özarslan et al., 2018; Lee et al., 2019), potentially overshadow the relatively small D?
a.
On the other hand, shrinkage during tissue fixation has been suggested as a potential shortcom-
ing of histology (Barazany et al., 2009; Horowitz et al., 2015), implying that in vivo axons are
thicker than their histologically reported values.
This study aimed at investigating what the dMRI signal can measure in terms of axon radius, as
well as provide insight into the aforementioned debate. Our wide range of diffusion weightings in

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 13 of 27

 Research article Neuroscience

0.25
WM Data (GM)
GM Finite radius
0.2 Exchange

0.15

S
0.1

0.05

0
0 0.2 0.4 0.6

Figure 9. Signal decay in the GM. The spherically-averaged signal decay in the WM and GM is shown for all
pffiffiffi pffiffiffi
human subjects as a function of 1= b. The consistent non-linear scaling of the signal as a function of 1= b
demonstrates deviations from the ‘stick’ model in the cortical GM. In contrast to the WM, the convex signal decay
in the GM is better described by an anisotropic exchange model of two compartments (Equation 4), than the
finite radius model (Equation 3). In all plots, b is expressed in ms=m2 .

both human and preclinical dMRI enables a suppression of the extra-axonal contribution (that other-
wise biases the radii Burcaw et al., 2015; Fieremans et al., 2016), thereby allowing us to shed light
on this controversy. We claim that the effective MR radii measured in this study (rMR ) quantitatively
agree with those derived from histology — to the extent that histology correctly captures the tail of
the axonal radii distribution hðrÞ. That is, rMR obtained from dMRI appears to be a self-averaging
quantity in each imaging voxel, as large MRI voxels ensure that the moments of hðrÞ sampled from a
voxel represent well the ‘true’ underlying hðrÞ in that WM region, so that the spatial variations in rMR
stem mainly from genuine biological variations of the tails of axon distributions across the brain.

Mesoscopic fluctuations
Histology-derived reff are prone to mesoscopic fluctuations due to small sampling sizes, Figure 7.
Despite a good correspondence of the bulk of axon radii distributions obtained from different histo-
logical studies and samples (Aboitiz et al., 1992; Caminiti et al., 2009; Liewald et al., 2014;
Tang et al., 1997), the tail of the distribution is typically coarsely sampled, with only a few spikes
representing the occasional observation of large axons within the relatively small histological sec-
tions, Figure 7. It is precisely for the detectable large r, that the relative fluctuations for the bin
counts Ni are observed for bin values of Ni ~ 1 (Table 2 of Aboitiz et al. (1992) and Figure 5),
according to the Poissonian statistics governing each Ni . Not surprisingly, reff derived from discrete
histological histograms exhibits strong fluctuations, as depicted by dotted vertical lines in Figure 7a
and the error bars in Figure 5.

Humans
Although the average radius r, as reported in human literature (Aboitiz et al., 1992;
Innocenti et al., 2015), only ranges between 0.54 and 0.69 m, the corresponding reff varies from
0.91 to 2.9 m. In comparison, the average dMRI-derived rMR estimated from the four Connectom
data sets within the same region-of-interest, the midbody of the CC, only varied from 2.48 to
2.82 m (Figure 7b).

Rodents
The average radius, as measured in this study, varies between 0.54 and 0.68 m across the 16
patches of the genu of the CC, while the associated reff varies from 0.81 to 1.30 m. The MR-derived

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 14 of 27

 Research article Neuroscience

effective axon radii rMR vary similarly, that is 0.94 to 1.4 m across several voxels within the genu of
the CC for all three scanned samples.
For dMRI, the variability in the estimation of rMR is determined by thermal MRI noise, and genuine
anatomical – inter-voxel and inter-subject – differences. For human MRI, the mesoscopic fluctuations
are much weaker, due to the large MRI voxels in comparison to the histological samples. Indeed, the
variance in the estimation of rMR is expected to decay inversely with the number of axons within a
field of view. However, for rodent MRI, in which the MRI voxels have the same surface area as the
histological patches, the precision in the estimation of the effective radius is similar for both
modalities.
Overall, dMRI provides a precise measurement of the largest axons, which are captured within an
MRI voxel. In contrast, histology, so far, mainly probes the bulk that consists of smaller axons with
high precision. Therefore, both modalities are complementary, especially in human MRI for which
the voxels are significantly larger than a typical histological sample.

Measuring the bulk of the axon distribution using MRI

As the signal attenuation inside axons, Equations (5)-(6), scales as ln S ~ g2 reff 4
, getting to two-
times smaller reff would require another four-fold increase in gradients. However, even with stronger
gradient systems, the main bottleneck might be the missing prior knowledge about the shape of the
expected axon radius distribution hðrÞ. Even when assuming a particular functional form of hðrÞ, one
is limited to estimating a single parameter to describe the axon radius distribution, whereas realistic
distributions such as the generalized extreme value distribution (Sepehrband et al., 2016) are
parameterized by at least two variables. Hence, the reconstruction of hðrÞ from only reff is technically
ill-posed and, as such, prone to mis– or over–interpretation due to biases towards user-defined dis-
tribution shapes and parameters, even more so if confounding factors such as dispersion or fixed dif-
fusivities are ignored (Assaf et al., 2008; Barazany et al., 2009; Horowitz et al., 2015;
McNab et al., 2013).
With unknown hðrÞ, only a single metric representing the entire distribution, that is reff , for which
the strength of tail-weighting is determined by the gradient pulse width, can be estimated reliably.
In the best case, that is the narrow-pulse limit d  tc , see ’From D? a to effective MR radius’, reff will
depend on the fourth rather than the sixth order moment of hðrÞ, that is
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
reff  hr 4 i=hr 2 i (Burcaw et al., 2015), thereby reducing, but not eliminating the difference between
reff and r . Other methods, such as oscillating gradient diffusion weighting or double diffusion encod-
ing, may provide other sources of contrast encoding different aspects of the size distribution
(Jiang et al., 2016), although the low-frequency limit of the oscillating-gradient attenuation has
been shown to be equivalent to the Neuman’s limit, not providing any extra information
(Novikov et al., 2019). It can be hypothesized that the combination of methods could perhaps
recover more accurate information on the underlying hðrÞ.

Dot fraction
The presence of isotropic immobile water fim has been conjectured by Stanisz et al. (1997) as water
possibly restricted inside the soma of various cell types, such as neurons or oligodendrocytes. Sev-
eral previous studies, for example Veraart et al. (2019), Tax et al. (2019), and Dhital et al. (2019),
demonstrated with various diffusion encoding strategies that in vivo dMRI is practically insensitive,
that is < 0.2%, to such signal contributions in the healthy white matter of the living human brain,
excluding the cerebellum (Tax et al., 2019). The lack of sensitivity of dMRI to immobile water might
be explained by a small volume fraction, a short T2 relaxation time, and/or a fast water exchange
rate on the scale of our diffusion time D ¼ 30 ms for treating them as coming from separate compart-
ments. In contrast, the dot compartment has been observed in fixed brain samples in various studies
(Stanisz et al., 1997; Alexander et al., 2010). The origin of this signal contribution is not well under-
stood yet, but the still water compartment needs to be considered when validating or studying bio-
physical models in fixed tissue.
In this work, for the human experiments, we build upon the previously reported observations to
fix fim ¼ 0 in the healthy white matter to avoid fitting degeneracies that are associated with the poor
conditioning of model (iii). However, any underestimation of the dot compartment, for example due
to fixing fim ¼ 0, leads directly to an underestimation of the effective MR radius, see Figure 1.

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 15 of 27

 Research article Neuroscience

Therefore, in future studies, especially those that focus on the developing, aging, or pathological
brain, we encourage the independent measurement of the dot compartment to complement the
axon radius acquisitions. The fast measurement of the dot compartment is promoted by the avail-
ability of spherical diffusion-encoding, as demonstrated by Dhital et al. (2019), and Tax et al.
(2019).
In our ex vivo experiments, the measurement of the dot compartment is based on the diffusion-
weighing in the direction parallel to the average fiber direction in the CC at the maximal b-value of
100 ms/mm2. The measured signal fraction of such a still water compartment in our fixed brain sam-
ples was in the range of 8 to 17%, in line with Stanisz et al. (1997). Applying a radial or planar diffu-
sion-weighting filter prior to this measurement would suppress any contribution of anisotropic signal
compartments, such as crossing or dispersed axons, to the isotropically restricted dot compartment.
Although we aimed to minimize this confounding factor by using a very high b-value (Dhital et al.,
2019), the dot compartment fraction, and as such the effective MR radius, might be slightly overesti-
mated because of various complex fiber configurations. Additional confounding factors are listed in
the following section.

Confounding factors
The apparent discrepancy between histology and dMRI, when confounding factors such as extra-
axonal water (Burcaw et al., 2015; Fieremans et al., 2016; Lee et al., 2018) and orientational dis-
persion (Drobnjak et al., 2016; Nilsson et al., 2012) are addressed, is mainly a result of the differ-
ence between r and reff – that is between the bulk and the tail of axonal distribution. This already
provides an important insight into the discussion on why the radii reported in the literature vary so
much between the methods. When comparing apples-to-apples, despite the excellent agreement
observed in this study between rMR and its histological counterpart reff , in our own histological vali-
dation we still observed a small, yet consistent overestimation of between 5 and 20% in axon radius
rMR using dMRI. Aside from the previously discussed dot compartment, various other factors might
contribute to this discrepancy.
First, an underlying assumption of all studies targeting the measurement of the axon radius is
specificity: that the signal observed at these high b-values could be attributed exclusively to the
intra-axonal space. However, this assumption is not established nor in our opinion is it justified given
that water resides in all cellular compartments of the central nervous system. We cannot exclude
that water trapped in other ‘stick’-like features such as the radiating processes of astrocytes contrib-
ute observable signals; it has been previously reported, but also observed in our histological sample,
that such glial processes can have large diameters, up to 7 m in our sample. In the future, this con-
tribution could be investigated using the increased cellular specificity of (diffusion-weighted) spec-
troscopy (Palombo et al., 2016; Shemesh et al., 2017; Ligneul et al., 2019).
Second, Stanisz et al. (1997) and, more recently, Palombo et al. (2019) demonstrated that at
shorter diffusion times, the signal contribution from cell bodies might be characterized by a specific
b-value dependent signature (Neuman, 1974 and Murday and Cotts, 1968) that might enable the
extraction of MR effective cell body sizes in both the white and gray matter (Palombo et al., 2019).
In this study, the potential b-value dependent signal contribution of cell bodies was unaccounted for,
and, given our and other (e.g. Sampaio-Baptista et al., 2019) observations of a finite cell body vol-
ume fraction, the axon radius measurements could be biased. However, deviations to the power law
scaling due to the presence of cell bodies are more likely to be expected in the GM because of
larger volume density of large somas in comparison to the WM (Palombo et al., 2019). In our histo-
logical sample, we observed a significant volume fraction of cell bodies in the genu of the CC, that is
5%, but due to unknown compartmental relaxation times, the associated, yet more important, signal
fraction is unknown (Lampinen et al., 2019). Regardless, a biophysical model parameterized by vari-
ous volume fractions, axon radii, soma sizes, and compartmental diffusivities may be poorly condi-
tioned and degenerate.
Third, along-axon undulations (Nilsson et al., 2012) and curvature (Özarslan et al., 2018) might
result in an increased apparent radial diffusivity and, as such, contribute to an overestimation of the
axon radius using dMRI, especially for increasingly long diffusion times (Lee et al., 2019;
Brabec, 2019).
Finally, the estimation of the MR effective axon radius depends on the unknown intrinsic diffusiv-
ity D0 of the axoplasm. In ex vivo samples of a well-aligned WM bundle, one could estimate D0

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 16 of 27

 Research article Neuroscience

directly by exploring the time dependence of the apparent diffusivity at very short diffusion times,
(Mitra et al., 1993). In this study, we were not able to achieve a reproducible and precise estimate
of D0 and opted to use the longitudinal diffusivity Dka as a proxy for D0 , with Dka  D0 . Therefore, we
might actually underestimate the positive bias in the estimation of ^reff ~ ðD0 Þ1=4 (Equation 10). How-
ever, the propagation of the error in the estimation of D0 to ^reff is strongly reduced by the fourth
root relation between both metrics.

Inter-species variability
In our study, the effective MR radius in humans was significantly higher than in rats when comparing
similar regions of interest, for example the midbody of the CC. This difference is in agreement with
several studies that compared the callosal fiber composition as a function of the brain size of various
mammals and concluded that large brains have more large axons and an increased maximal radius,
whereas the bulk of axons is not altered (Olivares et al., 2001; Schüz and Preibl, 1996;
Liewald et al., 2014). Since the effective MR radius is predominantly sensitive to the larger axons,
observed differences between humans and rats will be amplified when comparing effective MR radii.
Overall, this observation favors future application of MR axon radius mapping in species with rela-
tively large brain sizes.

Gray matter
Although this work mainly focuses on the WM, we do report significantly different signal scaling for
the cortical GM. We suggest that the proton exchange between dendrites and interstitial water
might explain this scaling behavior, in particular due to the convex scaling with b 1=2 . However, the
abundance of cell bodies in the gray matter might confound this analysis (Palombo et al., 2019).
Moreover, the study of the cortical GM is challenged by its low SNR and susceptibility to partial vol-
uming. Nonetheless, we conclude that the stick assumption does not hold in the cortical GM and
that biophysical models building upon that assumption must be interpreted with caution if applied
to tissue regions outside of WM.

Conclusion
In summary, we provide a realistic perspective on MR axon radius mapping by showing MR-derived
effective radii that have good quantitative agreement with histology. First, we compared the MR-
derived axon radii directly to confocal microscopy of the same rat brain samples. Second, the distri-
bution of dMRI-derived rMR of the living human brain falls almost entirely within the range spanned
by histology-derived reff that has been reported in the literature — even without introducing a puta-
tive axonal shrinkage factor. This estimation is inherently bound to a single scalar reff that encodes
moments of the axon distribution, which is – by virtue of the signal encoding – dominated by the
largest axons. Therefore, the average axon radius r and reff can be practically considered as two
complementary metrics probing the underlying axon caliber distribution: histology, so far, mainly
probes its bulk, that is r , while dMRI probes rMR ¼ ^reff , that is its tail. Due to the intrinsic bias of MR-
derived axon radii to larger axons, clinical applications should focus on pathologies that specifically
target those larger axons, until other methods are developed that probe the smaller axon diameter.

Materials and methods

Key resources table
Additional
Reagent type (species) or resource Designation Source or reference Identifiers information
Antibody anti-Neurofilament 160/200 Sigma Aldrich Cat# N2912 2.5 mg/mL
(Mouse monoclonal) (clone RMdo20)
Antibody anti-GFAP Thermo Cat# PA1-10019 1:1000
(rabbit polyclonal) Fisher Scientific
software. ImageJ imagej.nih.gov/ij/ RRID: SCR003070 1.52q
algorithm
Continued on next page

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 17 of 27

 Research article Neuroscience

Continued

Additional
Reagent type (species) or resource Designation Source or reference Identifiers information
software. FSL fsl.fmrib.ox.ac.uk/fsl/ RRID: SCR002823 v6
algorithm
software. MRtrix www.mrtrix.org RRID: SCR006971 v3.0
algorithm
software. FreeSurfer surfer.nmr.mgh. RRID: SCR001847 v6.0.0
algorithm harvard.edu
Other DAPI Sigma Aldrich Cat# D9542 500nM

MRI of fixed rat brain tissue

Ethics
Animals used in this study were handled in agreement with the European FELASA guide-lines and all
procedures were approved by the Champalimaud Animal Welfare Body and by the national authori-
ties, Direção Geral de Alimentação e Veterinária, Lisbon, Portugal, under the approved protocol
number 0421/000/000/2016. All animal care procedures were conducted in agreement with the
European Directive 2010/63, at the vivarium of the Champalimaud Foundation, a research facility
part of CONGENTO, project number Lisboa-01–0145-FEDER-022170.

Sample preparation
Three Long Evans rats (Female, 12-weeks-old) were transcardially perfused using 4% paraformalde-
hyde. The extracted brains were kept for 24 hr in 4% paraformaldehyde and washed using PBS over
two days (changed daily). Given our focus on the CC of the rat brains, we will refer to the samples as
CC#1, CC#2, and CC#3.

MRI scanning
i. Multi-shell dMRI data: The three samples were scanned on an 16.4T MR scanner (Bruker Bio-
Spin) at room temperature with D=d ¼ 20=7:1 ms interfaced with an AVANCE IIIHD console
and a micro2.5 imaging probe with maximal gradient amplitude Gmax ¼ 1500 mT=m. Diffu-
sion-weighting was applied using a RARE sequence in the midsagittal plane along 60 gradi-
ent directions for a densely sampled spectrum of 18 different b-values up to 100 ms/mm2.
Furthermore, TR=TE ¼ 2400=30:4 ms and the spatial resolution was 100  100  850mm3.
ii. ‘dot fraction’ fim : we acquired 60 repeated measurements of diffusion-weighing applied in
the direction parallel to the average fiber direction in the corpus callosum (CC) at the maxi-
mal b-value of 100 ms/mm2. The average fiber orientation was defined as the first eigenvector
of the dyadic tensor (Jones, 2003) that was computed from the voxelwise first eigenvectors
of the diffusion tensors that were estimated by fitting the DTI model to the lowest b-values,
that is b<5 ms=m2 , of the multi-shell data in each voxel within the manually segmented CC
(Basser et al., 1994).
The average SNR for Sjb¼0 was 195 and some examples of the acquired images at various low and
high b-values are shown in Figure 10, where the quality of the raw data can be evaluated. Notably,
since images are spherically averaged over many directions, the signal is characterized by high SNR
even at high b-values.

Data analysis
From the multi-shell data, the spherically-averaged signals SðbÞ are estimated per b-value as the
zeroth order spherical harmonic using a Rician maximum likelihood estimator of the even order
spherical harmonic coefficients up to the 6th order (Sijbers et al., 1998).
The spatially localized dot fraction fim is computed as the signal estimated from the repeated
(N ¼ 60) measurements in the direction parallel to the principal fibre direction using a Rician maxi-
mum likelihood estimator with pre-computed noise level (Veraart et al., 2016), normalized by the
respective non-diffusion weighted signal. We compute the dot-free signal in each voxel as follows

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 18 of 27

 Research article Neuroscience

(a) Preclincal (b) Human

2 5
b=0

b=0
0 0
b = 50 ms/µm2

b = 11 ms/µm2
0.5 0.5

0 0
b = 100 ms/µm2

b = 25 ms/µm2
0.25 0.25

0 0

Figure 10. Raw data. The spherically-averaged diffusion-weighted images, prior to any other image corrections, are shown for various low and high b-
values for one rat brain sample (a) and one human subject (b).

S* ðbÞ ¼ SðbÞ fim . In the remainder of the work, analyses were done on both ‘dot contaminated’ SðbÞ
and ‘dot free’ S* ðbÞ signals.
^?
The intra-axonal radial diffusivity Da is estimated voxelwise by fitting Equation 2 with fim ¼ 0 to
*
S ðb  20Þ using a nonlinear least squares estimator (code is available for download on GitHub
[Veraart and Novikov, 2019; https://github.com/NYU-DiffusionMRI/AxonRadiusMapping; copy
archived at https://github.com/elifesciences-publications/AxonRadiusMapping]).
Thereafter, the estimated effective axon radius rMR is derived from D ^?
a using Equation 6. The
^ a and fim from Sðb  20Þ is very poorly
alternative approach, that is the simultaneous estimation of D ?

conditioned. Hence, disentangling both parameters from only the linearly-encoded multi-shell data
is impossible, even at unrealistically high SNR.

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 19 of 27

 Research article Neuroscience

Histology of fixed rat brain tissue

Full details of the immunohistochemistry for sample preparation, confocal microscopy, and image
analysis are provided in Nunes et al. (2017). Study-specific elements are described below.

Sample preparation
After MRI scanning, free-floating horizontal sections 50 m-thick were collected from the medial lat-
eral center of two rat brains, CC#1 and CC#2, corresponding to the MR imaged volume. For CC#2,
we used antibodies against neurofilaments 160/200 (axonal marker; Sigma-Aldrich, cat.# N2912) and
GFAP (astrocytes marker; ThermoFisher Scientific, cat.# PA1-10019), as well as a staining for cell
bodies using DAPI (Sigma-Adrich, cat.# D9542). For CC#1, the staining was limited to the neurofila-
ments to focus on the axon radius count.

Confocal microscopy
A Zeiss LSM 710 laser scanning confocal microscope was used for immunohistochemistry image
acquisition. A tile scan using a 10 objective (EC Plan Neofluar, numerical aperture = 0.3, Zeiss, Ger-
many) was used to cover the entire CC. Various ROIs were imaged using a 63 immersion objective
(Plan Apochromat, numerical aperture = 1.4, Zeiss, Germany) in confocal mode, with pixel resolution
of 65  65  150 nm3 and field-of-view of 135  135 mm2 (Figure 11). The placement of the ROIs is
shown in Figures 4 and 5 for CC#1. and CC#2, respectively.

Confocal microscopy data analysis

Images were processed using the ImageJ software. Noise suppression of the confocal single frames
was done using a subsequent application of a 2D anisotropic diffusion filter and bandpass filtering in
the frequency domain, (Nunes et al., 2017). Thereafter, axons were identified as particles with a
minimum area size of 0.2 mm2 and a circularity larger than 0.4 in the confocal images that were
stained for neurofilaments. The number of extracted axons varied from about 500 to 2000, depend-
ing on the placement of the ROI. The long axes of fitted ellipsoids served as proxies for the respec-
tive axon radii. For each ROI, we obtain a distribution of axon radii hðrÞ from which we compute the
associate effective radius reff using Equation 9.

In vivo MRI of human brain

Ethics
Data were acquired after obtaining written informed consent and consent to publish. The project
was approved by the Cardiff University School of Psychology Ethics Committee (approval number
EC.06.05.02.891).

Subjects
Four healthy volunteers (3 males and 1 female between 22 and 45 years) were recruited for this
study. We will refer to the human subjects as 22M, 25F, 32M, and 45M to encode both the age and
gender. The data were collected under the approval of the Cardiff University School of Psychology
Ethics Committee.

MRI scanning
All four subjects were scanned on a Siemens Connectom 3T MR scanner using a 32-channel receiver
coil. The 300 mT/m gradient system was used to achieve b-values up to 25 ms=m2 . The diffusion
gradients were characterized by D=d ¼ 30=13 ms and maximal gradient amplitude of 289 mT/m. Dif-
fusion weighting was applied along 60 isotropically distributed gradient directions (Jones et al.,
1999) for b ¼ 1, 3, 5, 7, 9, 11, 12.1, 13.5, 15, 16.9, 19.1, 21.7, and 25 ms=m2 , with TR/TE : 3500/
62 ms, matrix: 74  74, and 42 slices with a spatial resolution of 3  3  3 mm3 . The average SNR for
Sjb¼0 was 52. See Figure 11 for the image quality and contrast at various b-values.
The dot compartment was not measured directly (see Dhital et al., 2018 and Tax et al., 2019).

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 20 of 27

 Research article Neuroscience

Figure 11. For two brain samples, MR scanning (a, color encoded FA map) was followed by low (b) and high (c) resolution confocal microscopy with
staining for neurofilaments to identify the axons. The low-resolution image was used to position various ROIs, whereas the axon caliber distributions
were extracted from the high-resolution image of the corresponding ROIs. The long axes of fitted ellipsoids served as proxies for the respective axon
diameters (d).

Data analysis
Image processing was done according to the DESIGNER pipeline (Ades-Aron et al., 2018) using the
FSL (Smith et al., 2004) and MRtrix (Tournier et al., 2019) software packages. In particular, MPPCA
noise estimation and denoising (Veraart et al., 2016) were used for estimating noise maps sðxÞ by
exploiting the inherent redundancy in dMRI data. The positive signal bias, inherent to low-SNR mag-
nitude MR data, was removed by using the method of moments (Koay and Basser, 2006), where
the denoised signal was used as a proxy for the Rician expectation value. Denoised and Rice-floor-
corrected images were subsequently corrected for Gibbs ringing (Kellner et al., 2016), geometric
eddy current distortions and subject motion (Andersson and Sotiropoulos, 2016). The pipeline is
available on https://github.com/NYU-DiffusionMRI/DESIGNER (Ades-Aron and Veraart, 2018). We
used tract-density imaging (Calamante et al., 2010) based on whole-brain probabilistic fiber-track-
ing (Tournier et al., 2019) of the b ¼ 5 ms=m2 -shell for identifying all WM voxels. To avoid voxels
affected by partial voluming with the gray matter (GM), an additional, more conservative, segmenta-
tion was obtained by omitting all voxels with a fractional anisotropy smaller than 0.6. In addition, the
cortical GM was segmented using FreeSurfer (Dale et al., 1999).

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 21 of 27

 Research article Neuroscience

Acknowledgements
All authors would like to thank Prof. Mark D Does (Vanderbilt University) for the remmiRARE
sequence used in this study, supported by grant number NIH EB019980, and Dr. Erika Raven for dis-
cussions and comments on the manuscript. JV is a Postdoctoral Fellow of the Research Foundation -
Flanders (FWO; grant number 12S1615N). DN and NS are supported by the European Research
Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Start-
ing Grant, agreement No. 679058). EF and DSN were supported by the NIH/NINDS award
R01NS088040 and research was performed as part of the Center of Advanced Imaging Innovation
and Research (CAI2R, www.cai2r.net), an NIBIB Biomedical Technology Resource Center (NIH P41
EB017183). The Connectom data were acquired at the UK National Facility for in vivo MR Imaging of
Human Tissue Microstructure funded by the EPSRC (grant EP/M029778/1), and The Wolfson Foun-
dation. DKJ is supported by a Wellcome Trust Investigator Award (096646/Z/11/Z) and a Wellcome
Trust Strategic Award (104943/Z/14/Z).

Additional information
Funding
Funder Grant reference number Author
Fonds Wetenschappelijk On- 12S1615N Jelle Veraart
derzoek
National Institute of Neurolo- R01NS088040 Els Fieremans
gical Disorders and Stroke Dmitry S Novikov
National Institute of Biomedi- P41 EB017183 Els Fieremans
cal Imaging and Bioengineer- Dmitry S Novikov
ing
H2020 European Research 679058 Noam Shemesh
Council Daniel Nunes
Engineering and Physical EP/M029778/1 Derek K Jones
Sciences Research Council
Wellcome 096646/Z/11/Z Derek K Jones
Wellcome 104943/Z/14/Z Derek K Jones

The funders had no role in study design, data collection and interpretation, or the
decision to submit the work for publication.

Author contributions
Jelle Veraart, Conceptualization, Data curation, Software, Formal analysis, Funding acquisition, Vali-
dation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and edit-
ing; Daniel Nunes, Conceptualization, Data curation, Software, Formal analysis, Visualization,
Methodology, Writing - original draft, Writing - review and editing; Umesh Rudrapatna, Data cura-
tion, Writing - original draft, Writing - review and editing; Els Fieremans, Conceptualization, Funding
acquisition, Writing - original draft, Writing - review and editing; Derek K Jones, Resources, Data
curation, Funding acquisition, Writing - original draft, Writing - review and editing; Dmitry S Novikov,
Conceptualization, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visu-
alization, Methodology, Writing - original draft, Writing - review and editing; Noam Shemesh, Con-
ceptualization, Resources, Data curation, Supervision, Visualization, Methodology, Writing - original
draft, Writing - review and editing

Author ORCIDs
Jelle Veraart https://orcid.org/0000-0003-0781-0420
Daniel Nunes https://orcid.org/0000-0001-8882-3228
Els Fieremans https://orcid.org/0000-0002-1384-8591
Dmitry S Novikov https://orcid.org/0000-0002-4213-3050
Noam Shemesh https://orcid.org/0000-0001-6681-5876

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 22 of 27

 Research article Neuroscience

Ethics
Human subjects: Human studies were carried out under a protocol (EC.06.05.02.891) approved by
Cardiff University School of Psychology Ethics Committee. Written informed consent and consent to
publish was obtained from all participants.
Animal experimentation: Animals used in this study were handled in agreement with the European
FELASA guide-lines and all procedures were approved by the Champalimaud Animal Welfare Body
and by the national authorities, Direção Geral de Alimentação e Veterinária, Lisbon, Portugal, under
the approved protocol number 0421/000/000/2016. All animal care procedures were conducted in
agreement with the European Directive 2010/63, at the vivarium of the Champalimaud Foundation,
a research facility part of CONGENTO, project number Lisboa-01-0145-FEDER-022170.

Decision letter and Author response

Decision letter https://doi.org/10.7554/eLife.49855.sa1
Author response https://doi.org/10.7554/eLife.49855.sa2

Additional files
Data availability
All source data files generated or analysed during this study have been deposited in Dryad Digital
Repository (http://doi.org/10.5061/dryad.4qrfj6q66).

The following dataset was generated:

Database and
Author(s) Year Dataset title Dataset URL Identifier
Veraart J, Nunes D, 2019 Data from: Nonivasive quantication http://doi.org/10.5061/ Dryad Digital
Rudrapatna U, of axon radii using diffusion MRI dryad.4qrfj6q66 Repository, 10.5061/
Fieremans E, Jones dryad.4qrfj6q66
DK, Novikov DS,
Shemesh N

References
Aboitiz F, Scheibel AB, Fisher RS, Zaidel E. 1992. Fiber composition of the human corpus callosum. Brain
Research 598:143–153. DOI: https://doi.org/10.1016/0006-8993(92)90178-C
Ades-Aron B, Veraart J, Kochunov P, McGuire S, Sherman P, Kellner E, Novikov DS, Fieremans E. 2018.
Evaluation of the accuracy and precision of the diffusion parameter EStImation with Gibbs and NoisE removal
pipeline. NeuroImage 183:532–543. DOI: https://doi.org/10.1016/j.neuroimage.2018.07.066
Ades-Aron B, Veraart J. 2018. Desginer: diffusion parameter estimation with Gibbs and noise removal . GitHub.
https://github.com/NYU-DiffusionMRI/DESIGNER
Alexander DC, Hubbard PL, Hall MG, Moore EA, Ptito M, Parker GJM, Dyrby TB. 2010. Orientationally invariant
indices of axon diameter and density from diffusion MRI. NeuroImage 52:1374–1389. DOI: https://doi.org/10.
1016/j.neuroimage.2010.05.043
Andersson JLR, Sotiropoulos SN. 2016. An integrated approach to correction for off-resonance effects and
subject movement in diffusion MR imaging. NeuroImage 125:1063–1078. DOI: https://doi.org/10.1016/j.
neuroimage.2015.10.019
Assaf Y, Blumenfeld-Katzir T, Yovel Y, Basser PJ. 2008. Axcaliber: a method for measuring axon diameter
distribution from diffusion MRI. Magnetic Resonance in Medicine 59:1347–1354. DOI: https://doi.org/10.1002/
mrm.21577
Assaf Y, Alexander DC, Jones DK, Bizzi A, Behrens TEJ, Clark CA, Cohen Y, Dyrby TB, Huppi PS, Knoesche TR,
LeBihan D, Parker GJM, Poupon C. 2013. The CONNECT project: combining macro- and micro-structure.
NeuroImage 80:273–282. DOI: https://doi.org/10.1016/j.neuroimage.2013.05.055
Barazany D, Basser PJ, Assaf Y. 2009. In vivo measurement of axon diameter distribution in the corpus callosum
of rat brain. Brain 132:1210–1220. DOI: https://doi.org/10.1093/brain/awp042
Baron CA, Kate M, Gioia L, Butcher K, Emery D, Budde M, Beaulieu C. 2015. Reduction of diffusion-weighted
imaging contrast of acute ischemic stroke at short diffusion times. Stroke 46:2136–2141. DOI: https://doi.org/
10.1161/STROKEAHA.115.008815
Basser PJ, Mattiello J, LeBihan D. 1994. MR diffusion tensor spectroscopy and imaging. Biophysical Journal 66:
259–267. DOI: https://doi.org/10.1016/S0006-3495(94)80775-1
Beaulieu C. 2002. The basis of anisotropic water diffusion in the nervous system - a technical review. NMR in
Biomedicine 15:435–455. DOI: https://doi.org/10.1002/nbm.782

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 23 of 27

 Research article Neuroscience

Behrens TEJ, Woolrich MW, Jenkinson M, Johansen-Berg H, Nunes RG, Clare S, Matthews PM, Brady JM, Smith
SM. 2003. Characterization and propagation of uncertainty in diffusion-weighted MR imaging. Magnetic
Resonance in Medicine 50:1077–1088. DOI: https://doi.org/10.1002/mrm.10609
Brabec J. 2019. Time-dependent diffusion in undulating structures: impact on axon diameter estimation. arXiv.
https://arxiv.org/abs/1903.04536.
Burcaw LM, Fieremans E, Novikov DS. 2015. Mesoscopic structure of neuronal tracts from time-dependent
diffusion. NeuroImage 114:18–37. DOI: https://doi.org/10.1016/j.neuroimage.2015.03.061
Burnham KP, Anderson DR. 2002. Information and Likelihood Theory: A Basis for Model Selection and Inference.
In: Burnham K. P, Anderson D. R (Eds). Model Selection and Multimodel Inference. Springer. p. 49–97.
Calamante F, Tournier J-D, Jackson GD, Connelly A. 2010. Track-density imaging (TDI): Super-resolution white
matter imaging using whole-brain track-density mapping. NeuroImage 53:1233–1243. DOI: https://doi.org/10.
1016/j.neuroimage.2010.07.024
Callaghan PT, Jolley KW, Lelievre J. 1979. Diffusion of water in the endosperm tissue of wheat grains as studied
by pulsed field gradient nuclear magnetic resonance. Biophysical Journal 28:133–141. DOI: https://doi.org/10.
1016/S0006-3495(79)85164-4
Callaghan PT, Eccles CD, Xia Y. 1988. NMR microscopy of dynamic displacements: k-space and q-space imaging.
Journal of Physics E: Scientific Instruments 21:820–822. DOI: https://doi.org/10.1088/0022-3735/21/8/017
Callaghan PT. 1991. Principles of Nuclear Magnetic Resonance Microscopy. Oxford: Clarendon Press.
Callaghan PT. 1997. A simple matrix formalism for spin Echo analysis of restricted diffusion under generalized
gradient waveforms. Journal of Magnetic Resonance 129:74–84. DOI: https://doi.org/10.1006/jmre.1997.1233
Caminiti R, Ghaziri H, Galuske R, Hof PR, Innocenti GM. 2009. Evolution amplified processing with temporally
dispersed slow neuronal connectivity in primates. PNAS 106:19551–19556. DOI: https://doi.org/10.1073/pnas.
0907655106, PMID: 19875694
Campbell GR, Worrall JT, Mahad DJ. 2014. The central role of mitochondria in axonal degeneration in multiple
sclerosis. Multiple Sclerosis Journal 20:1806–1813. DOI: https://doi.org/10.1177/1352458514544537
Dale AM, Fischl B, Sereno MI. 1999. Cortical surface-based analysis. I. segmentation and surface reconstruction.
NeuroImage 9:179–194. DOI: https://doi.org/10.1006/nimg.1998.0395, PMID: 9931268
Dhital B, Kellner E, Kiselev VG, Reisert M. 2018. The absence of restricted water pool in brain white matter.
NeuroImage 182:398–406. DOI: https://doi.org/10.1016/j.neuroimage.2017.10.051
Dhital B, Reisert M, Kellner E, Kiselev VG. 2019. Intra-axonal diffusivity in brain white matter. NeuroImage 189:
543–550. DOI: https://doi.org/10.1016/j.neuroimage.2019.01.015
Drakesmith M, Harms R, Rudrapatna SU, Parker GD, Evans CJ, Jones DK. 2019. Estimating axon conduction
velocity in vivo from microstructural MRI. NeuroImage 203:116186. DOI: https://doi.org/10.1016/j.neuroimage.
2019.116186
Drobnjak I, Siow B, Alexander DC. 2010. Optimizing gradient waveforms for microstructure sensitivity in
diffusion-weighted MR. Journal of Magnetic Resonance 206:41–51. DOI: https://doi.org/10.1016/j.jmr.2010.05.
017
Drobnjak I, Zhang H, Ianuş A, Kaden E, Alexander DC. 2016. PGSE, OGSE, and sensitivity to axon diameter in
diffusion MRI: insight from a simulation study. Magnetic Resonance in Medicine 75:688–700. DOI: https://doi.
org/10.1002/mrm.25631
Eichert N, Verhagen L, Folloni D, Jbabdi S, Khrapitchev AA, Sibson NR, Mantini D, Sallet J, Mars RB. 2019. What
is special about the human arcuate fasciculus? lateralization, projections, and expansion. Cortex 118:107–115.
DOI: https://doi.org/10.1016/j.cortex.2018.05.005
Evangelou N, Konz D, Esiri MM, Smith S, Palace J, Matthews PM. 2001. Size-selective neuronal changes in the
anterior optic pathways suggest a differential susceptibility to injury in multiple sclerosis. Brain 124:1813–1820.
DOI: https://doi.org/10.1093/brain/124.9.1813, PMID: 11522583
Fieremans E, Jensen JH, Helpern JA. 2011. White matter characterization with diffusional kurtosis imaging.
NeuroImage 58:177–188. DOI: https://doi.org/10.1016/j.neuroimage.2011.06.006
Fieremans E, Burcaw LM, Lee H-H, Lemberskiy G, Veraart J, Novikov DS. 2016. In vivo observation and
biophysical interpretation of time-dependent diffusion in human white matter. NeuroImage 129:414–427.
DOI: https://doi.org/10.1016/j.neuroimage.2016.01.018
Glasser MF, Smith SM, Marcus DS, Andersson JLR, Auerbach EJ, Behrens TEJ, Coalson TS, Harms MP, Jenkinson
M, Moeller S, Robinson EC, Sotiropoulos SN, Xu J, Yacoub E, Ugurbil K, Van Essen DC. 2016. The human
connectome project’s neuroimaging approach. Nature Neuroscience 19:1175–1187. DOI: https://doi.org/10.
1038/nn.4361
Horowitz A, Barazany D, Tavor I, Bernstein M, Yovel G, Assaf Y. 2015. In vivo correlation between axon diameter
and conduction velocity in the human brain. Brain Structure and Function 220:1777–1788. DOI: https://doi.org/
10.1007/s00429-014-0871-0
Huang SY, Nummenmaa A, Witzel T, Duval T, Cohen-Adad J, Wald LL, McNab JA. 2015. The impact of gradient
strength on in vivo diffusion MRI estimates of axon diameter. NeuroImage 106:464–472. DOI: https://doi.org/
10.1016/j.neuroimage.2014.12.008
Innocenti GM, Caminiti R, Aboitiz F. 2015. Comments on the paper by Horowitz et al. (2014). Brain Structure and
Function 220:1789–1790. DOI: https://doi.org/10.1007/s00429-014-0974-7
Jbabdi S, Sotiropoulos SN, Haber SN, Van Essen DC, Behrens TE. 2015. Measuring macroscopic brain
connections in vivo. Nature Neuroscience 18:1546–1555. DOI: https://doi.org/10.1038/nn.4134
Jensen JH, Russell Glenn G, Helpern JA. 2016. Fiber ball imaging. NeuroImage 124:824–833. DOI: https://doi.
org/10.1016/j.neuroimage.2015.09.049

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 24 of 27

 Research article Neuroscience

Jespersen SN, Kroenke CD, Østergaard L, Ackerman JJH, Yablonskiy DA. 2007. Modeling dendrite density from
magnetic resonance diffusion measurements. NeuroImage 34:1473–1486. DOI: https://doi.org/10.1016/j.
neuroimage.2006.10.037
Jespersen SN, Bjarkam CR, Nyengaard JR, Chakravarty MM, Hansen B, Vosegaard T, Østergaard L, Yablonskiy
D, Nielsen NC, Vestergaard-Poulsen P. 2010. Neurite density from magnetic resonance diffusion measurements
at ultrahigh field: comparison with light microscopy and electron microscopy. NeuroImage 49:205–216.
DOI: https://doi.org/10.1016/j.neuroimage.2009.08.053
Jespersen SN, Lundell H, Sønderby CK, Dyrby TB. 2013. Orientationally invariant metrics of apparent
compartment eccentricity from double pulsed field gradient diffusion experiments. NMR in Biomedicine 26:
1647–1662. DOI: https://doi.org/10.1002/nbm.2999
Jiang X, Li H, Xie J, Zhao P, Gore JC, Xu J. 2016. Quantification of cell size using temporal diffusion
spectroscopy. Magnetic Resonance in Medicine 75:1076–1085. DOI: https://doi.org/10.1002/mrm.25684
Jones DK, Horsfield MA, Simmons A. 1999. Optimal strategies for measuring diffusion in anisotropic systems by
magnetic resonance imaging. Magnetic Resonance in Medicine 42:515–525. DOI: https://doi.org/10.1002/(SICI)
1522-2594(199909)42:3<515::AID-MRM14>3.0.CO;2-Q
Jones DK. 2003. Determining and visualizing uncertainty in estimates of fiber orientation from diffusion tensor
MRI. Magnetic Resonance in Medicine 49:7–12. DOI: https://doi.org/10.1002/mrm.10331
Jones DK. 2010. Diffusion MRI: Theory, Methods and Applications. Oxford University Press.
Jones DK, Alexander DC, Bowtell R, Cercignani M, Dell’Acqua F, McHugh DJ, Miller KL, Palombo M, Parker
GJM, Rudrapatna US, Tax CMW. 2018. Microstructural imaging of the human brain with a ‘super-scanner’: 10
key advantages of ultra-strong gradients for diffusion MRI. NeuroImage 182:8–38. DOI: https://doi.org/10.
1016/j.neuroimage.2018.05.047
Kaden E, Kruggel F, Alexander DC. 2016. Quantitative mapping of the per-axon diffusion coefficients in brain
white matter. Magnetic Resonance in Medicine 75:1752–1763. DOI: https://doi.org/10.1002/mrm.25734
Kärger J. 1985. NMR self-diffusion studies in heterogeneous systems. Advances in Colloid and Interface Science
23:129–148. DOI: https://doi.org/10.1016/0001-8686(85)80018-X
Kay S. 1993. Fundamentals of Statistical Signal Processing. Prentice Hall PTR.
Kellner E, Dhital B, Kiselev VG, Reisert M. 2016. Gibbs-ringing artifact removal based on local subvoxel-shifts.
Magnetic Resonance in Medicine 76:1574–1581. DOI: https://doi.org/10.1002/mrm.26054
Kjellström C, Conradi NG. 1993. Decreased axonal calibres without axonal loss in optic nerve following chronic
alcohol feeding in adult rats: a morphometric study. Acta Neuropathologica 85:117–121. DOI: https://doi.org/
10.1007/BF00227757, PMID: 8442403
Koay CG, Basser PJ. 2006. Analytically exact correction scheme for signal extraction from noisy magnitude MR
signals. Journal of Magnetic Resonance 179:317–322. DOI: https://doi.org/10.1016/j.jmr.2006.01.016
Kroenke CD, Ackerman JJH, Yablonskiy DA. 2004. On the nature of the NAA diffusion attenuated MR signal in
the central nervous system. Magnetic Resonance in Medicine 52:1052–1059. DOI: https://doi.org/10.1002/
mrm.20260
Lampinen B, Szczepankiewicz F, Novén M, van Westen D, Hansson O, Englund E, Mårtensson J, Westin CF,
Nilsson M. 2019. Searching for the neurite density with diffusion MRI: challenges for biophysical modeling.
Human Brain Mapping 40:2529–2545. DOI: https://doi.org/10.1002/hbm.24542, PMID: 30802367
Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval-Jeantet M. 1986. MR imaging of intravoxel
incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 161:401–407.
DOI: https://doi.org/10.1148/radiology.161.2.3763909
Le Bihan D. 2003. Looking into the functional architecture of the brain with diffusion MRI. Nature Reviews
Neuroscience 4:469–480. DOI: https://doi.org/10.1038/nrn1119
Lebel C, Beaulieu C. 2009. Lateralization of the arcuate fasciculus from childhood to adulthood and its relation to
cognitive abilities in children. Human Brain Mapping 30:3563–3573. DOI: https://doi.org/10.1002/hbm.20779
Lee H-H, Fieremans E, Novikov DS. 2018. What dominates the time dependence of diffusion transverse to axons:
intra- or extra-axonal water? NeuroImage 182:500–510. DOI: https://doi.org/10.1016/j.neuroimage.2017.12.
038
Lee H-H, Yaros K, Veraart J, Pathan JL, Liang F-X, Kim SG, Novikov DS, Fieremans E. 2019. Along-axon diameter
variation and axonal orientation dispersion revealed with 3D electron microscopy: implications for quantifying
brain white matter microstructure with histology and diffusion MRI. Brain Structure and Function 224:1469–
1488. DOI: https://doi.org/10.1007/s00429-019-01844-6
Liewald D, Miller R, Logothetis N, Wagner H-J, Schüz A. 2014. Distribution of axon diameters in cortical white
matter: an electron-microscopic study on three human brains and a macaque. Biological Cybernetics 108:541–
557. DOI: https://doi.org/10.1007/s00422-014-0626-2
Ligneul C, Palombo M, Hernández-Garzón E, Carrillo-de Sauvage M-A, Flament J, Hantraye P, Brouillet E,
Bonvento G, Escartin C, Valette J. 2019. Diffusion-weighted magnetic resonance spectroscopy enables cell-
specific monitoring of astrocyte reactivity in vivo. NeuroImage 191:457–469. DOI: https://doi.org/10.1016/j.
neuroimage.2019.02.046
McKinnon ET, Jensen JH, Glenn GR, Helpern JA. 2017. Dependence on b-value of the direction-averaged
diffusion-weighted imaging signal in brain. Magnetic Resonance Imaging 36:121–127. DOI: https://doi.org/10.
1016/j.mri.2016.10.026
McNab JA, Edlow BL, Witzel T, Huang SY, Bhat H, Heberlein K, Feiweier T, Liu K, Keil B, Cohen-Adad J, Tisdall
MD, Folkerth RD, Kinney HC, Wald LL. 2013. The human connectome project and beyond: initial applications of
300mT/m gradients. NeuroImage 80:234–245. DOI: https://doi.org/10.1016/j.neuroimage.2013.05.074

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 25 of 27

 Research article Neuroscience

Mitra PP, Sen PN, Schwartz LM. 1993. Short-time behavior of the diffusion coefficient as a geometrical probe of
porous media. Physical Review B 47:8565–8574. DOI: https://doi.org/10.1103/PhysRevB.47.8565
Mollink J, Kleinnijenhuis M, Cappellen van Walsum A-V, Sotiropoulos SN, Cottaar M, Mirfin C, Heinrich MP,
Jenkinson M, Pallebage-Gamarallage M, Ansorge O, Jbabdi S, Miller KL. 2017. Evaluating fibre orientation
dispersion in white matter: comparison of diffusion MRI, histology and polarized light imaging. NeuroImage
157:561–574. DOI: https://doi.org/10.1016/j.neuroimage.2017.06.001
Moseley ME, Cohen Y, Mintorovitch J, Chileuitt L, Shimizu H, Kucharczyk J, Wendland MF, Weinstein PR. 1990.
Early detection of regional cerebral ischemia in cats: comparison of diffusion- and T2-weighted MRI and
spectroscopy. Magnetic Resonance in Medicine 14:330–346. DOI: https://doi.org/10.1002/mrm.1910140218
Murday JS, Cotts RM. 1968. Self-diffusion coefficient of liquid lithium. The Journal of Chemical Physics 48:4938–
4945. DOI: https://doi.org/10.1063/1.1668160
Neuman CH. 1974. Spin Echo of spins diffusing in a bounded medium. The Journal of Chemical Physics 60:4508–
4511. DOI: https://doi.org/10.1063/1.1680931
Nilsson M, Lätt J, Ståhlberg F, Westen D, Hagslätt H. 2012. The importance of axonal undulation in diffusion MR
measurements: a monte carlo simulation study. NMR in Biomedicine 25:795–805. DOI: https://doi.org/10.1002/
nbm.1795
Novikov DS, Jensen JH, Helpern JA, Fieremans E. 2014. Revealing mesoscopic structural universality with
diffusion. PNAS 111:5088–5093. DOI: https://doi.org/10.1073/pnas.1316944111, PMID: 24706873
Novikov DS, Veraart J, Jelescu IO, Fieremans E. 2018. Rotationally-invariant mapping of scalar and orientational
metrics of neuronal microstructure with diffusion MRI. NeuroImage 174:518–538. DOI: https://doi.org/10.1016/
j.neuroimage.2018.03.006
Novikov DS, Fieremans E, Jespersen SN, Kiselev VG. 2019. Quantifying brain microstructure with diffusion MRI:
theory and parameter estimation. NMR in Biomedicine 32:e3998. DOI: https://doi.org/10.1002/nbm.3998
Nunes D, Cruz TL, Jespersen SN, Shemesh N. 2017. Mapping axonal density and average diameter using non-
monotonic time-dependent gradient-echo MRI. Journal of Magnetic Resonance 277:117–130. DOI: https://doi.
org/10.1016/j.jmr.2017.02.017
Olivares R, Montiel J, Aboitiz F. 2001. Species differences and similarities in the fine structure of the mammalian
corpus callosum. Brain, Behavior and Evolution 57:98–105. DOI: https://doi.org/10.1159/000047229
Ong HH, Wright AC, Wehrli SL, Souza A, Schwartz ED, Hwang SN, Wehrli FW. 2008. Indirect measurement of
regional axon diameter in excised mouse spinal cord with q-space imaging: simulation and experimental
studies. NeuroImage 40:1619–1632. DOI: https://doi.org/10.1016/j.neuroimage.2008.01.017
Ong HH, Wehrli FW. 2010. Quantifying axon diameter and intra-cellular volume fraction in excised mouse spinal
cord with q-space imaging. NeuroImage 51:1360–1366. DOI: https://doi.org/10.1016/j.neuroimage.2010.03.
063
Özarslan E, Yolcu C, Herberthson M, Knutsson H, Westin C-F. 2018. Influence of the size and curvedness of
neural projections on the orientationally averaged diffusion MR signal. Frontiers in Physics 6:17. DOI: https://
doi.org/10.3389/fphy.2018.00017
Packer KJ, Rees C. 1972. Pulsed NMR studies of restricted diffusion. I. droplet size distributions in emulsions.
Journal of Colloid and Interface Science 40:206–218. DOI: https://doi.org/10.1016/0021-9797(72)90010-0
Palombo M, Ligneul C, Najac C, Le Douce J, Flament J, Escartin C, Hantraye P, Brouillet E, Bonvento G, Valette
J. 2016. New paradigm to assess brain cell morphology by diffusion-weighted MR spectroscopy in vivo. PNAS
113:6671–6676. DOI: https://doi.org/10.1073/pnas.1504327113, PMID: 27226303
Palombo M, Ianus A, Nunes D, Guerreri M, Alexander DC, Shemesh N, Zhang H. 2019. SANDI: a compartment-
based model for non-invasive apparent soma and neurite imaging by diffusion MRI. arXiv. https://arxiv.org/abs/
1907.02832.
Raffelt D, Tournier J-D, Rose S, Ridgway GR, Henderson R, Crozier S, Salvado O, Connelly A. 2012. Apparent
fibre density: a novel measure for the analysis of diffusion-weighted magnetic resonance images. NeuroImage
59:3976–3994. DOI: https://doi.org/10.1016/j.neuroimage.2011.10.045
Reisert M, Kellner E, Dhital B, Hennig J, Kiselev VG. 2017. Disentangling micro from mesostructure by diffusion
MRI: a bayesian approach. NeuroImage 147:964–975. DOI: https://doi.org/10.1016/j.neuroimage.2016.09.058
Rushton WAH. 1951. A theory of the effects of fibre size in medullated nerve. The Journal of Physiology 115:
101–122. DOI: https://doi.org/10.1113/jphysiol.1951.sp004655
Sampaio-Baptista C, Diosi K, Johansen-Berg H. 2019. Oligodendrocytes. In: Lyons D. A, Kegel L (Eds). Magnetic
Resonance Techniques for Imaging White Matter. Springer. p. 397–407.
Schüz A, Preibl H. 1996. Basic connectivity of the cerebral cortex and some considerations on the corpus
callosum. Neuroscience & Biobehavioral Reviews 20:567–570. DOI: https://doi.org/10.1016/0149-7634(95)
00069-0, PMID: 8994195
Sepehrband F, Alexander DC, Clark KA, Kurniawan ND, Yang Z, Reutens DC. 2016. Parametric probability
distribution functions for axon diameters of corpus callosum. Frontiers in Neuroanatomy 10:1–9. DOI: https://
doi.org/10.3389/fnana.2016.00059
Shemesh N, Rosenberg JT, Dumez J-N, Grant SC, Frydman L. 2017. Distinguishing neuronal from astrocytic
subcellular microstructures using in vivo double diffusion encoded 1H MRS at 21.1 T. PLOS ONE 12:e0185232.
DOI: https://doi.org/10.1371/journal.pone.0185232, PMID: 28968410
Shepherd TM, Thelwall PE, Stanisz GJ, Blackband SJ. 2009. Aldehyde fixative solutions alter the water relaxation
and diffusion properties of nervous tissue. Magnetic Resonance in Medicine 62:26–34. DOI: https://doi.org/10.
1002/mrm.21977

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 26 of 27

 Research article Neuroscience

Sijbers J, den Dekker AJ, Scheunders P, Van Dyck D. 1998. Maximum-likelihood estimation of rician distribution
parameters. IEEE Transactions on Medical Imaging 17:357–361. DOI: https://doi.org/10.1109/42.712125
Smith SM, Jenkinson M, Woolrich MW, Beckmann CF, Behrens TEJ, Johansen-Berg H, Bannister PR, De Luca M,
Drobnjak I, Flitney DE, Niazy RK, Saunders J, Vickers J, Zhang Y, De Stefano N, Brady JM, Matthews PM. 2004.
Advances in functional and structural MR image analysis and implementation as FSL. NeuroImage 23:S208–
S219. DOI: https://doi.org/10.1016/j.neuroimage.2004.07.051
Sotiropoulos SN, Behrens TEJ, Jbabdi S. 2012. Ball and rackets: inferring fiber fanning from diffusion-weighted
MRI. NeuroImage 60:1412–1425. DOI: https://doi.org/10.1016/j.neuroimage.2012.01.056
Stanisz GJ, Wright GA, Henkelman RM, Szafer A. 1997. An analytical model of restricted diffusion in bovine optic
nerve. Magnetic Resonance in Medicine 37:103–111. DOI: https://doi.org/10.1002/mrm.1910370115
Stejskal EO. 1965. Use of spin echoes in a pulsed magnetic-field gradient to study anisotropic, restricted
diffusion and flow. The Journal of Chemical Physics 43:3597–3603. DOI: https://doi.org/10.1063/1.1696526
Tang Y, Nyengaard JR, Pakkenberg B, Gundersen HJG. 1997. Age-Induced white matter changes in the human
brain: a stereological investigation. Neurobiology of Aging 18:609–615. DOI: https://doi.org/10.1016/S0197-
4580(97)00155-3
Tanner JE. 1979. Self diffusion of water in frog muscle. Biophysical Journal 28:107–116. DOI: https://doi.org/10.
1016/S0006-3495(79)85162-0
Tax M, Szczepankiewicz F, Jones DK. 2019. The dot-compartment revealed? diffusion MRI with ultra-strong
gradients and spherical tensor encoding in the living human brain. bioRxiv. DOI: https://doi.org/10.1101/
584730
Tournier J-D, Smith R, Raffelt D, Tabbara R, Dhollander T, Pietsch M, Christiaens D, Jeurissen B, Yeh C-H,
Connelly A. 2019. MRtrix3: a fast, flexible and open software framework for medical image processing and
visualisation. NeuroImage 202:116–137. DOI: https://doi.org/10.1016/j.neuroimage.2019.116137
van Gelderen P, Des Pres D, van Zijl PCM, Moonen CTW. 1994. Evaluation of restricted diffusion in cylinders.
Phosphocreatine in rabbit leg muscle. Journal of Magnetic Resonance, Series B 103:255–260. DOI: https://doi.
org/10.1006/jmrb.1994.1038, PMID: 8019777
Veraart J, Fieremans E, Novikov DS. 2016. Diffusion MRI noise mapping using random matrix theory. Magnetic
Resonance in Medicine 76:1582–1593. DOI: https://doi.org/10.1002/mrm.26059
Veraart J, Novikov DS, Fieremans E. 2018. TE dependent diffusion imaging (TEdDI) distinguishes between
compartmental T2 relaxation times. NeuroImage 182:360–369. DOI: https://doi.org/10.1016/j.neuroimage.
2017.09.030, PMID: 28935239
Veraart J, Fieremans E, Novikov DS. 2019. On the scaling behavior of water diffusion in human brain white
matter. NeuroImage 185:379–387. DOI: https://doi.org/10.1016/j.neuroimage.2018.09.075, PMID: 30292815
Veraart J, Novikov DS. 2019. Axon radius mapping. https://github.com/NYU-DiffusionMRI/AxonRadiusMapping
Waxman SG. 1980. Determinants of conduction velocity in myelinated nerve fibers. Muscle & Nerve 3:141–150.
DOI: https://doi.org/10.1002/mus.880030207
Wegiel J, Kaczmarski W, Flory M, Martinez-Cerdeno V, Wisniewski T, Nowicki K, Kuchna I, Wegiel J. 2018. Deficit
of corpus callosum axons, reduced axon diameter and decreased area are markers of abnormal development
of interhemispheric connections in autistic subjects. Acta Neuropathologica Communications 6:143.
DOI: https://doi.org/10.1186/s40478-018-0645-7
Xu J, Li H, Harkins KD, Jiang X, Xie J, Kang H, Does MD, Gore JC. 2014. Mapping mean axon diameter and
axonal volume fraction by MRI using temporal diffusion spectroscopy. NeuroImage 103:10–19. DOI: https://
doi.org/10.1016/j.neuroimage.2014.09.006
Zhang H, Schneider T, Wheeler-Kingshott CA, Alexander DC. 2012. NODDI: practical in vivo neurite orientation
dispersion and density imaging of the human brain. NeuroImage 61:1000–1016. DOI: https://doi.org/10.1016/j.
neuroimage.2012.03.072

Veraart et al. eLife 2020;9:e49855. DOI: https://doi.org/10.7554/eLife.49855 27 of 27