peripheral nerve perfusion by dynamic.3 - clinical...

Peripheral Nerve Perfusion by Dynamic Contrast-EnhancedMagnetic Resonance Imaging

Demonstration of Feasibility

Philipp Baumer, MD, MSc,* Maximilian Reimann,* Clemens Decker, PhD,!Alexander Radbruch, MD, JD,* Martin Bendszus, MD,* Sabine Heiland, PhD,! and Mirko Pham, MD*

Purpose: The aim of this study was to establish dynamic contrast-enhancedperfusion in peripheral nerves for determination of blood-nerve permeability(Ktrans) and nerve blood volume (NBV) in peripheral neuropathies as com-pared with healthy controls.Methods: The study was approved by the institutional ethics committee, andwritten informed consent was obtained from all participants. Forty-threecontrols (24 women, 19 men; age, 48.7 T 17.5 years) and 59 patients withperipheral neuropathy (28 women, 31 men; age, 52.7 T 12.4 years) were ex-amined by a standard protocol including a T1-weighted dynamic contrast-enhanced sequence (time of repetition/time of echo, 4.91/1.64; 10 slices;resolution 0.8 ! 0.6 ! 3.0 mm3). Time - signal intensity analysis wasperformed by normalizing to preYbolus arrival and calculating the meancontrast uptake (MCU) for each patient. Further analyses were performed bycustomized software to calculate Ktrans and NBV. Statistical analysis included2-sided Student’s t tests of controls versus patients, receiver operating char-acteristic analysis, and subgroup analysis of patients according to etiologies ofneuropathy.Results: TimeYsignal intensity analysis showed significantly increased contrastuptake in patients as compared with controls (MCU, 1.29 T 0.15 vs 1.18 T 0.08;P G 0.001). This was caused mainly by an increase in Ktrans (0.046 T 0.025 vs0.026 T 0.016 minj1; P G 0.001) and less by an increase in NBV (3.9 T 2.6 vs3.0 T 1.9 mL/100 mL; P = 0.12). This trend was true for all etiologies exceptentrapment neuropathies. Excluding these, receiver operating characteristicanalysis found an area under the curve of 0.78 (95% confidence interval,0.69Y0.89) for MCU and 0.77 (95% confidence interval, 0.65Y0.90) for Ktrans

to discriminate neuropathy from control.Conclusions: Dynamic contrast-enhanced perfusion is a feasible technique toassess Ktrans and NBV in peripheral nerves and may be used in future in-vestigations on peripheral neuropathies.

Key Words: peripheral nerves, MR Neurography, MRI, perfusion, permeability

(Invest Radiol 2014;49: 518Y523)

The peripheral nervous system (PNS) is increasingly investigatedby means of magnetic resonance neurography (MRN).1Y4 Al-

though dynamic contrast-enhanced (DCE) perfusion by magneticresonance imaging (MRI) has long been established as a standard

functional imaging technique for the central nervous system(CNS),5,6 it has, to date, not been implemented in MRN scans. Dy-namic contrast-enhanced MRI can be used for estimation of perfu-sion parameters such as blood volume or permeability, providingpotentially valuable information that is not as readily obtainableby other methods.7,8 In peripheral nerves, the blood-nerve barrier(BNB) forms a tight and highly regulated interface in the humanbody similar to the blood-brain interface.9 Pathogenic factors in pe-ripheral neuropathies are nerve edema, demyelination, and inflam-mation, all of which might be associated with perturbations in theBNB.10 Information on nerve permeability is therefore potentially ofhigh relevance for diagnostic MRN examinations in neuropathiesof inflammatory, ischemic, or other etiology. Moreover, the patho-genesis of various prevalent and poorly understood metabolic or in-flammatory polyneuropathies might be further addressed by thistechnique.

The technical challenge of DCE MRI perfusion in peripheralnerves is their small caliber, ranging from less than 10 mm2 for upperarm nerves such as the radial nerve to approximately 50 mm2 for thesciatic nerve.11 This might have hampered previous attempts in theapplication and optimization of DCE MRI for MRN. However, cur-rent structural MRN sequences are able to resolve peripheral nervesto the level of individual fascicles so that structural resolution of anerve trunk should, in principle, be achievable for DCE sequences.

In this study, we implemented a peripheral nerve perfusionprotocol based on DCE and quantitative evaluation methods andaimed to assess these techniques for the characterization of healthynerve perfusion by MRI and its potential to detect altered perfusion inneuropathy.

PATIENTS AND METHODS

Clinical and Demographic Patient DataThis study was approved by the institutional ethics board

(S398-2012), and written informed consent was obtained from allpatients. Patients undergoing a clinical MRI study of the extremitiesfor any indication (eg, musculoskeletal or MRN) and receiving con-trast agent were asked to participate in the study. Overall, 43 subjectswho underwent examination of an extremity for indications otherthan clinically apparent neuropathy served as healthy controls (24women, 19 men; age, 48.7 T 17.5 years). A total of 59 subjects wereincluded as patients with neuropathy (28 women, 31 men; age, 52.7 T12.4 years). Eight subjects underwent the same protocol but were notincluded in the study because MRN imaging did not show any ab-normalities and therefore appeared incongruent with clinical infor-mation. Diagnosis in patients was based on clinical symptoms and/orelectrophysiological evidence of motor or sensory impairment in thedistribution of nerves examined by MRN.

Patients were further classified according to the etiology of thedisease as entrapment neuropathy, inflammatory neuropathy, trau-matic nerve injury, hereditary polyneuropathy, or, if clinical, elec-trophysiological, and imaging examinations had been inconclusive

ORIGINAL ARTICLE

518 www.investigativeradiology.com Investigative Radiology & Volume 49, Number 8, August 2014

Received for publication November 26, 2013; and accepted for publication, afterrevision January 22, 2014.

From the *Department of Neuroradiology and †Section of Experimental Radiology,Department of Neuroradiology, Heidelberg University Hospital, Heidelberg,Germany.

Conflicts of interest and sources of funding: This study was supported by aPostdoctoral-Fellowship granted to P.B. from the Medical Faculty of the Uni-versity of Heidelberg. M.P. is supported by the EFSD/JDRF/Novo NordiskEuropean Programme in Type 1 Diabetes Research.

Reprints: Philipp Baumer, MD, MSc, Department of Neuroradiology, HeidelbergUniversity Hospital, Im Neuenheimer Feld 400, 69120 Heidelberg, Germany.E-mail: [email protected].

Copyright * 2014 by Lippincott Williams & WilkinsISSN: 0020-9996/14/4908Y0518

Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

at the time of examination, as disseminated polyneuropathy of yetundetermined cause. Entrapment neuropathy was diagnosed if a pa-tient presented with symptoms in the distribution of 1 nerve com-patible with carpal tunnel syndrome or ulnar neuropathy at the elbowand electrophysiological findings supported this diagnosis. Patientswere classified as inflammatory neuropathy if a polyneuropathy wassupported by clinical and electrophysiological examinations or if amononeuropathy at a localization atypical for compression neurop-athies was detected and the underlying disease was known, such asmultifocal motor neuropathy or chronic inflammatory demyelinatingpolyneuropathy. If the underlying etiology was not known at the timeof examination, patients were classified as disseminated neuropathyof yet undetermined cause. If a hereditary cause for the neuropathywas known, such as hereditary sensory and motor neuropathy, etiol-ogy was classified as hereditary polyneuropathy, and finally, in casesof known blunt or sharp trauma to a nerve, etiology was classified astraumatic nerve injury.

MRN ImagingExaminations were conducted at 3 T magnetic field strength

(Magnetom VERIO; Siemens AG, Erlangen, Germany) betweenJanuary and July 2013. First, a T2-weighted turbo-spin-echo se-quence with spectral fat saturation and at high spatial resolution forreliable recognition and segmentation of peripheral nerves was ac-quired. The slab position of this first sequencewas chosen based on thepatient’s known or most likely localization of maximum nerve lesion assuspected by clinical and electrophysiological findings. At this posi-tion, a T1-weighted DCE volume interpolated breathhold examina-tion sequence was acquired for the detection of contrast uptake andfurther quantitative analysis. A contrast agent (DOTAREM; Guerbet,Villepinte, France) was administered at the beginning of the third frameof the sequence, that is, after 12 to 18 seconds, at a standard concen-tration of 0.1 mmol/kg and with a flow rate of 3.5 mL/s. A total of35 frames were recorded at a rate of 6.11 seconds per frame. A T1-weighted sequence with fat saturation after administration of contrastagent was then acquired at the same position in 79 of the 102 subjects.Sequence parameters are given in Table 1.

A knee 15-channel transmit/receive phased array radiofrequencycoil was used in 88 subjects and awrist 8-channel receive radiofrequencycoil was used in 14 subjects.

Quantitative Image AnalysisTimeYSignal Intensity Curve Analysis

Using the software mean curve on a Syngo-workstation(Syngo VE 32 B; Siemens), peripheral nerve signal was analyzedfor all time points, and timeYsignal intensity curves were plotted.Since no automated method is established for the registration of pe-ripheral nerves, masks for peripheral nerves were obtained by precisemanual segmentation around the epineurial contour (Fig. 1) of allperipheral nerves at all slice positions (median, ulnar, and radialnerves for upper arm examinations; median and ulnar nerves forelbow and wrist; sciatic nerve for thigh; and peroneal and tibial nervefor knee examinations). The most proximal and distal positions

(1 and 10) were discarded because of potential 3-dimensionalaliasing artifacts at the extremes of the imaging slab. An additionalmask was placed in the largest artery contained in the imaging slab toderive the arterial input function (AIF) and to determine the exact timepoint of bolus arrival in the examined body region. Absolute signalintensity values for each mask were read out and then normalized tothe arithmetic mean of the preYbolus arrival frames (6Y8 frames) bydivision. These values then allowed plotting of an individual timeYsignal intensity curve for each nerve. Furthermore, all signal intensityvalues recorded later than 10 frames (61 seconds) after bolus arrivalwere averaged for each nerve to calculate 1 single value, denoted asmean contrast uptake (MCU).

Quantitative Analysis Using the Patlak ModelWe used the Patlak model to estimate the transfer constant

(Ktrans) between intravascular and extravascular space and nerve bloodvolume (NBV). Although the Patlak model was developed for irre-versible tracer uptake,12 it can also be used in first-order kinetics if theinfluence of the diffusion of contrast agent from the extravascular backto the intravascular space is negligible, which is a valid assumptionwhen Ktrans is relatively low and the observation time is short. In thiscase, Ktrans and NBV can be determined by linear regression of

x "Xt

0AIF#T$ dT

AIF#t$

and

y " c#t$AIF#t$ ;

considering the equation

y#t$ " NBV % k transqx#t$;

where AIF(t) is the concentration of contrast agent in the supplyingartery at a given time point t and c(t) is the contrast agent concentra-tion measured in tissue.

We used a customized software package developed in ourdepartment (O2dicom) to determine Ktrans (in minj1) and NBV(in mL/100 mL) in the peripheral nerve masks. O2dicom is written inJAVA and is based on the Patlak model12 to calculate Ktrans and NBVas described above semiautomatically either on a pixel-by-pixel basisor in regions of interest. A linear relationship between measuredsignal intensity and contrast concentration was assumed. All signalvalues up to the time point of bolus arrival were used for normali-zation. Computed analysis in several cases failed to calculate pa-rameters because the subjects’ AIF was insufficient for analysis bythe Patlak model. Furthermore, pulsation artifacts by adjacent vesselsoccasionally impaired quantitative analysis. We therefore set respec-tive upper and lower limits for Ktrans as 0.1 and 0.0 minj1 and NBVas 10 and 0.0 mL/100 mL and excluded all subjects in whom theselimits were passed at any slice position.

TABLE 1. Sequence Parameters

Sequence/Parameters TR, ms TE, ms No. Slices ST, mm Gap, mm FoV, mm Resolution, mm2 n Time of Acquisition

T1 DCE VIBE 4.91 1.64 10 3.0 0.6 160 0.8 ! 0.6 1 3:34 minT2 fs 7020 52 45 3.0 0.3 130 0.3 ! 0.3 3 7:17 minT1 fs CE 850 16 45 3.0 0.3 170 0.4 ! 0.3 2 4:41 min

CE indicates contrast enhanced; DCE, dynamic contrast enhanced; FoV, field of view; fs, fat-saturated; n, number of averages; ST, slice thickness; TE, time ofecho; TR, time of repetition; VIBE, volume interpolated breathhold examination.

Investigative Radiology & Volume 49, Number 8, August 2014 Peripheral Nerve Perfusion by DCE MRI

* 2014 Lippincott Williams & Wilkins www.investigativeradiology.com 519


Overall, by timeYsignal intensity curve analysis and Patlakquantitative analysis, 3 quantitative read-out parameters were calcu-lated for each nerve (Fig. 1) and used for further statistical analysis.

Statistical AnalysisData visualization and statistical analyses were performed using

Origin Pro 9.0 (Northampton, MA). Mean values were calculated forMCU, Ktrans, and NBV in each subject for each nerve. In controls, ifmore than 1 nerve was present in the imaging sections, values wereaveraged. In patients, only those nerves affected by neuropathy, that is,those with objectifiable symptoms and/or electrophysiological evi-dence of dysfunction, were averaged and used for statistical analysis ofpatients. Graphs mapping the timeYsignal intensity curves for groupmean values and box plots for Ktrans and NBV were charted in OriginPro 9.0. Mean values were tested against each other for statistical sig-nificance using a 2-tailed Student’s t test, with a P value of G0.05considered significant. The Bonferroni-Holm correction was used toadjust for the family-wise error rate in multiple comparisons. Pearsoncorrelation analysis was performed for age versus MCU, Ktrans, andNBV. Receiver operating characteristic (ROC) analysis was performedin Origin Pro 9.0.

RESULTSA total of 102 participants participated in the study: 43 were

included as control subjects and 59 subjects were included as patientsbased on clinical and electrophysiological results. Patients were furtherclassified by etiology into 12 patients with an entrapment neuropathy,22 patients with an inflammatory neuropathy, 4 patients with traumaticnerve injury, 2 patients with a known hereditary polyneuropathy, and 19patients with a polyneuropathy of yet undetermined cause. Neuropa-thies caused by tumors of the PNS were excluded from the study. Bybody region, examinations covered the upper arm in 12 controls and 21patients, respectively, the elbow in 10 and 14, the wrist in 5 and 9, thethigh in 11 and 8, and the knee in 5 and 7.

TimeYsignal intensity curve analysis showed that contrastagent uptake in the nerves of patients with neuropathy was signifi-cantly higher compared with normal nerve tissue (Fig. 2).

Patlak analysis could be applied to 27 controls and 42 patients(Fig. 3). Quantitative read-out parameters were compared betweenpatients and controls (Fig. 4). Mean contrast uptake was significantlyincreased in patients (1.29 T 0.15 in patients vs 1.18 T 0.08 in con-trols; P G 0.001). Likewise, Ktrans was significantly higher in patients(0.046 T 0.025 vs 0.026 T 0.016 minj1; P G 0.001). The increase inNBV in patients did not reach statistical significance (3.9 T 2.6 vs3.0 T 1.9 mL/100 mL; P = 0.12).

Subgroup analyses for neuropathies with different etiologieswere performed. All 3 read-out parameters were significantly higherin traumatic nerve injury than in any other group (Table 2). Averageperfusion parameter values in entrapment neuropathies were found tofall in the range of control subjects and not in the range of neurop-athies of other etiologies.

The ROC analysis was performed to assess the potential ofperfusion read-out parameters as diagnostic markers for the presenceof peripheral neuropathy. Whereas NBV was found to be a weakmarker, MCU and Ktrans provided considerable diagnostic accuracyfor the detection of peripheral neuropathy (ROC values are givenin Fig. 5). Since values in entrapment neuropathies fell in the rangeof controls and not in that of neuropathies of other etiologies, theirexclusion yielded improved diagnostic accuracy (Table 3).

Perfusion parameter values were independent of the body regionin which they were assessed in healthy controls; for example, no sta-tistically significant differences were found between proximal versusdistal locations or between upper and lower extremities. Because

FIGURE 1. Quantitative analysis of peripheral nerve perfusion.Precise regions of interest were drawn around the epineurialcontour of peripheral nerves in axial T1-vibe source images asillustrated in the upper left corner. T2- and T1-weightedimages were used as reference if exact delineation of nerveswas unclear in T1-vibe. Quantitative analysis was performed bytimeYsignal intensity course analysis, which yielded a value forMCU, normalized to baseline before bolus arrival. Quantitativeparameter maps were calculated using the Patlak model andyielded values for Ktrans and NBV.

FIGURE 2. TimeYsignal intensity curve analysis of healthycontrols and patients with neuropathy. Values are normalizedto baseline before contrast (time points =j7 toj1). Contrastarrival in the arterial system is at time point = 0. Patients withneuropathy show distinctly increased uptake already withinthe first frames, which persists for more than 2 minutes.

Baumer et al Investigative Radiology & Volume 49, Number 8, August 2014

520 www.investigativeradiology.com * 2014 Lippincott Williams & Wilkins


peripheral nerves are known to show age-related degenerative changesand blood volume is known to decrease with age in the CNS, correla-tion analyses between perfusion parameters and agewere performed. Inhealthy controls, this revealed a negative linear correlation betweenNBV and age (P = 0.047) (Fig. 6). No significant correlations of agewith Ktrans or MCU were observed.

DISCUSSIONWe here report a DCE MRI technique to characterize peripheral

nerve perfusion by parameters of contrast uptake, blood-nerve perme-ability (Ktrans), and NBV. These measures represent hitherto unusedmetabolic markers for human neuropathies and seem particularlypromising to diagnose and understand nerve diseases for which nerveconduction studies assessing electrophysiological function have previ-ously been the only investigative method. In addition to being the firstinvestigation to test the feasibility of this method, our study also dem-onstrates that in a total of 59 patients and 43 controls, perfusion pa-rameters are significantly altered in the presence of neuropathy.

We found that symptomatic nerves demonstrate significantlyincreased contrast enhancement and that this is caused by an increase inKtrans and not in NBV. This corresponds to a disruption of the BNB,which, under normal circumstances, forms a tight and highly regulatedinterface in the human body similar to the blood-brain barrier.9 Eval-uation of time-dependent signal increase after contrast administrationshowed a relatively slow and continuous influx of contrast agent toperipheral nervous tissue during the sequence acquisition. This obser-vation allowed use of the Patlak model for further analysis as a well-known model for quantifying the unidirectional influx constant forlow-permeating substances.12 The values we report here for Ktrans andNBVin control nerves are similar tovalues in normal CNSwhite matterreported in the literature. For example, Leenders et al,13 in a study usingpositron emission tomography, found a cerebral blood volume (CBV)of 2.7 mL/100 mL, which is close to the 3.0 mL/100 mL we report forhealthy peripheral nerves. Other studies using DCE T1 MRI arrive atsimilar values for CBV.14Y16 The reported increase in permeability in

patients with neuropathy in this study is plausible given the patho-physiological processes in neuropathy with edema and permeabilityincrease of the BNB, demyelination, and axon loss.10 The increase inNBVand permeability in neuropathies is analogous to increases in in-flammatory brain lesions detected by DCE MRI16,17 and suggest thatthe technique as used here yields accurate estimates for perfusionparameters.

In addition to these findings, a trend of decreasing NBV withincreasing age was observed in our control group. Little experimental

FIGURE 3. Parameter maps for Ktrans and NBV in the wrist and thigh. Upper row shows wrist images with median nerve, lower rowshows thigh images with sciatic nerve already divided into tibial (t) and peroneal (p) fascicles. Axial T2-weighted images with fatsaturation on the left show nerves at high spatial resolution for visual recognition and anatomical orientation. Ktrans and NBV werecalculated using the Patlak model and are visualized in quantitative parameter maps. These parameter maps also allow relativelygood differentiation between nerve and surrounding epineurial and fatty tissue, as well as differentiation between tibial andperoneal branches of the sciatic nerve as displayed in the lower row.

FIGURE 4. Statistical analysis of perfusion parameters incontrols versus patients. A, Mean-contrast uptake issignificantly increased in patients with peripheral neuropathy.B, Blood-nerve permeability as assessed by Ktrans is likewisesignificantly increased in patients. C, A trend for increasedNBV is observed in patients but does not reach statisticalsignificance.




and clinical evidence exists about peripheral nerve perfusion changesassociated with aging. Thickening of basal lamina ensheathment duringaging has been reported18 as well as a decreasing response to vasodi-lators,19 both of which could factor into decreasing intraneural bloodvolume during aging. For the CNS, correlations between perfusionparameters and age have been established using positron emission to-mography.13,20 From these, CBV is known to decrease with increasingage in white matter as well as the entire brain. The correlation in ourgroup of healthy controls is analogous and further supports our model.

To our knowledge, this is the first report on investigating hu-man peripheral nerve perfusion in vivo by MRI using quantitativeparameters. One previous study investigated carpal tunnel syndromein 9 patients but remained qualitative in the analysis.21 Severalgroups have investigated the potential of sonography to assess nerveperfusion.22Y25 Although technical advances and postprocessing willenhance the application of sonography for measuring perfusion pa-rameters in nerves,26 the use of sonography has, until now, beenlimited to measuring intraneural flow, whereas MRI data allow cal-culation of both permeability and blood volume. In addition, so-nography was applied in these studies only in superficial entrapmentneuropathies, while nerves frequently affected by polyneuropathies

and located deep within tissue, such as the sciatic nerve, are difficultto examine. Moreover, sonography is highly operator dependent. Incontrast, MRI allows an observer-independent and fully quantitativeanalysis.

The diagnostic accuracy calculated by ROC analysis showedperfusion measures to be diagnostic signs of moderate to good accuracyin detecting peripheral neuropathy. In this range, they are comparablewith the diagnostic accuracy of diffusion tensor imaging as the hithertoonly available functional imaging method for the PNS.27 Refinement ofthe used sequence and postprocessing may yield even better diagnosticquality.We anticipate that perfusionMRI for peripheral nerveswill findapplication in the investigation of the time course and localization ofthe highly prevalent and poorly understood diseases such as inflam-matory and ischemic polyneuropathies. Furthermore, the techniquemay also prove useful in monitoring of therapeutic drug effects in thepolyneuropathies.

Our study has several limitations. First, the general assumptionsand limitations of the Patlak model apply to the technique, includingunidirectionality of permeability at least in the first 2 minutes. Second, anumber of patients were excluded from evaluation of Ktrans and NBV,either because AIF did not allow calculation or because of pulsationartifacts. Further refinement of sequence parameters to improve AIFmeasurement and of software parameters may solve this in the future.Third, absolute values as reported here are always dependent on themodel chosen; thus, investigators should verify a normal range ofvalues at their own centers.28,29 Finally, our study consisted of a large

TABLE 2. Perfusion Parameter Values According to Etiology of Neuropathy

CompressiveNeuropathy(n = 12/3)

TraumaticNeuropathy(n = 4/1)

InflammatoryNeuropathy(n = 21/9)

DisseminatedNeuropathy(n = 21/9)

HereditaryNeuropathy(n = 2/0)

Controls(n = 43/16)

MCU 1.19 T 0.12 1.47 T 0.12 1.30 T 0.18 1.28 T 0.14 1.34 T 0.11 1.18 T 0.08vs controls P = 0.81 P G 0.001 P G 0.001 P = 0.003 P = 0.046vs other etiologies P = 0.049 P = 0.043 P = 0.51 P = 0.97 P = 0.59 P G 0.001

Ktrans, 1/min 0.035 T 0.023 0.082 T 0.036 0.045 T 0.017 0.041 T 0.020 0.073 T 0.005 0.026 T 0.016vs controls P = 0.19 P G 0.001 P = 0.003 P = 0.043 P = 0.003vs other etiologies P = 0.16 P = 0.005 P = 0.87 P = 0.29 P = 0.12 P G 0.001

NBV, mL/100 mL 3.3 T 2.2 8.5 T 4.0 3.7 T 2.2 3.6 T 2.1 3.2 T 0.8 3.0 T 1.9vs controls P = 0.66 P = 0.002 P = 0.33 P = 0.37 P = 0.86vs other etiologies P = 0.46 P = 0.006 P = 0.79 P = 0.51 P = 0.71 P = 0.12

Perfusion parameters are given for each subgroup of neuropathy patients according to etiology. Subgroup values were tested against controls and against themean of all other neuropathy patients.

P values are adjusted for multiple comparisons by the Bonferroni-Holm correction. Numbers in parentheses indicate total number of subjects and number ofsubjects excluded from the analysis because of insufficient AIF or pulsation artifacts for accurate calculation of Ktrans and NBV.

MCU indicates mean contrast uptake; Ktrans, blood-nerve permeability; NBV, nerve blood volume; AIF, arterial input function.

FIGURE 5. ROC analysis. Parameters of MCU and Ktrans havesignificant diagnostic accuracy in discriminating neuropathiesfrom healthy controls with AUC of 0.78 and 0.77, respectively.

TABLE 3. ROC Analysis of Diagnostic Performance for PerfusionRead-Out Parameters

All EtiologiesExcluding Entrapment

Neuropathies

AUC 95% CI AUC 95% CI

MCU 0.72 0.46Y0.74 0.78 0.69Y0.89Ktrans 0.74 0.60Y0.87 0.77 0.65Y0.90NBV 0.60 0.41Y0.83 0.62 0.47Y0.76

ROC indicates receiver operating characteristic; AUC, area under thecurve; CI, confidence interval; MCU, mean contrast uptake; Ktrans, blood-nerve permeability; NBV, nerve blood volume.

Baumer et al Investigative Radiology & Volume 49, Number 8, August 2014

522 www.investigativeradiology.com * 2014 Lippincott Williams & Wilkins


but heterogeneous patient group. Although this was effective to test thetechnique at different anatomical positions and in different forms ofneuropathy, future investigations may focus on 1 restricted diagnosis oretiology.

In conclusion, we have developed a method for quantifica-tion of peripheral nerve perfusion parameters. The plausibility of thetechnique is supported by findings in healthy controls and in a largenumber of patients with neuropathy. Dynamic contrast-enhanced T1MRI may be used in future investigations on peripheral neuropathyand in clinical MRN.

REFERENCES1. Bendszus M, Stoll G. Technology insight: visualizing peripheral nerve injury

using MRI. Nat Clin Pract Neurol. 2006;1:45Y53.2. Gasparotti R, Lodoli G, Meoded A, et al. Feasibility of diffusion tensor

tractography of brachial plexus injuries at 1.5 T. Invest Radiol. 2013;48:104Y112.

3. Baumer P, Weiler M, Ruetters M, et al. MR neurography in ulnar neuropathy assurrogate parameter for the presence of disseminated neuropathy. PLoS One.2012;7:e49742.

4. Pham M, Baumer P, Meinck HM, et al. Anterior interosseous nerve syndrome:fascicular motor lesions of median nerve trunk. Neurology. 2014;82:598Y606.

5. Runge VM, Schoerner W, Niendorf HP, et al. Initial clinical evaluation ofgadolinium DTPA for contrast-enhanced magnetic resonance imaging. MagnReson Imaging. 1985;3:27Y35.

6. Le Bihan D. Theoretical principles of perfusion imaging: application to mag-netic resonance imaging. Invest Radiol. 1992;27(suppl 2):S6YS11.

7. Cuenod CA, Balvay D. Perfusion and vascular permeability: basic conceptsand measurement in DCE-CT and DCE-MRI. Diagn Interv Imaging. 2013;94:1187Y1204.

8. Sourbron S. Technical aspects of MR perfusion. Eur J Radiol. 2010;76:304Y313.9. Weerasuriya A, Mizisin AP. The blood-nerve barrier: structure and functional

significance. Methods in molecular biology. 2011;686:149Y173.

10. Mizisin AP, Weerasuriya A. Homeostatic regulation of the endoneurial mi-croenvironment during development, aging and in response to trauma, diseaseand toxic insult. Acta Neuropathol. 2011;121:291Y312.

11. Cartwright MS, Passmore LV, Yoon JS, et al. Cross-sectional area referencevalues for nerve ultrasonography.Muscle Nerve. 2008;37:566Y571.

12. Patlak CS, Blasberg RG. Graphical evaluation of blood-to-brain transfer con-stants from multiple-time uptake data: generalizations. J Cereb Blood FlowMetab. 1985;5:584Y590.

13. Leenders KL, Perani D, Lammertsma AA, et al. Cerebral blood flow, bloodvolume and oxygen utilization: normal values and effect of age. Brain. 1990;113(pt 1):27Y47.

14. Larsson HB, Courivaud F, Rostrup E, et al. Measurement of brain perfusion,blood volume, and blood-brain barrier permeability, using dynamic contrast-enhanced T(1)-weighted MRI at 3 tesla.Magn Reson Med. 2009;62:1270Y1281.

15. Sourbron S, Ingrisch M, Siefert A, et al. Quantification of cerebral blood flow,cerebral blood volume, and blood-brain-barrier leakage with DCE-MRI. MagnReson Med. 2009;62:205Y217.

16. Cramer SP, Simonsen H, Frederiksen JL, et al. Abnormal blood-brain barrierpermeability in normal appearing white matter in multiple sclerosis investigatedby MRI. NeuroImage Clin. 2013;4:182Y189.

17. Ingrisch M, Sourbron S, Morhard D, et al. Quantification of perfusion andpermeability in multiple sclerosis: dynamic contrast-enhanced MRI in 3D at3 T. Invest Radiol. 2012;47:252Y258.

18. Jacobs JM, Love S. Qualitative and quantitative morphology of human suralnerve at different ages. Brain. 1985;108(pt 4):897Y924.

19. Cowen T, Ulfhake B, King RHM. Aging in the Peripheral Nervous System.4th ed. Philadelphia, PA: Elsevier Saunders; 2005.

20. Yamaguchi T, Kanno I, Uemura K, et al. Reduction in regional cerebral met-abolic rate of oxygen during human aging. Stroke. 1986;17:1220Y1228.

21. Sugimoto H, Miyaji N, Ohsawa T. Carpal tunnel syndrome: evaluation of mediannerve circulation with dynamic contrast-enhanced MR imaging. Radiology.1994;190:459Y466.

22. WangY, Tang P, Zhang L, et al. Gray-scale contrast-enhanced ultrasonography forquantitative evaluation of the blood perfusion of the sciatic nerves with crushinjury. Acad Radiol. 2011;18:1285Y1291.

23. Wang Y, Tang P, Zhang L, et al. Quantitative evaluation of the peripheral nerveblood perfusion with high frequency contrast-enhanced ultrasound. AcadRadiol. 2010;17:1492Y1497.

24. Ghasemi-Esfe AR, Khalilzadeh O, Vaziri-Bozorg SM, et al. Color and powerDoppler US for diagnosing carpal tunnel syndrome and determining its se-verity: a quantitative image processing method. Radiology. 2011;261:499Y506.

25. Akcar N, Ozkan S, Mehmetoglu O, et al. Value of power Doppler and gray-scaleUS in the diagnosis of carpal tunnel syndrome: contribution of cross-sectionalarea just before the tunnel inlet as compared with the cross-sectional area at thetunnel. Korean J Radiol. 2010;11:632Y639.

26. Sridharan A, Eisenbrey JR, Liu JB, et al. Perfusion estimation using contrast-enhanced 3-dimensional subharmonic ultrasound imaging: an in vivo study.Invest Radiol. 2013;48:654Y660.

27. Guggenberger R, Markovic D, Eppenberger P, et al. Assessment of median nervewith MR neurography by using diffusion-tensor imaging: normative and patho-logic diffusion values. Radiology. 2012;265:194Y203.

28. Buckley DL. Uncertainty in the analysis of tracer kinetics using dynamiccontrast-enhanced T1-weighted MRI. Magn Reson Med. 2002;47:601Y606.

29. Kershaw LE, Buckley DL. Precision in measurements of perfusion and mi-crovascular permeability with T1-weighted dynamic contrast-enhanced MRI.Magn Reson Med. 2006;56:986Y992.

FIGURE 6. Correlation analysis between NBV and age inhealthy controls. A trend of decreasing NBV with increasingage is observed (r = j0.41, P = 0.047).




peripheral nerve perfusion by dynamic.3 - clinical...

Documents