Review Article

Focus on Diffusion MR Investigations ofMusculoskeletal Tissue to Improve Osteoporosis Diagnosis:A Brief Practical Review

Silvia Capuani,1,2 Guglielmo Manenti,3 Riccardo Iundusi,2 and Umberto Tarantino2

1CNR-IPCF UOS Roma Sapienza, Physics Department, “Sapienza” University of Rome, Italy2Department of Orthopaedics and Traumatology, PTV Foundation, “Tor Vergata” University of Rome, Italy3Department of Diagnostic and Interventional Radiology, Molecular Imaging and Radiotherapy, PTV Foundation,“Tor Vergata” University of Rome, Italy

Correspondence should be addressed to Silvia Capuani; silvia.capuani@roma1.infn.it

Received 28 October 2014; Accepted 18 February 2015

Academic Editor: Jimin Ren

Copyright © 2015 Silvia Capuani et al. Gis is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Nowadays, a huge number of papers have documented the ability of diMusion magnetic resonance imaging (D-MRI) to highlightnormal and pathological conditions in a variety of cerebral, abdominal, and cardiovascular applications. To date, however, the roleof D-MRI to investigate musculoskeletal tissue, speciUcally the cancellous bone, has not been extensively explored. In order todetermine potentially useful applications of diMusion techniques in musculoskeletal investigation, D-MRI applications to detectosteoporosis disease were reviewed and further explained.

1. Introduction

Since the early publications by Le Bihan [1] and Basser [2–4] a great number of papers have documented the ability ofdiMusionmagnetic resonance imaging (D-MRI) to deUne anddiscriminate normal and pathological conditions in a varietyof cerebral [5, 6], abdominal [7–10] and cardiovascularapplications [11, 12]. Ge enormous importance of D-MRItechniques in clinical diagnostics is due to the fact thatbiological water in tissues is an endogenous molecular probeand its diMusion re^ects tissue conUguration at a microscopiclevel. Gerefore, diMusion magnetic resonance (MR) imagescanmeasurewater proton displacements, within and betweentissues, by probing molecules motion on the micrometerlength scale which is orders of magnitude smaller than themacroscopic MR image resolution (usually 1–3 millimeters,in clinical applications).

To date, however, the potential role of D-MRI in thediagnosis of musculoskeletal disorders has not been exten-sively explored [13]. Gis is due to diMerent causes of variousnature that we have focused and described here, on the

base of our experience as interdisciplinary group consistingof academics researchers in physics, radiologist clinicians,orthopedic surgeons and clinicians operating in the Ueld ofmusculoskeletal diseases.

In the past, technical restriction limited the use of D-MRI techniques to the brain [5, 6, 12]. However, during thelast 10 years, with the improvement in scanners technologyand the availability of new MR sequences, investigation ofthe musculoskeletal system was made possible [14–23]. Byusing “water diMusion imaging” and “musculoskeletal” as keywords for literature search, about twenty reviews are found.Some of these are focused on D-MRI for diMerentiation ofacute benign and neoplastic vertebral compression fractures[14–16], painful spinal disorders [17], musculoskeletal tumorimaging [18, 19], and spine, muscles, and peripheral nervesdisorders [20]. Some others are focused on D-MRI in body[21] and in pediatric radiology [22, 23]. However, noneof these reviews is speciUcally focused on water diMusionMRI to detect osteoporosis. On the other hand, by usingthe following key words for literature search, “water diMu-sion imaging” and “osteoporosis,” only one relevant review,

Hindawi Publishing CorporationBioMed Research InternationalVolume 2015, Article ID 948610, 10 pageshttp://dx.doi.org/10.1155/2015/948610

2 BioMed Research International

published in 2010, is obtained [14]. In this review, only veryfew studies about vertebral marrow di@usion in osteoporosisare reported, which show contradictory results.

Recently, in parallel with technological development, anew point of view about the study of the bone marrow waterin cancellous bone has been introduced. Bis new point ofview provides an explanation of the previous contradictoryresults extending the water di@usion investigation to otherskeletal sites in addition to the vertebral one.

From the above considerations, the need for this review,which includes and explains the previous and the last recentresults on the basis of simple and intuitive considerationsregarding the biophysicalmechanismofwater di@usion in thetissues. Recommendations and suggestions about practicalsteps of optimization closely related to the microstructuralcharacteristics of the di@erent skeletal sites are also reported.

Osteoporosis is a systemic and metabolic disorder char-acterized by a progressive reduction of mineral bone massand microarchitectural deterioration of bone tissue, whichincreases the risk of bone fractures [24, 25]. Be clinicaldiagnosis of osteoporosis is currently based on the quantiK-cation of bone mineral density (BMD) performed by dual-energy X-ray absorptiometry (DXA) [26]. Although DXAexamination is economical and noninvasive, it is knownto have reduced sensitivity and a low predictive value onpatients’ risk of reporting bone fracture [27, 28]. Bis lack ofsensitivity is likely to be due to the partial information thatBMD provides on cancellous bone characteristics, assessingexclusively its mineral component [27, 29]. Indeed, bonestrength depends on BMD and bone quality [25, 30, 31].Bone quality refers to topological properties of trabecularmicrostructure, bone turnover, and composition of bonemarrow.Bese and other factors, such as collagen frameworkand cell viability [32], may contribute to determining bonestrength and its resistance to fracture [29]. In this regard,magnetic resonance MR techniques allow investigation ofboth trabecular networks and bone marrow [25, 33–37]providing some additional information on the physiologicaland functional changes associated with osteoporosis [29].

In general, D-MRI techniques are more challengingwhen applied to extracranial tissues due to technical issuesmainly related to magnetic susceptibility variation Δ� andmotion sensitivity [13]. In particular, water di@usion in bonemarrow and cancellous bone is a@ected by strong internalmagnetic Keld gradients ( �) generated by Δ� [38, 39] andmainly localized at the interface between bone and watercomponents [40].

Be setting, optimization, and operation of D-MRI tech-niques in cancellous bone and muscle investigations aremarkedly di@erent from conventional MRI of the muscu-loskeletal system primarily based on image contrast due todi@erences in tissues relaxation times: !1, !2, and !∗2 [41,42]. Berefore, D-MRI applications demand a greater levelof knowledge of the di@usion fundamentals, attention, andforethought. Currently, thanks to the development of highperformance magnetic Keld gradients, new di@usion sensi-tized sequences, and parallel imaging techniques, the maintechnical problems are drastically reduced. Moreover, a newpoint of view on human cancellous bone [40, 43], described

as a porous system in which water experiences a regime ofrestricted di@usion in the space between fat andmineral bonein each pore, forces to develop new MRI protocols based onmolecular di@usion of water to obtain new microstructuralinformation for improving osteoporosis diagnosis.

Be basic principles of di@usion and its assessment ofD-MRI in musculoskeletal tissue were summarized here,together with a brief description about microstructural char-acteristics of musculoskeletal tissues. Subsequently, somestrategies and suggestions to optimize D-MRI protocols fortheir potential applications in the diagnosis of osteoporosiswere described.

2. Microstructural Characteristics ofMusculoskeletal Tissues

Be principal musculoskeletal tissues that are related toosteoporosis and can be investigated with D-MRI are bones[33–38], bone marrow, muscles, and fat [44–48].

Bone tissue is a complex biomaterial composed of asolid mineral matrix, which is Klled by bone marrow (a so[interstitialmaterial).Be solidmatrix ismainly constituted bymineral components while bonemarrow ismainly composedby water and fat at di@erent relative percentages. Be relativeconcentration of each bonemarrow component is dependenton both its anatomical skeletal location and the age ofconsidered individuals [49]. Human bone may be classiKedas cortical or trabecular. Be former is mainly present inthe sha[ of long bones and is much denser (with a porosityranging from 5% to 10%) than cancellous bone. Conversely,cancellous bone is much more porous (with a porosityranging from 50% to 90%), and it is metabolically moreactive. From a physical point of view, cancellous bone is aporous system that can be described as a solid with holes andcavities, that is, presenting connected void spaces, randomlydistributed within a solid matrix. Moreover, water is moreprevalent in the boundary zone while fat occupies primarilythe central zone of each pore, as recently demonstrated [40].Water in the boundary zones of pores is also due to the pres-ence of the endosteum, a thin membrane (with size about 3–7 "m) of so[ tissue (principally water) comprised of a linearchain of cells that lines the medullary cavity [50]. Moreover,due to a biological division of the bonemarrow compartment,granulocytes and other nonfat entities accumulate at theboundary of the bone marrow compartment adjacent tothe endosteum. Pore sizes in human cortical bone are inthe range of several nanometers to few micrometers, whilepore sizes in cancellous bone are in the range 1"m–1mm,where intertrabecular plates space ranges approximately from100 "m to 1-2mm. On the other hand, the space betweenfat and bone in the boundary of each pore ranges fromless than 1 "m (when endosteum is damaged or absent dueto pathologies) to 10–20"m [50] or more, in the case oftrabecular disruption or bone marrow irradiation. Bonemarrow located inside the central cavities of long bones is notforced in pores. From a microstructural point of view, bothbonemarrow in trabecular bone and bonemarrow free in thelarger bone cavities are so[ tissues characterized by severalparticles of spherical/elliptical shape and average size ranging

BioMed Research International 3

from 6 microns of red blood cells to approximately 100microns of the fat globules. Adipose tissue is characterized byapproximately the same kind of bonemarrow fat [51]. Skeletalmuscles are attached to and bring about the movement of thevarious bones of the skeleton. Fe whole muscle is enclosedin a sheath of connective tissue that folds inwards into thesubstance of themuscle to surround a large number of smallerbundles, called fasciculi.Fese fasciculi consist of still smallerbundles of elongated, cylindrical cells named Hbres that arecharacterized by diameter size of about 10–20 �m.

With the development of osteoporosis, a change of theaforementioned microstructures size occurs. Bone microar-chitecture is rearranged or disrupted (such as conversionof plates to rods and loss of connectivity) [41, 52, 53] andbone marrow which Hlls the trabecular holes is altered [44–46, 54]. SpeciHcally, an increase of the marrow fat thatdecreases the interstitial space size in which water diPuse anda microstructural deterioration with a pore enlargement thatincreases the interstitial space size and their interconnectionoccur [43]. Fese changes do not result equally along allskeletal sites. As an example, in calcaneus, with the onsetof osteoporosis, no bone marrow increase is detected butonly a trabecular bone deterioration is observed [43, 55].Moreover, recent investigations show a correlation betweenthe cancellous bone deterioration and the loss of muscle masswith a change of Hbres diameter size, referred to as sarcopenia[47, 48].

3. Magnetic Resonance Diffusion:Introduction and Concepts

DiPusion is the process by which matter is transported fromone part to another as a result of random molecular motion,also called Brownian motion [56]. Fis motion results fromthe thermal energy carried by diPusing molecules and itis characterized by a diPusion coeUcient that providesinformation about the behavior of the molecular motionwhich in turn depends on the characteristics of the mediumin which the diPusion occurs. Ferefore, measurementof water molecules in human tissues, as a consequence ofthe interactions between molecules and obstacles that hinderand or restrict their motion, gives information about size,orientation, and shape of cellular structures [1, 3].

Currently, D-MRI is the only means able to measure andmonitor the diPusion coeUcient in vivo, with a completelynoninvasive modality and without requiring exogenouscontrast agents. MRI measurements of diPusion indirectlymeasure the displacements of diPusing molecules in onedimension.

In Gaussian diPusion approximation, the mean squaredisplacement (MSD) of diPusing molecules is linearly pro-portional to the diPusion coeUcient and the time ! duringwhich the diPusion process is observed [56]:

MSD = 2" !, (1)

where " is the “dimensionality” of the space over whichdiPusion distances are measured.

By using relation (1), it is possible to evaluate the lengthscale # or the characteristic length probed by diPusingmolecules:

# = MSD1/2. (2)

By using relation (1) and (2) at human body temperature, dur-ing typical diPusion times of about ! = 50ms, random watermolecules move in a homogeneous and isotropic mediumover distances around # = 30 �m in the tridimensional case(i.e., " = 3) and around # = 15 �m along a chosen spatialdirection (i.e., " = 1). However, this is the case of freediPusion of bulk water, for which there are no barriers tomolecules motion and ≈ 3 ∗ 10−3mm2/s.

Because most human cells and subcellular structureshave smaller dimensions than those listed above, it is highlylikely that water molecules will have many interactions withcellular components over such ! measurement interval [3].As a consequence, water movement in tissues is neitherentirely free nor random, beingmodiHed by interactions withhydrophobic lipid-containing cell membranes, intracellularorganelles, and macromolecules and by ]ows within tubularchannels such as blood vessels, capillaries, interconnectedpores, and ducts. Water motion in tissue is therefore relatedto its microscopic and submicroscopic biological structure.As depicted in Figure 1, by using relation (1) and (2) andmeasuring , it is possible to detect diPerent water diPusionbehavior, such as free, hindered, and restricted diPusion,in structures characterized by diPerent size, and, moreimportantly, it is possible to estimate microstructures size.SpeciHcally, if water diPusion is limited by a container (suchas water in cortical bone pores, or intracellular water), whenthe molecules reach the boundaries, they are re]ected andthe MSD results conHned in the container [56]. As a conse-quence, both and # are lower than those of free bulk water.4. Magnetic Resonance Diffusion

Measurement: DWI and ADC

Fe molecular diPusion is uniquely assessed by diPusionweighted imaging (DWI) techniques [1, 3]. DWI techniquesare able to measure the water diPusivity by the application ofa couple of diPusion sensitizing gradients (motion probinggradients) to a simple &2-weighted spin-echo sequence,named pulse Held gradient spin-echo (PFGSE) [57]. Signallosses on DWI are proportional to both diPusion coeUcientof water molecules and the diPusion gradient strength used,as described by the signal attenuation '()) of DWI, given by

' ()) = ' (0) exp (−) ) , (3)

where '(0) is the signal amplitude in absence of diPusinggradients and b is the so-called )-value that indicates thestrength and duration of application of diPusion sensitizinggradients together with the diPusion time ! during which thediPusion phenomenon is observed:

) = (/46)2 (Δ − 43) , (4)

4 BioMed Research International

7

t (ms)

9

4

80

4515

Free water

Hindered water

Hindered water

Restricted di8usion 9!m

Restricted di8usion 3!m

Restricted di8usion 2!m

MS

D (!

m2)

Figure 1: Schematic representation of water MSD as a function ofthe di@usion time �, obtained in a heterogeneous system depicted atthe bottom. Solid lines represent curves obtained by using relation(1). In the case of restricted di@usion, the two solid lines representcurves obtained considering data at short and long times.

where is the gyromagnetic ratio, ! is the strength of thegradient (di@usion gradient) used to encode di@usion, " isthe duration of the gradient pulse, and Δ, which is the timebetween the leading edges of the pair of di@usion gradientpulses in PFGSE, is the di@usion time � of relation (1).In the practice, the $-value which determines the degreeof signal attenuation due to di@usion allows the selectionof di@erent di@usion behaviors characterized by di@erentdi@usion coeNcients of water di@using in heterogeneoussystems (this issue will be discussed in paragraph 7).

By observing Figure 1, it is possible to understand thatif the di@usion process is observed for time � which is tooshort, the di@using water molecules do not reach obstaclesand walls of pore; in this case, a free di@usion behavior and afree di@usion coeNcient are detected.Qerefore, an adequatedi@usion time must be selected to obtain microstructuralinformation from DWI investigations. As an example, todetect the presence of pores characterized by pore size ofabout 9%m, � greater than 45ms must be used with & ≈3∗10−4mm2/s, as shown in Figure 1.Moreover, free di@usionof water at human body temperature shows the highest

& value approximately equal to 3 ∗ 10−3mm2/s. Hinderedand restricted water di@usion assume smaller values withincreasing of obstacles and with the decreasing of pores andstructures size within which the water is conZned.

Because in anisotropic and heterogeneous systems di@u-sion coeNcient depends on both the direction along whichthe di@usion gradient acts and the di@usion time � = Δ, DWIin tissues refers to ADC, the apparent di@usion coeNcient,rather than &, the self-di@usion constant of a pure isotropicand homogeneous solution. By using relation (3), the ADCcan be computed in each voxel of DW image to obtain anADC map according to the following:

ADC = −(1$) ln [- ($)- (0)] . (5)

5. Diffusion Tensor Imaging: DTI

Molecular mobility in tissues which are in general heteroge-neous and anisotropic is not the same in all directions [2–4].As an example, water di@usion along muscles Zbers is lesshindered than that along directions perpendicular to Zbers[58]. In the presence of anisotropy, di@usion can no longerbe completely characterized by a single scalar coeNcientbut required a tensor, &, which fully describes molecularmobility along each direction and correlation between thesedirections. Because & is symmetric, only six ADC measure-ments along six independent directions are needed, andrelation (3) is now expressed by the following [2]:

- (&) = - (0) exp (−$ & − $!!&!! − $""&""

− 2$ !& ! − 2$ "& " − 2$!"&!") ,(6)

where, the on-diagonal terms & , &!!, and &"" representthe ADC along axes ;, <, and >, respectively, while the o@-diagonal terms& !,& ", and&!" recect correlation betweenmolecular displacements in perpendicular directions [2–4].

In practice, to determine &, one must Zrst collect DWimages along at least six independent directions, by usingdi@usion sensitized MRI pulse sequences with di@usiongradients along six independent directions, such as ;, <, >,;<, ;>, and ><, plus a spin-echo image that provides the-(0) term in relation (6). Qerefore, DTI protocol lasts sixtimesmore than the DWI protocol; however, from the tensor,several pieces of information can be extracted. Qe sun ofthe on-diagonal elements is rotationally invariant measure ofdi@usion, that is, a measure of di@usion independent on the

BioMed Research International 5

reference frame. 4is is very important in diagnostic appli-cations because di<usion quanti>cation obtained at di<erenttimes in a given patient or in di<erent patients and healthysubjects or in di<erent hospitals can be compared. One-thirdof the trace is oAen referred to as the mean di<usivity (MD)and constitutes a directionally averaged di<usion coeEcientthat can also be obtained by averaging the ADC measuredalong �, , and ! directions.4e fractional anisotropy (FA), asort of normalized standard deviation, measures the fractionof the magnitude of " that can be ascribed to anisotropicdi<usion. FA varies between 0 (isotropic di<usion) and 1(maximum anisotropy). As an example, biceps femoris shorthead in a healthy subject is characterized by FA= 0.4 [59], butthe same tissue in the presence of edema shows FA values lessthan 0.3.

6. D-MRI Pulses Sequences

Di<usion gradients can be applied to many standard MRIsequences, but, particularly in musculoskeletal applications,the images produced are prone to artifacts resulting fromphysiological motions, eddy currents, and internal gradientsdue to Δ$. Spin-echo DWI sequences such as pulse gra-dient spin-echo (PGSE) [57] or pulse gradient stimulatedecho (PGSTE) [60] that is particularly useful to investigatedi<usion in tissues characterized by %2 ≪ %1 (such as themost musculoskeletal tissues) yield high signal-to-noise ratio(SNR) images and are more resistant to >eld inhomogeneitycreated by the application of di<usion gradients. However,long acquisition times of several minutes limit their clinicaluse due to the increased likelihood of encountering motion-related artifacts.

Because the reliability of di<usion contrast depends onSNR of DWI images [61], it must be approximately higherthan 2 [62] to exclude an underestimation of the MD and anoverestimation of the anisotropy quanti>ed by FA [62, 63].

Single shot echo planar imaging (SS-EPI) sequences o<ersigni>cantly faster acquisition times, thus overcoming DWIsensitivity to patientmovement while still o<ering a relativelyhigh SNR.

Amajor limitation of di<usion sensitized SS-EPI imagingis magnetic susceptibility artifacts, at di<erent tissue inter-faces common in musculoskeletal imaging, such as boneand soA tissue, which can lead to severe geometric imagedistortion, relatively low spatial resolution due to rapid %∗2decay of signal, and in general low image quality.

Recent advances in SS-EPI to overcome these problemsinclude improved gradient systems with reduced eddy cur-rent e<ects which can help reduce geometric distortions andparallel imaging techniques employing multiple receiver coilelements to improve acquisition times and increase spatialresolution.

An alternative approach is to acquire EPI data usinga segmented or multishot, echo planar readout (MS-EPI),which divides the echo train into several shorter parts. 4ismakes the image less sensitive to susceptibility artifacts,reduces image distortion, and increases spatial resolution butat the price of longer acquisition times and therefore at agreater risk of motion artifact.

Recently [64], by using a di<usion sensitized segmentedEPI with EPI factor equal to 7 and a scan time duration for theDTI protocol equal to 17:50 minutes, we have obtained the>rst MD and FA maps of femoral neck performed in threedi<erent groups of women based on their BMD (healthy,osteopenic, and osteoporotic women) [64]. Each DW image,acquired by using four averaged scans, was characterized by asuEcient SNR to extract signi>cant values ofMD and FA andan in-plane resolution of about 2.2mm with a slice thicknessof 5mm [64]. Subsequently, the same di<usion sensitizedsegmented EPI sequence has been used to evaluate ADC incalcanei of healthy and osteoporotic subjects [43, 55].

7. D-MRI Parameters for Optimization

4e choice of pulse sequence parameters has a profoundinbuence on the accuracy and precision in the estimatesADC, MD, and FA and other di<usion metrics parameters.Currently, the most of these sequence parameters are alreadypreset in clinical scanner to perform brain, spinal cord, ormuscles investigations. For example, it is well known that

DTI protocol with ' = 1000 s/mm2 is used in conventionalneurological clinical applications [6, 7, 12], while ' = 500–

800 s/mm2 optimizes di<usionmeasurement inmuscles [65–67] and spinal cord [68]. Conversely, no suggestions can beusually found to performdi<usion investigation in cancellousbone.

In principle, the optimal '-value is approximately equal tothe reciprocal of ADC (or MD) that one is trying to estimate:

' ≈ 1ADC

. (7)

However, the best compromise between a suEcient SNR anda suEcient di<usion weighted contrast to detect di<usion ofbone marrow water restricted between fat and bone must bereached.

Bone marrow fat extracted from the vertebrae, thefemoral neck, and the calcaneus of healthy postmenopausalwomen was found to range from 50% to 70%, 60% to 88%,and 78% to 98%, respectively [40, 44, 45, 55, 64, 69–71]. Anincrease in the amount of bonemarrow fat involves a decreaseof the interstitial space between bone and fat (region in whichwater di<uses), leading to a lower water ADC. As a con-sequence, water ADC is higher in the vertebrae, intermediatein femoral neck, and lower in the calcaneus; that is, di<usionof water in the calcaneus ismore restricted compared towaterin femoral neck or in vertebrae, as displayed in Figure 2.

4erefore, these three di<erent skeletal sites must beprobed with di<erent length scale (because they are charac-terized by di<erent pores size) and di<erent '-values, becausethey are characterized by di<erent water ADC. Speci>cally,ADC values of bone marrow water measured in humancancellous bone of healthy subjects showed the following

approximate values: ADC ≈ 10−3mm2/s, ADC ≈ 4 ∗10−4mm2/s, andADC ≈ 5∗10−5mm2/s for the vertebrae, thefemoral neck, and the calcaneus, respectively [40, 43, 55, 64,70, 72, 73]. In principal, by using relation (7), '-values around' = 1000 s/mm2, ' = 2500 s/mm2, and ' = 20000 s/mm2

6 BioMed Research International

Healthy cancellous bone

Osteoporotic cancellous bone

ADC

ADC

Water

Bone

Fat

Vertebrae

ADC ≈ 10−3 mm2/s

Bone

FatW

Femoral neck

ADC ≈ 10−4 mm2/s

Bone

Fat

W

Calcaneus

ADC ≈ 10−5 mm2/s

Figure 2: Schematic representation of cancellous bone at di?erent skeletal sites in healthy (top) and osteoporotic (bottom) cancellous bone.Fat content increases from axial skeleton to peripheral skeletal sites. As a consequence, water ADC is higher in the vertebrae and lowerin the calcaneus. Indeed, di?usion of water in calcaneus is more restricted compared to water in femoral neck or in vertebrae. With thedevelopment of osteoporosis, two di?erent and opposite mechanisms occur: an increase of the marrow fat which decreases the ADC valueand a microstructure deterioration with a pore enlargement that increases ADC. In the calcaneus, only the last mechanism results (adaptedfrom reference [43] with permission).

should be used to measure ADC in vertebrae, femoral neck,and calcaneus, respectively. In reality, most tissues are madeof multiple subcompartments (for example, the intra- andextracellular water compartments); each of these is charac-terized by a peculiar ADC. As a consequence, the ADC thatis measured could depend on the range used for the �-values,as measurements with low �-values would be more sensitiveto fast (near free or less hindered) di?usion components andmeasurements with high �-values would be more sensitive toslow (more restricted and entrapped) di?usion components.In the ideal case, a very large range of � values wouldbe required. However, this means long acquisition timesincompatible with clinical applications. In any case, it is clearthat it is better to estimate the averaged ADC in a tissue byusing a range of �-values (from four to eight) and a Wttingprocedure of experimental data to relation (3) rather thanestimate it using a single �-value.

As previously underlined, in cancellous bone, morethan in other tissues, the magnetic susceptibility mismatchbetween the solid matrix and interstitial water bone marrowgenerates internal gradients of magnitude depending on thegeometry and orientation of the trabeculae with respect tothe static magnetic Weld direction scaling with magnetic Weldstrength. Xerefore, the � strength that water moleculessense as they di?use near the interface between fatty marrowand bone increases with the decreasing of the interstitialspace and with the increasing of the static magnetic Weld

[40, 55, 74]. In particular, when cancellous bone in calcaneusis investigated, a positive coupling between �� and di8usiongradients occurs that produce an e8ective -value higherthan that computed by using relation (4). Finally, it is wellknown that the di8usive motion of water molecules throughthe �� leads to irreversible dephasing [40, 55, 74–77]. As aconsequence, the e8ective di8usion coeGcient ACDe! (mea-sured by using a DWI protocol) results in smaller than theideal ADC obtained without taking into account the Δ" atthe interfaces between water and bone [55, 78, 79]. Becauseof the above considerations, particularly in calcaneus D-MRIinvestigations, water ADC is underestimated and the use of

a -value range up to approximately 10000 s/mm2 should beappropriate.

By considering relation (3) and (4), in order to achievesigniTcant signal attenuation when the di8usion gradientsare applied, the pulses must be strong. However, this placesstrong demands on the magnetic Teld gradient hardware.Verefore, D-MRI investigations in cancellous bone requiredhigh gradient unit performance of MRI scanner. Ve choiceto select di8usion gradients applied for a long time (i.e., long#) to increase di8usion gradients strength is not applicablein cancellous bone and muscles investigations. Indeed, waterin bone marrow and muscles is characterized by lower $2,compared to other tissues (such as brain tissue). Verefore,long # can lead to extended echo times, which consequentlydetermines poor SNR in the resulting DW images. Moreover,

BioMed Research International 7

due to the reduced �2 of water in musculoskeletal tissues, theuse of PGSTE (which requires Δ < �1) rather than PGSE(which requires Δ < �2) sequences would be preferred.

Be !-value includes both diEusion time Δ and diEusiongradients strength ". Berefore, one can obtain certain !-value ranges by Hxing the diEusion timeΔ and varying" or byHxing " and varying Δ. Bese two modalities do not provideequal investigations and equal results. Bis is because Δ alsodeHnes the length scale throughwhich a tissue is probedwhile" only determines the degree of signal attenuation due todiEusion.

Recently [43, 55], ADC values of bone marrow watermeasured in calcaneus showed an increasing trend from

healthy (ADC ≈ 4 ∗ 10−11mm2/s) to osteoporotic subjects(ADC ≈ 7 ∗ 10−11mm2/s) and a signiHcant negative corre-lation between ADC and BMD values.Bis study, performed

with ! = 8000 s/mm2 and Δ = 30ms, suggests that by using% ≈ 3 &m healthy, osteopenic, and osteoporotic subjects canbe signiHcantly discriminated. Bis is most likely due to thefact that the spaces sizes between fat and bone (where thewater diEuses) are around 3 &m and they increase with thedevelopment of osteoporosis, according to the ADC increase[43, 55]. On the other hand, healthy and osteoporotic subjects

were signiHcantly discriminated by selecting ! = 2500 s/mm2

andΔ = 40ms, to probe femoral neck with % ≈ 10 &m [64]. Inthis DTI study, both MD and FA values obtained in femoralneck showed a decreasing trend from healthy to osteoporoticsubjects. Bese results suggest that, with the development ofthe osteoporosis, the space between fat and bone probed bywater decreases due to marrow fat increases and it becomesmore isotropic due to the structural rearrangement of theendosteum cell lines or of the trabeculae surface [64].

8. Conclusion

D-MRI in cancellous bone provides a new tool to probe tissuemicrostructures changes that occur with the developmentof osteoporosis. Bis is because at microscopic level manytissue features related to osteoporosis inXuence MR diEusionmeasurements.

Currently, diEusion in musculoskeletal tissue is quitediYcult to use successfully, principally because of the occur-rence of strong internal gradients due to magnetic sus-ceptibility mismatch between soZ tissue and mineral boneand short �2. Bese obstacles, however, can sometimes beovercome with ad hoc optimized pulse sequences, hardware,and a theoretical background of diEusion process.

Abbreviations by Order of Appearancein the Paper

MR: Magnetic resonanceMRI: Magnetic resonance imagingD-MRI: DiEusion magnetic resonance imagingBMD: Bone mineral densityDXA: Dual-energy X-ray absorptionΔ': Magnetic susceptibility variation(!: Internal magnetic Held gradientsMSD: Mean square displacement

): DiEusion coeYcient%: Length scaleDWI: DiEusion weighted imagingDW: DiEusion weightedPFGSE: Pulse Held gradient spin-echo!: !-value*: Gyromagnetic ratio": Strength of the magnetic Held gradient+: Duration of the magnetic Held gradient

pulseΔ: DiEusion time: the time between the pair

of diEusion gradient pulsesADC: Apparent diEusion coeYcientDTI: DiEusion tensor imaging): DiEusion tensorMD: Mean diEusivityFA: Fractional anisotropyPGSE: Pulse gradient spin-echoPGSTE: Pulse gradient stimulated echo�2: Transverse relaxation time�1: Longitudinal relaxation timeSNR: Signal-to-noise ratioEPI: Echo planar imagingSS-EPI: Single shot echo planar imagingMS-EPI: Multishot echo planar imaging.

Conflict of Interests

Be authors declare that there is no conXict of interestsregarding the publication of this paper.

Acknowledgment

Bis work was supported by Agenzia Spaziale Italiana (Asi)“Studio multidisciplinare degli eEetti della microgravita sullecellule ossee” (SMEMCO).Call for research n.DC-DTE-2011-033.

References

[1] D. Le Bihan, “Molecular diEusion nuclear magnetic resonanceimaging,” Magnetic Resonance Quarterly, vol. 7, no. 1, pp. 1–30,1991.

[2] P. J. Basser, J. Mattiello, and D. LeBihan, “MR diEusion tensorspectroscopy and imaging,” Biophysical Journal, vol. 66, no. 1,pp. 259–267, 1994.

[3] P. J. Basser, “Inferring microstructural features and the physi-ological state of tissues from diEusion-weighted images,” NMRin Biomedicine, vol. 8, no. 7, pp. 333–344, 1995.

[4] P. J. Basser and C. Pierpaoli, “Microstructural and physiologicalfeatures of tissues elucidated by quantitative-diEusion-tensorMRI,” Journal of Magnetic Resonance Series B, vol. 111, no. 3, pp.209–219, 1996.

[5] D. Le Bihan, R. Turner, P. Douek, and N. Patronas, “DiEusionMR imaging: clinical applications,” 9e American Journal ofRoentgenology, vol. 159, no. 3, pp. 591–599, 1992.

[6] M. A. HorsHeld and D. K. Jones, “Applications of diEusion-weighted and diEusion tensor MRI to white matter diseases—areview,”NMR in Biomedicine, vol. 15, no. 7-8, pp. 570–577, 2002.

8 BioMed Research International

[7] D.-M. Koh and D. J. Collins, “Di<usion-weighted MRI in thebody: applications and challenges in oncology,” American Jour-nal of Roentgenology, vol. 188, no. 6, pp. 1622–1635, 2007.

[8] M. A. Haider, T. H. van der Kwast, J. Tanguay et al., “CombinedT2-weighted and di<usion-weighted MRI for localization ofprostate cancer,”American Journal of Roentgenology, vol. 189, no.2, pp. 323–328, 2007.

[9] M. P. Lichy, P. Ascho<, C. Plathow et al., “Tumor detectionby di<usion-weightedMRI and ADC-mapping—initial clinicalexperiences in comparison to PET-CT,” Investigative Radiology,vol. 42, no. 9, pp. 605–613, 2007.

[10] S. Gourtsoyianni, N. Papanikolaou, S. Yarmenitis, T. Maris, A.Karantanas, and N. Gourtsoyiannis, “Respiratory gated di<u-sion-weighted imaging of the liver: value of apparent di<u-sion coe_cient measurements in the di<erentiation betweenmost commonly encountered benign and malignant focal liverlesions,” European Radiology, vol. 18, no. 3, pp. 486–492, 2008.

[11] M.-T. Wu, W.-Y. I. Tseng, M.-Y. M. Su et al., “Di<usion tensormagnetic resonance imaging mapping the aber architectureremodeling in human myocardium aber infarction: correlationwith viability and wall motion,” Circulation, vol. 114, no. 10, pp.1036–1045, 2006.

[12] H. P.Adams Jr., G.Del Zoppo,M. J. Alberts et al., “Guidelines forthe early management of adults with ischemic stroke: a guide-line from the American heart association/American strokeassociation stroke council, clinical cardiology council, cardio-vascular radiology and intervention council, and the atheroscle-rotic peripheral vascular disease and quality of care outcomes inresearch interdisciplinary working groups,” Stroke, vol. 38, no. 5,pp. 1655–1711, 2007.

[13] H. C.foeny and F. Keyzer, “Extracranial applications of di<u-sion-weighted magnetic resonance imaging,” European Radiol-ogy, vol. 17, no. 6, pp. 1385–1393, 2007.

[14] A. Bi<ar, O. Dietrich, S. Sourbron, H.-R. Duerr, M. F. Reiser,and A. Baur-Melnyk, “Di<usion and perfusion imaging of bonemarrow,” European Journal of Radiology, vol. 76, no. 3, pp. 323–328, 2010.

[15] M. M. Y. Khoo, P. A. Tyler, A. Saifuddin, and A. R. Padhani,“Di<usion-weighted imaging (DWI) in musculoskeletal MRI: acritical review,” Skeletal Radiology, vol. 40, no. 6, pp. 665–681,2011.

[16] S. Colagrande, S. F. Carbone, L. M. Carusi, M. Cova, andN. Vil-lari, “Magnetic resonance di<usion-weighted imaging: extra-neurological applications,” Radiologia Medica, vol. 111, no. 3, pp.392–419, 2006.

[17] P. F. Beattie, “Di<usion-weighted magnetic resonance imag-ing of the musculoskeletal system: an emerging technologywith potential to impact clinical decision making,” Journal ofOrthopaedic and Sports Physical?erapy, vol. 41, no. 11, pp. 887–895, 2011.

[18] O. Dietrich, A. Bi<ar, M. F. Reiser, and A. Baur-Melnyk, “Di<u-sion-weighted imaging of bone marrow,” Seminars in Muscu-loskeletal Radiology, vol. 13, no. 2, pp. 134–144, 2009.

[19] T. A. Bley, O. Wieben, and M. Uhl, “Di<usion-weighted MRimaging in musculoskeletal radiology: applications in trauma,tumors, and inmammation,”Magnetic Resonance ImagingClinicsof North America, vol. 17, no. 2, pp. 263–275, 2009.

[20] J.-F. Budzik, V. Balbi, S. Verclytte, V. Pansini, V. le fuc, andA. Cotten, “Di<usion tensor imaging in musculoskeletal dis-orders,” Radiographics, vol. 34, no. 3, pp. E56–E72, 2014.

[21] O. Dietrich, A. Bi<ar, A. Baur-Melnyk, and M. F. Reiser, “Tech-nical aspects of MR di<usion imaging of the body,” EuropeanJournal of Radiology, vol. 76, no. 3, pp. 314–322, 2010.

[22] D. Jaramillo, “Whole-body MR imaging, bone di<usion imag-ing: how and why?” Pediatric Radiology, vol. 40, no. 6, pp. 978–984, 2010.

[23] H. K. Kim, D. M. Lindquist, S. D. Serai et al., “Magneticresonance imaging of pediatric muscular disorders. Recentadvances and clinical applications,” Radiologic Clinics of NorthAmerica, vol. 51, no. 4, pp. 721–742, 2013.

[24] A. Baur-Melnyk and H. Boehm, “Osteoporosis,” in ScreeningandPreventiveDiagnosis with Radiological Imaging,M. F. Reiser,G. van Kaick, C. Fink, and S. O. Schoenberg, Eds., pp. 249–260,Springer, Berlin, Germany, 2008.

[25] H. K. Genant and Y. Jiang, “Advanced imaging assessment ofbone quality,” Annals of the New York Academy of Sciences, vol.1068, no. 1, pp. 410–428, 2006.

[26] World Health Organization, “Assessment of fracture risk andits application to screening for postmenopausal osteoporosis:report of a WHO study group,” World Health OrganizationTechnical Report Series vol. 843, World Health Organization,Geneva, Switzerland, 1994.

[27] J. A. Kanis, “Diagnosis of osteoporosis and assessment of frac-ture risk,”?e Lancet, vol. 359, no. 9321, pp. 1929–1936, 2002.

[28] B. Cortet and X. Marchandise, “Bone microarchitecture andmechanical resistance,” Joint Bone Spine, vol. 68, no. 4, pp. 297–305, 2001.

[29] J. F. Gri_th and H. K. Genant, “Bone mass and architecturedetermination: state of the art,” Best Practice and Research:Clinical Endocrinology and Metabolism, vol. 22, no. 5, pp. 737–764, 2008.

[30] P. McDonnell, P. E. McHugh, and D. O’Mahoney, “Vertebralosteoporosis and trabecular bone quality,” Annals of BiomedicalEngineering, vol. 35, no. 2, pp. 170–189, 2007.

[31] J. S. Bauer and T. M. Link, “Advances in osteoporosis imaging,”European Journal of Radiology, vol. 71, no. 3, pp. 440–449, 2009.

[32] NIH Consensum Development Panel on Osteoporosis Preven-tion, “Osteoporosis prevention, diagnosis, and therapy,” ?eJournal of the American Medical Association, vol. 285, no. 6, pp.785–795, 2001.

[33] F. W.Wehrli, J. C. Ford, M. Attie, H. Y. Kressel, and F. S. Kaplan,“Trabecular structure: preliminary application ofMR interfero-metry,” Radiology, vol. 179, no. 3, pp. 615–621, 1991.

[34] F.W.Wehrli, J. A.Hopkins, S. N.Hwang,H.K. Song, P. J. Snyder,and J. G. Haddad, “Cross-sectional study of osteopenia withquantitative MR imaging and bone densitometry,” Radiology,vol. 217, no. 2, pp. 527–538, 2000.

[35] T. M. Link, S. Majumdar, P. Augat et al., “Proximal femur:assessment for osteoporosis with T2∗ decay characteristics atMR imaging,” Radiology, vol. 209, no. 2, pp. 531–536, 1998.

[36] P. K. Saha, B. R. Gomberg, and F. W. Wehrli, “free-dimen-sional digital topological characterization of cancellous bonearchitecture,” International Journal of Imaging Systems andTech-nology, vol. 11, no. 1, pp. 81–90, 2000.

[37] R. Krug, S. Banerjee, E. T. Han, D. C. Newitt, T. M. Link, andS. Majumdar, “Feasibility of in vivo structural analysis of high-resolution magnetic resonance images of the proximal femur,”Osteoporosis International, vol. 16, no. 11, pp. 1307–1314, 2005.

[38] S. Majumdar, D.fomasson, A. Shimakawa, and H. K. Genant,“Quantitation of the susceptibility di<erence between trabec-ular bone and bone marrow: experimental studies,” MagneticResonance in Medicine, vol. 22, no. 1, pp. 111–127, 1991.

BioMed Research International 9

[39] S. N. Hwang and F. W. Wehrli, “>e calculation of the suscep-tibility-induced magnetic Feld from 3D NMR images withapplications to trabecular bone,” Journal of Magnetic Resonance,Series B, vol. 109, no. 2, pp. 126–145, 1995.

[40] S. de Santis, M. Rebuzzi, G. Di Pietro, F. Fasano, B. Maraviglia,and S. Capuani, “In vitro and in vivo MR evaluation of internalgradient to assess trabecular bone density,” Physics in Medicineand Biology, vol. 55, no. 19, pp. 5767–5785, 2010.

[41] F. W. Wehrli, “Magnetic resonance of calciFed tissues,” Journalof Magnetic Resonance, vol. 229, pp. 35–48, 2013.

[42] T. H. Berquist, MRI of the Musculoskeletal System, LippincottWilliams &Wilkins, 5th edition, 2006.

[43] S. Capuani, “Water di\usion in cancellous bone,” Microporousand Mesoporous Materials, vol. 178, pp. 34–38, 2013.

[44] D. Schellinger, C. S. Lin, J. Lim, H. G. Hatipoglu, J. C. Pezzullo,and A. J. Singer, “Bone marrow fat and bone mineral density onproton MR spectroscopy and dual-energy X-ray absorptiom-etry: their ratio as a new indicator of bone weakening,” <eAmerican Journal of Roentgenology, vol. 183, no. 6, pp. 1761–1765,2004.

[45] D. K.W. Yeung, J. F. Gricth, G. E. Antonio, F. K. H. Lee, J. Woo,and P. C. Leung, “Osteoporosis is associated with increasedmarrow fat content and decreased marrow fat unsaturation: aproton MR spectroscopy study,” Journal of Magnetic ResonanceImaging, vol. 22, no. 2, pp. 279–285, 2005.

[46] J. F.Gricth,D.K.W.Yeung,G. E.Antonio et al., “Vertebral bonemineral density, marrow perfusion, and fat content in healthymen and men with osteoporosis: dynamic contrast-enhancedMR imaging and MR spectroscopy,” Radiology, vol. 236, no. 3,pp. 945–951, 2005.

[47] M. C. Walsh, G. R. Hunter, and M. B. Livingstone, “Sarcopeniain premenopausal and postmenopausal women with osteo-penia, osteoporosis and normal bone mineral density,” Osteo-porosis International, vol. 17, no. 1, pp. 61–67, 2006.

[48] U. Tarantino, J. Baldi, M. Celi et al., “Osteoporosis and sarcope-nia: the connections,”Aging Clinical and Experimental Research,vol. 25, no. 1, pp. S93–S95, 2013.

[49] G. P. Liney, C. P. Bernard, D. J. Manton, L. W. Turnbull, and C.M. Langton, “Age, gender, and skeletal variation in bone mar-row composition: a preliminary study at 3.0 Tesla,” Journal ofMagnetic Resonance Imaging, vol. 26, no. 3, pp. 787–793, 2007.

[50] F. H. Netter, R. T. Woodburne, E. S. Crelin, and F. Kaplan,Mus-culoskeletal System, Part 1: Anatomy, Physiology, and MetabolicDisorders, vol. 8 of Netter Collection of Medical Illustrations,Ciba Pharmaceutical, 1987.

[51] J. Ren, I. Dimitrov, A.D. Sherry, andC. R.Malloy, “Compositionof adipose tissue and marrow fat in humans by 1H NMR at 7Tesla,” Journal of Lipid Research, vol. 49, no. 9, pp. 2055–2062,2008.

[52] S. A. Athavale, S. D. Joshi, and S. S. Joshi, “Internal architectureof calcaneus: correlations with mechanics and pathoanatomy ofcalcaneal fractures,” Surgical and Radiologic Anatomy, vol. 32,no. 2, pp. 115–122, 2010.

[53] S. Capuani, E. Piccirilli, G. di Pietro, M. Celi, and U. Tarantino,“Microstructural di\erences between osteoporotic and osteo-arthritic femoral cancellous bone: an in vitro magnetic reso-nance micro-imaging investigation,” Aging Clinical and Experi-mental Research, vol. 25, no. 1, pp. S51–S54, 2013.

[54] C. J. Rosen and M. L. Bouxsein, “Mechanisms of disease: isosteoporosis the obesity of bone?” Nature Clinical PracticeRheumatology, vol. 2, no. 1, pp. 35–43, 2006.

[55] M. Rebuzzi, V. Vinicola, F. Taggi, U. Sabatini, F. W. Wehrli,and S. Capuani, “Potential diagnostic role of the MRI-derivedinternal magnetic Feld gradient in calcaneus cancellous bonefor evaluating postmenopausal osteoporosis at 3T,” Bone, vol.57, no. 1, pp. 155–163, 2013.

[56] W. S. Price, “Pulsed-Feld gradient nuclear magnetic resonanceas a tool for studying translational di\usion. Part 1. Basictheory,” Concepts in Magnetic Resonance, vol. 9, no. 5, pp. 299–335, 1997.

[57] E. O. Stejskal and J. E. Tanner, “Spin di\usion measurements:spin echoes in the presence of a time-dependent Feld gradient,”<e Journal of Chemical Physics, vol. 42, no. 1, pp. 288–292, 1965.

[58] U. Sinha, S. Sinha, J. A. Hodgson, and R. V. Edgerton, “Humansoleus muscle architecture at di\erent ankle joint angles frommagnetic resonance di\usion tensor imaging,” Journal ofApplied Physiology, vol. 110, no. 3, pp. 807–819, 2011.

[59] J. F. Budzik, V. Le >uc, X. Demondion, M. Morel, D. Chechin,and A. Cotten, “In vivoMR tractography of thighmuscles usingdi\usion imaging: initial results,” European Radiology, vol. 17,no. 12, pp. 3079–3085, 2007.

[60] J. E. Tanner, “Use of stimulated echo in NMR-di\usion studies,”<e Journal of Chemical Physics, vol. 52, no. 5, pp. 2523–2526,1970.

[61] D. K. Jones and P. J. Basser, “‘Squashing peanuts and smashingpumpkins’: how noise distorts di\usion-weighted MR data,”Magnetic Resonance in Medicine, vol. 52, no. 5, pp. 979–993,2004.

[62] M. E. Bastim, P. A. Armitage, and I. Marshall, “A theoreticalSNR: non monotonic behaviors of tensor contrasts,” MagneticResonance Imaging, vol. 16, pp. 773–785, 1998.

[63] B. A. Landman, J. A. D. Farrell, H. Huang, J. L. Prince, andS. Mori, “Di\usion tensor imaging at low SNR: nonmonotonicbehaviors of tensor contrasts,”Magnetic Resonance Imaging, vol.26, no. 6, pp. 790–800, 2008.

[64] G. Manenti, S. Capuani, E. Fanucci et al., “Di\usion tensorimaging and magnetic resonance spectroscopy assessment ofcancellous bone quality in femoral neck of healthy, osteopenicand osteoporotic subjects at 3T: preliminary experience,” Bone,vol. 55, no. 1, pp. 7–15, 2013.

[65] B. M. Damon, Z. Ding, A. W. Anderson, A. S. Freyer, and J. C.Gore, “Validation of di\usion tensor MRI-based muscle Fbertracking,”Magnetic Resonance in Medicine, vol. 48, no. 1, pp. 97–104, 2002.

[66] U. Sinha and L. Yao, “In vivo di\usion tensor imaging of humancalf muscle,” Journal of Magnetic Resonance Imaging, vol. 15, no.1, pp. 87–95, 2002.

[67] A. vanDoorn, P.H.M. Bovendeerd, K.Nicolay,M.R.Drost, andJ. D. Janssen, “Determination of muscle Fbre orientation usingdi\usioni-weightedMRI,” European Journal of Morphology, vol.34, no. 1, pp. 5–10, 1996.

[68] C. Rossi, A. Boss, G. Steidle et al., “Water di\usion anisotropyin white and gray matter of the human spinal cord,” Journal ofMagnetic Resonance Imaging, vol. 27, no. 3, pp. 476–482, 2008.

[69] F. Schick, “Bone marrow NMR in vivo,” Progress in NuclearMagnetic Resonance Spectroscopy, vol. 29, no. 3-4, pp. 169–227,1996.

[70] Y.Ueda, T.Miyati,N.Ohno et al., “Apparent di\usion coeccientand fractional anisotropy in the vertebral bonemarrow,” Journalof Magnetic Resonance Imaging, vol. 31, no. 3, pp. 632–635, 2010.

[71] F. W. Wehrli, L. Hilaire, M. Fernandez-seara et al., “Quantita-tive magnetic resonance imaging in the calcaneus and femur

10 BioMed Research International

of women with varying degrees of osteopenia and vertebraldeformity status,” Journal of Bone and Mineral Research, vol. 17,no. 12, pp. 2265–2273, 2002.

[72] J. F. GriIth,D.K.W.Yeung,G. E.Antonio et al., “Vertebralmar-row fat content and diSusion and perfusion indexes in womenwith varying bone density: MR evaluation,” Radiology, vol. 241,no. 3, pp. 831–838, 2006.

[73] D. K. W. Yeung, S. Y. S. Wong, J. F. GriIth, and E. M.C. Lau, “Bone marrow diSusion in osteoporosis: evaluationwith quantitative MR diSusion imaging,” Journal of MagneticResonance Imaging, vol. 19, no. 2, pp. 222–228, 2004.

[74] S. Capuani, C. Rossi, M. Alesiani, and B. Maraviglia, “DiSusiontensor imaging to study anisotropy in a particular porous sys-tem: the trabecular bone network,” Solid State Nuclear MagneticResonance, vol. 28, no. 2–4, pp. 266–272, 2005.

[75] G. Di Pietro, M. Palombo, and S. Capuani, “Internal magnetic_eld gradients in heterogeneous porous systems: comparisonbetween Spin-Echo and DiSusion Decay Internal Field (DDIF)method,” Applied Magnetic Resonance, vol. 45, no. 8, pp. 771–784, 2014.

[76] J. Mitchell, T. C. Chandrasekera, M. L. Johns, L. F. Gladden,and E. J. Fordham, “Nuclear magnetic resonance relaxation anddiSusion in the presence of internal gradients: the eSect of mag-netic _eld strength,” Physical Review E: Statistical, Nonlinear,and So@ Matter Physics, vol. 81, no. 2, Article ID 026101, 2010.

[77] E. E. Sigmund, H. Cho, and Y.-Q. Song, “High-resolution MRIof internal _eld diSusion-weighting in trabecular bone,” NMRin Biomedicine, vol. 22, no. 4, pp. 436–448, 2009.

[78] J. Zhong, R. P. Kennan, and J. C. Gore, “ESects of susceptibilityvariations on NMR measurements of diSusion,” Journal ofMagnetic Resonance, vol. 95, no. 2, pp. 267–280, 1991.

[79] C. Rossi, A. Boss, P.Martirosian et al., “Inguence of steady back-ground gradients on the accuracy of molecular diSusion aniso-tropy measurements,” Magnetic Resonance Imaging, vol. 26, no.9, pp. 1250–1258, 2008.