OOM

HMAa

pJb

S

AvtlfimdfdtsmwosssfOaSsh(l5thfistdl

Km

*EAfnibtTs

Neuroscience 172 (2011) 547–553

0d

LFACTORY BULB VENTRICLES AS A FREQUENT FINDING—A MYTHR REALITY? EVALUATION USING HIGH RESOLUTION 3 TESLA

AGNETIC RESONANCE IMAGING

TttmbieichcataOstssfdad(aitra

thiwgvnvpobtnattvT

. P. BURMEISTER,a* T. BITTER,b P. A. T. BALTZER,a

. DIETZEL,a O. GUNTINAS-LICHIUS,b H. GUDZIOLb

ND W. A. KAISERa

Institute of Diagnostic and Interventional Radiology, University Hos-ital - Friedrich Schiller University Jena, Philosophenweg 3, D-07740ena, Germany

Department of Otorhinolaryngology, University Hospital - Friedrichchiller University Jena, Lessingstrasse 2, 07743 Jena, Germany

bstract—Data on the prevalence of persistent olfactory bulbentricles (OBV) in humans remain contradictory. The aim ofhis study was to investigate the hypothesis of large cystic-ike OBVs filled with cerebrospinal fluid (CSF) as a frequentnding in magnetic resonance imaging (MRI). Fifty normos-ic volunteers (25 men and 25 women, mean 40 years) un-erwent 3 Tesla MRI of the anterior skull base. Normal smellunction was determined by testing of the odor thresholdiscrimination identification score using the Sniffin’ Sticksest kit. The voxel size of the constructive interference inteady state (CISS) sequence was 0.4�0.4�0.4 mm (TR 12.18s, TE 6.09 ms) using a 12-channel head coil. Image qualityas rated by three observers according to predefined criterian an ordinal scale. Additionally, contrast-to-noise (CNR) andignal-to-noise (SNR) ratios were calculated. Quantitativeignal intensity (SI) measurement of olfactory bulb (OB)tructures and small Virchow-Robin spaces (VRS) was per-ormed using multi planar reconstruction mode. Ninety-oneBs were eligible for evaluation. Image quality was rated asdequate in 55% and as excellent in 36% of cases. CNR andNR calculations resulted in values of 21.59 and 19.06, re-pectively. Wilcoxon signed rank test revealed significantigher SI values for OB center compared to OB surfaceP<0.001) and to OB base (P<0.001) but also significantower SI values compared to small VRS (P<0.001) in 94.5%. In.5%, SI measurement revealed signs for CSF-filled struc-ures in the OB. High-resolution 3 Tesla MRI did not verify theypothesis of large cystic CSF-filled OBVs as a frequentnding although evidence is growing that the hyperintenseignal in the center of OBs might be associated with intersti-ial or finely dispersed CSF/fluid or with tiny, histologicallyetectable remnants of OBVs. Crown Copyright © 2011 Pub-

ished by Elsevier Ltd on behalf of IBRO. All rights reserved.

ey words: olfactory bulb ventricles, MRI, brain develop-ent, volumetry.

Corresponding author. Tel: �49-3641-935826; fax: �49-3641-936767.-mail address: hartmut.burmeister@med.uni-jena.de (H. P. Burmeister).bbreviations: CBS, central bright structure; CISS, constructive inter-

erence in steady state-3D-fourier transformation; CNR, contrast-to-oise ratio; CSF, cerebrospinal fluid; MRI, magnetic resonance imag-

ng; OB, olfactory bulb; OBC, olfactory bulb cistern; OBV, olfactoryulb ventricle; ROI, region of interest; SI, signal intensity; SNR, signal-

o-noise ratio; SVZ, subventricular zone; T, Tesla; TA, acquisition time;

pE, echo time; TR, repetition time; VOE, ventriculo-olfactory exten-ion; VRS, Virchow-Robin spaces.

306-4522/11 $ - see front matter. Crown Copyright © 2011 Published by Elsevier Loi:10.1016/j.neuroscience.2010.10.068

547

he olfactory bulb (OB) is the first cerebral olfactory struc-ure, which processes afferent olfactory information fromhe olfactory receptor neurons (Gottfried, 2006). Sinceagnetic resonance imaging (MRI) of the olfactory systemecame possible in the late 1980s, (Suzuki et al., 1989)

maging of the OB focused on the depiction of the OB in itsntirety. Today, OB imaging in clinical practice is predom-

nantly limited to following aspects. First, imaging is indi-ated to rule out OB absence or hypoplasia in congenitalyposmia or anosmia (Yousem et al., 1996a) or to assessongenital malformations (Yousem et al., 1993; Blustajn etl., 2008). Other indications for OB imaging are tumors ofhe anterior skull base that may affect the OB (Abolmaali etl., 2008). Furthermore, MRI allows the assessment of theB volume (Yousem et al., 1997). The OB exhibits a high

tructural plasticity, in which the OB volume is correlated tohe afferent neural activity (Abolmaali et al., 2008). Con-equently, an OB volume decrease has been demon-trated for posttraumatic (Yousem et al., 1996b), postin-ectious (Mueller et al., 2005), idiopathic olfactory disor-ers (Rombaux et al., 2010), schizophrenia (Turetsky etl., 2000), Alzheimer’s disease (Rombaux et al., 2009a),epression (Negoias et al., 2010), or sinonasal diseaseRombaux et al., 2008). One recent study could also shown OB volume increase after successful olfactory rehabil-

tation in sinonasal diseases (Gudziol et al., 2009). Fur-hermore, a positive correlation of OB volume with ageelated olfactory function has been reported (Yousem etl., 1998).

These dynamic changes in OB volume are based onhe following exceptional characteristic of the OB: Theuman OB kept its ability to renew cell population because

n the OB, progenitor cells differentiate into neurons untilell into adulthood (Bédard and Parent, 2004). It is sug-ested that those neuroblasts are received from the sub-entricular zone (SVZ)—known to be a major source ofeural stem cells in the adult human brain—via a lateralentricular extension (Curtis et al., 2007). Many speciesossess an open tube between the lateral ventricle andlfactory bulb ventricle (OBV) allowing a free flow of cere-rospinal fluid (CSF) as well as a continuous SVZ betweenhe two regions (Rae, 1994). Until now, this structure hasot been demonstrated in the adult human brain. Curtis etl. (2007) provided a characterization of the human ven-riculo-olfactory neurogenic system containing the SVZ,he rostral migration stream, and an extension of the lateralentricle entitled as ventriculo-olfactory extension (VOE).his VOE appears to be fluid-filled, is connected via a

atent duct in the olfactory tract to the OB, and ends at thetd on behalf of IBRO. All rights reserved.

Ocmpefg

pcpasmTpR(deoa5pT7uctbala(nMb

wt

mtartvCtf

P

Icmamistewiuftfcte

I

ATi(mmapsl

D

A(m

1osob3I

r1

Fp(vr(fi8sbs

H. P. Burmeister et al. / Neuroscience 172 (2011) 547–553548

BV. However, these findings are controversially dis-ussed (Sanai et al., 2007). The organization of the rostraligratory stream around the VOE could support the hy-othesis that the migration of neuroblasts may be influ-nced by CSF circulation (Sawamoto et al., 2006). There-

ore, a CSF-filled large cystic OBV may facilitate the mi-ration of neuroblasts.

Recently, imaging studies presented isolated exem-lary images of hyperintense structures within the OBenter (central bright structure (CBS)) with pronounced T2rolongation using T2-weighted fast spin echo (Curtis etl., 2007; Duprez and Rombaux, 2010) (Fig. 1) or con-tructive interference in steady state-3D-fourier transfor-ation (CISS) sequences (Abolmaali et al., 2008, 2009).hese hyperintense structures were contradictorily inter-reted as bulbomalacia / neuromalacic bulb (Duprez andombaux, 2010), atrophy or small dysontogenetic cysts

Abolmaali et al., 2008), unmyelinated white matter (Schnei-er and Floemer, 2009), central fluid compartment (Curtist al., 2007), OBV-like structures (Smitka et al., 2009) orlfactory bulb ventricles (Abolmaali et al., 2009; Smitka etl., 2009). Smitka et al. (2009) reported a prevalence of9% OBVs with at least 4 mm in length in their studyopulation using T2-weighted fast spin echo sequences.his stands in disagreement with an OBV prevalence of% histologically verified in postmortem adult cadaverssed as reference in the same study. Although this dis-repancy was clearly stated, it remains unclear whetherhose hyperintense structures within the OB center areased on large cystic CSF-filled ventricles or whether theyre associated with interstitial CSF/fluid or with tiny, histo-

ogically detectable remnants of OBVs. Smitka et al. (2009)ssumed the postmortem resorption of CSF from OBVsJohnston et al., 2004; Walter et al., 2006) to be an expla-ation for the discrepancy between their results based onRI compared to histopathology. Further, in human em-

ig. 1. Olfactory bulbs with central bright structures (previously sup-osed to be olfactory bulb ventricles). (A) Coronal T2 weighted imageTR 12.18 ms, TE 6.09 ms, matrix 512�512, field of view 230 mm,oxel size 0.4�0.4�2.0 mm, averages 2, TA 8:50 min) and (B) cor-esponding parasagittal constructive interference in steady stateCISS) sequence image (TR 12.18 ms, TE 6.09 ms, matrix 512�512,eld of view 210 mm, voxel size 0.4�0.4�4.0 mm, averages 1, TA:08 min) of the olfactory bulb at 3 Tesla depicting the (1) olfactoryulcus, (2) crista galli, (3) olfactory bulbs in olfactory fossa with centralright structures (CBS) in T2w (A), and (4) proper olfactory artery. S,uperior; I, inferior; R, right; L, left.

ryos the OBV has been reported to obliterate around 19 n

eeks of “menstrual age,” being completely closed at theime of birth (Smitka et al., 2009).

Until now, data on a possible detection of human OBicrostructures—in particular of OBVs—based on quanti-

ative analysis of signal intensities (SI) using high field MRIre missing. One major advantage of high field magneticesonance imaging (i.e. at 3 Tesla) is an increased signal-o-noise ratio. This higher signal-to-noise ratio can be in-ested in better spatial resolution (Schmitt et al., 2004).onsequently, this in vivo high field MRI study investigates

he hypothesis of CSF-filled olfactory bulb ventricles as arequent finding.

EXPERIMENTAL PROCEDURES

articipant recruitment and sampling

n this prospective study, we examined 100 bulbi olfactorii of aonsecutive series of 50 normosmic volunteers. To ensure nor-osmic individual olfactory function in relation to the volunteers’ge odor threshold discrimination identification score was deter-ined by the Sniffin’ Sticks test (Kobal et al., 2000). Further

nclusion criteria were as follows: at least 18 years of age, noymptoms of common cold, allergic rhinitis, and infections relatingo the paranasal sinuses or respiratory tract, no medication influ-ncing cerebrospinal fluid balance, and a subjective feeling ofell-being. Not eligible for this investigation were known contra-

ndications against MRI, corticosteroid medication or long-termse of decongestant nasal sprays, craniocerebral trauma withracture of the anterior skull base, any known previous surgery ofhe paranasal sinuses, structural brain lesions of the central ol-actory pathway, presence of parosmia or phantosmia, or insuffi-ient image quality. Institutional review board approval was ob-ained and all volunteers gave written informed consent for thexamination.

maging procedures

ll volunteers underwent a 3 Tesla MRI examination (Magnetomim Trio, Siemens, Erlangen, Germany) of the anterior skull base

n our institute. The slice thickness of the axial CISS sequenceCasselman et al., 1993) was 0.4 mm (TR 12.18 ms, TE 6.09 ms,atrix 512�512, field of view 210 mm, voxel size 0.4�0.4�0.4m, flip angle 50°, bandwidth 130 Hz/Px, averages 1, time ofcquisition (TA) 8:08 min) using a dedicated 12-channel head coilrovided by the manufacturer. The volunteers were examined inupine position; slice orientation was parallel to the horizontalamella of cribriform plates.

ata analysis

cquired images were transferred to a Numaris/4 workstationSyngo-Software MR B15 Numaris/4, Siemens, Erlangen, Ger-any) for analysis.

One radiologist, specialized in skull base imaging (Observer, HPB, experience of 8 years), and two otorhinolaryngologists,ne specialized in surgery of the anterior skull base, paranasalinuses, and head and neck imaging (Observer 2, HG, experiencef 30 years), the other experienced in surgery of the anterior skullase, paranasal sinuses, and head and neck imaging (Observer, TB, experience of 5 years) evaluated all images in consensus.f unanimity could not be achieved, majority decision was decisive.

Image quality—with a special focus on the depiction of the OBegion—was ordinally categorized as follows: 0�insufficient,�adequate, 2�excellent.

Criteria for insufficient rating of image quality were pro-

ounced occurrence of movement, pulsation, off-resonance, or

boc

wet

av

t(f

(p

lu

(

psscOs

ao(Otop(optCmi

acRsbdsCi

onp

S

A(

foOnfl

cmm

clt(

P

Amup2yRbiab

Fso(rsC(fibttIoeSs

H. P. Burmeister et al. / Neuroscience 172 (2011) 547–553 549

anding artifacts preventing reliable evaluation due to duplicationr extinction of OB or vascular structures in the olfactory bulbistern (OBC).

The image quality was rated adequate if following criteriaere present: mild occurrence of artifacts, no duplications orxtinctions, continuous delineation of OB or vascular structures inhe OBC.

The image quality was rated as excellent if no or insignificantrtifacts were observed, allowing excellent visualization of OB orascular structures in the OBC.

Additionally, to assess the image quality for the relevant struc-ures in the olfactory fossa, the mean contrast-to-noise ratioCNR) was calculated for CSF and the OB according to theollowing formula.

CNR�1n�

i�1

n �SIiCSF�SIi

OB��i

SICSF�SI of CSF; SIOB�SI of OB; ��SI standard deviation of theure image noise)

In addition, the mean signal-to-noise ratio (SNR) was calcu-ated to quantify how much the signal has been corrupted by noisesing the CISS sequence according to the following formula.

SNR�1n�

i�1

n SIiCSF

�i

SI�SI of CSF; ��SI standard deviation of the pure image noise)Based on known anatomical characteristics of the encephalo-

eripheral nervous system and its vascularization in the anteriorkull base (Leblanc, 2001), multiplanar reconstructions with eachlice orientation indicted on the other reconstruction planes asolor-coded cross hairs were used for identification of the OB andBC. This approach allows for exact correlation of anatomicaltructures in different planes.

To evaluate SI differences within the olfactory bulb indicatingfluid filled central cavity and to evaluate the prevalence of

lfactory bulb ventricles, SI measurements of the CSF in the OBCmeasuring point 1) and of three different areas of interest in theB (measuring points 2–4) were performed (Fig. 2). For the latter,

he OB was divided into three thirds: top third (surface) of thelfactory bulb adjacent to the OB cistern from below (measuringoint no. 2), middle third in the OB center (central bright structureCBS)) (measuring point no. 3), and the lower third (base) of thelfactory bulb adjacent to the horizontal lamella of the cribriformlate (measuring point no. 4). Measuring point no.3 was posi-ioned at the brightest point of the middle third in the CBS. If noBS could be identified on at least two adjoining slices in theiddle third of the OB, measurement was performed at the max-

mum diameter of the equatorial section.Physiological structures being filled with CSF, circular-shaped

nd approximately identical in size and SI (Kwee and Kwee, 2007)ompared to a reputedly existing OB ventricle are small Virchow-obin spaces (VRS). Therefore, SI of small Virchow-Robinpaces type I—located in the basal ganglia—or type III—locatedetween the cerebral peduncles (Kwee and Kwee, 2007)—with aiameter of 0.7–1.2 mm were measured (measuring point 5) andet as reference (Fig. 2). Further, we evaluated whether or notBS showed sharp contours at a presumptive CSF/soft tissue

nterface and compared to VRS (set as reference).All SI measurements were performed in standardized vertical

rder in parasagittal reconstructed images for measuring pointso. 1–4 and in paraxial images for measuring point no. 5 using a

ixel lens. (

tatistical analysis

ll data analysis was performed using SPSS 15.0 for WindowsSPSS, Chicago, III, USA).

The exact two-sided Wilcoxon signed rank test was per-ormed to compare SI measurements values of the top third of thelfactory bulb (OB surface), middle third (central bright structure,B CBS), lower third of the olfactory bulb (OB base), cerebrospi-al fluid in Virchow-Robin spaces (CSF VRS), and cerebrospinaluid in the olfactory bulb cistern (CSF OBC).

Further, nonparametric Spearman’s rank correlation coeffi-ients were calculated to test statistical dependency between SIeasurements of OB structures or VRS and sex, age, and theaximum diameter of the OB equatorial section.

All statistical analyses in this study were based on a signifi-ance level ��5%. In order to address possible � error accumu-ation, P-values in this paper are given as calculated, for interpre-ation of results classical Bonferroni correction was consideredAickin and Gensler, 1996).

RESULTS

articipant recruitment and sampling

ll 50 volunteers (25 men and 25 women) showed nor-osmic olfactory function, met all inclusion criteria andnderwent 3 Tesla magnetic resonance imaging in thisrospective study between March 9, 2009 and April 26,010. The volunteers’ age ranged between 19 and 68ears (mean age 40.4, standard deviation 14.0 years).egarding our exclusion criteria, in four volunteers motion,anding, of resonance and/or pulsation artifacts led to an

nsufficient image quality and in one case, artifacts led ton insufficient image quality of a left hand sided olfactoryulb. Consecutively, nine bulbi olfactorii were excluded

ig. 2. Parasagittal constructive interference in steady state (CISS)equence image (TR 12.18 ms, TE 6.09 ms, flip angle 50°) of thelfactory bulb (A) and paraaxial image (B) of the brain stem at 3 Tesla:A) Directional order of measurements (dashed line, superior to infe-ior) and localization of measuring points for signal intensity (SI) mea-urement and exemplary corresponding SI values (case no. 85): (1)erebrospinal fluid (CSF) in olfactory bulb cistern (OBC) (SI�1038),

2) top third (surface) of the olfactory bulb adjacent to the OB cisternrom below (SI�156), (3) middle third (central bright structure spread-ng over 3/3rds of the olfactory bulb) (SI�441) (measuring point atrightest point—see enlarged display detail (grey arrow)), (4) lower

hird (base) of the olfactory bulb adjacent to the horizontal lamella ofhe cribriform plate (SI�96), and (B) three Virchow-Robin spaces typeII (VRS) located between the cerebral peduncles with a (5) diameterf 0.7 mm (SI�466), (6) diameter of 1.0 mm (SI�598), and (7) diam-ter of 1.1 mm (SI�770), demonstrating the dependency of measuredI values on the diameter (SI value increases with diameter). S,uperior; I, inferior; A, anterior; P, posterior; R, right; L, left.

seven male and two female; 9%). Forty-six right hand

swfn

I

I5tr1col

A

TStbacbl(

eit4smboo

Os

OdSc

AfiaitVdhfidItimmOtOl

mvCc(f1oH(uTMtc

FR(tRcp

T(

S

H. P. Burmeister et al. / Neuroscience 172 (2011) 547–553550

ided and forty-five left hand sided bulbi olfactorii (n�91)ere eligible for statistical and image analyses as well as

or evaluation of the contrast-to-noise ratio and signal-to-oise ratio of the CISS sequences.

mage quality

mage quality of the OB region was rated as adequate in5% and as excellent in 36% of cases. The evaluation ofhe contrast-to-noise ratio and the signal-to-noise ratioesulted in mean values of 21.59 (SD �3.9) for CNR and9.06 (SD �0.7) for SNR, respectively. Further, an exactontour sharpness of the CSF/soft tissue interface wasbserved in every VRS whereas no CBS showed a cystic-

ike demarcation with approximate sharp contours.

nalysis of the olfactory bulbs

he Wilcoxon signed rank test revealed significant higherI values for CBS compared to OB surface (P�0.001) and

o OB base (P�0.001). Further, no difference were shownetween OB surface and OB base SI values (P�0.806). Inddition, SI values for CSF in VRS proved to be signifi-antly higher compared to CBS, OB surface and to OBase (P�0.001 each), whereas the CSF in VRS showed

ower SI values in comparison with CSF in the OBCP�0.001) (Fig. 3, Table 1).

In five of ninety-one cases (5.5%) SI values of CBSxceeded the absolute minimum value of all CSF signal

ntensities of VRS measured (SI�382, set as threshold). Inwo of ninety-one individual cases (2.2%) (no. 19 (female,2 years, right sided OB) and no. 85 (male, 44 years, rightided OB)) SI values of CBS not only exceeded the above-entioned threshold value of CSF signal intensity in VRSut also reached or exceeded the intra individual SI valuef CSF in VRS. In addition, only in the latter two cases SIf CBS exceeded intra individual maximum SI values of

ig. 3. Box-and-whisker diagram comparing signal intensity values.esults of signal intensity values of the top third of the olfactory bulb

OB surface), middle third (central bright structure, OB CBS), lowerhird of the olfactory bulb (OB base), cerebrospinal fluid in Virchow-obin spaces (CSF VRS), and cerebrospinal fluid in the olfactory bulb

Cistern (CSF OBC) with according P-values (Wilcoxon test) for com-arison of OB structures.

B surface or OB base for more than the quadruplicatedtandard deviation (Fig. 4).

Regarding a correlation between SI measurements ofB structures or VRS and sex, age, and the maximumiameter of the OB equatorial section nonparametricpearman’s rank correlation analyses showed no statisti-ally significant results (P�0.05).

DISCUSSION

ccording to our results, 3T high-resolution imaging veri-ed a higher SI for the OB center compared to the uppernd lower OB outer rim using a CISS sequence. However,

n 94.5% SI values of the CBS showed no intersection withhe lower threshold value of all CSF signal intensities inRS measured. Due to the latter and a lack of a cystic-likeemarcation with sharp contours, we cannot confirm theypothesis of large CSF-filled cystic OBVs as a frequentnding with at least 4–7 mm length and/or a transverseiameter of about 1.5 mm as stated by Curtis et al. (2007).n 3.3%, SI values of the CBS intersected with the lowerhreshold value for CSF in VRS. However, due to interndividual fluctuation range of SI values and due to the

issing intersection with intra individual CSF SI values itay not be concluded that these CBS represent CSF-filledBVs. Only in two cases (2.2%), evidence is growing that

hese CBSs might be associated with tiny remnants ofBVs (with regard to our reference standard equal to or

ess than 1.2 mm in size; refer to Data analysis).Due to its high soft tissue contrast, MRI is the imaging

ethod of choice to visualize pathology within the OB inivo. MRI-sequences with stronger T2 weighting (e.g.ISS) seems advantageous (Abolmaali et al., 2008) be-ause this technique generates bright signals from fluidse.g. perilymph and endolymph, and CSF) and very low SIrom all other tissues and structures (Casselman et al.,993). Regarding to the basic principles of the techniquef magnetic resonance imaging, an analogon to theounsfield scale as it is used in computed tomography

which allows a qualitative analysis based on specific val-es for various tissues) does not exist for MR imaging.herefore, the arbitrary chosen, dimensionless units forR signal intensity only allow for intraindividual quantita-

ive MR signal intensity analysis as it was taken into ac-ount in this study. The quantitative SI analyses of the

able 1. Descriptive statistics of signal intensity measurementsn�91)

Minimum Maximum Mean Std. deviation(SD)

ignal intensityvalues

OB surface 64.00 381.00 177.05 52.42OB CBS 150.00 563.00 275.50 69.89OB Base 68.00 372.00 180.65 59.40CSF VRS 382.00 966.00 641.21 132.91CSF OB Cistern 581.00 1333.00 937.62 147.74

ISS sequence used in this study—in particular, with re-

gvqactprprist

tmmipctOVca

(soap

Op

OtplfOO

oiuamSspoOsisalmpoptriepc

F(bis s excludet n of this a

H. P. Burmeister et al. / Neuroscience 172 (2011) 547–553 551

ard to the high resolution of 0.4 mm—resulted in excellentalues for CNR and SNR. These results confirm the se-uence used in this study to be most suitable regarding ourim to discriminate between nervous tissue of the OB andystic OBVs solely filled with CSF. The disadvantages ofhe CISS sequence are occurrence of artifacts and theoor contrast between gray and white matter. The occur-ence of artifacts cannot be avoided completely. It was notossible to eliminate these artifacts completely out of theegion of interest (Dietrich et al., 2008). However, accord-ng to our results of image quality rating, artifacts did notignificantly restrict the evaluation of the OB concerninghe aim of this study.

Another limitation that has to be mentioned is the spa-ial resolution of MR imaging. Taking into account theagnetic field strength of 3 Tesla and the technical equip-ent available (e.g., receiver coils) for human in vivo stud-

es a voxel resolution of 0.4�0.4�0.4 mm is at the limits oferformance. However, this resolution—in particular inonsideration of partial volume effects—is not high enougho verify single voxel hyperintensities as small CSF filledBVs. With regard to our reference standard of smallRSs, partial volume effects did not influence the identifi-ation of CSF-filled structures with a diameter between 0.7nd 1.2 mm (Fig. 2B) to a significant extent.

Signal intensity evaluation using a region of interestROI) measurement tool would have been favored in thistudy. However, due to a lack of exact contour sharpnessf CBSs, the tiny dimension of the structures measured,nd to be methodologically consistent we decided to use aixel lens as measurement tool.

The existence and high prevalence of large cysticBVs would be of immense interest regarding the inter-

ig. 4. Line chart showing all case-related signal intensity values ofblack markers), Virchow-Robin spaces type I and III (VRS) (yellow mulb (green markers), basal layer of olfactory bulb (blue markers). In

ndividual maximum SI values of OB surface or OB base for more thaignal intensity values (red circles). Missing values represent volunteero color in this figure legend, the reader is referred to the Web versio

retation of results of studies dealing with the subject of h

B volumetry. Neither instructions how to measure olfac-ory bulb volume (Rombaux et al., 2009b) nor recentlyublished volumetric studies took into account yet, that a

arge cystic OBVs volume should have been subtractedrom the OB volume. According to our results, very smallBVs might exist, but should not compromise results ofB volumetry to a significant extent.

Additional aspects support the complexity of evaluationf human OBs. In 2009, Schneider and Floemer (2009)

nterpreted the T2-hyperintensity in the OB center to benmyelinated white matter that—undergoing a predefinednd predictable maturational process—disappears at aedian age of 5 years. This runs counter to the results ofmitka et al. (2009)—published in the same year—showingimilar T2-hyperintensities in adults (19–79 years) inter-reted as OBVs. Although OBVs are confirmed as frequentlfactory structures in staged human embryos (Muller and’Rahilly, 2004), systematic data on the prevalence of per-isting human OBVs in childhood or adulthood are miss-

ng. In 1987, Roy et al. (1987) described a single rightided large cystic OBV that was found incidentally duringutopsy of a 43-year-old man. Further, Andrews and Hu-

ette (1993) reported an infant with macrocephaly, abnor-al neuronal migration, persistent large cystic OBV, and aersistence of the fetal connection between lateral andlfactory bulb ventricles. In addition, Curtis et al. (2007)resented a macroscopic preparation showing a large cys-ic-like structure in the OB center. Further, neither in neu-opathologic case reports nor in studies presenting MRmages of OBV like structures (Curtis et al., 2007; Smitkat al., 2009) in adults a systematic evaluation of size,osition, nor OBV extension was performed. In addition,ontradictory details are also given about whether in adult-

interest measured cerebrospinal fluid in olfactory bulb cistern (OBC)entral bright structure (CBS) (red markers), surface layer of olfactoryes (no. 19 and no. 85), signal intensity values of CBS exceed intradruplicated standard deviation as well as they do intersect with VRSd due to insufficient image quality. For interpretation of the referencesrticle.

areas ofarkers), c

two casn the qua

ood OBVs have a clear ependymal lining (Roy et al.,

1mwTi

rrhcmlOt(pioOppmcliit1osOtcwtisafiwiM

htiaifpivaaameOa

bedraA

Rohfissc

Actbof

A

A

A

A

B

B

B

C

C

D

D

G

G

J

H. P. Burmeister et al. / Neuroscience 172 (2011) 547–553552

987), are loosely lined with ependymal-like cells (ependy-al remnants found histologically) (Curtis et al., 2007), orhether ependyma is lacking (Smitka et al., 2009) at all.hose uncertainties should be resolved in future histolog-

cal studies.The OB is a paleocortical extension and differs from

egular cellular organization of the neocortex and the ce-ebral white matter. Cellular organization of the OB inuman fetuses is described as follows (from surface toenter): olfactory nerve fiber layer, external plexiform layer,itral cell layer, internal plexiform layer, subventricular

ayer, ventricular layer, olfactory bulb ventricle (Muller and’Rahilly, 2004). This differs from cellular organization of

he OB in human adults as described by Bhatnagar et al.1987): olfactory nerve layer, glomerular layer, externallexiform layer, mitral cell layer (a single layer of neurons),

nternal plexiform layer, granular layer, intrabulbar anteriorlfactory nucleus. Apart from developmental changes inB cytoarchitecture during infancy, we hypothesize thathysiological SI differences of the adult OB center com-ared to the OB outer rim using T2 weighted sequencesay be the result of a summation of outer and inner OB

ell layers (e.g. outer rim: olfactory nerve layer, glomerularayer, and external plexiform layer; CBS: mitral cell layer,nternal plexiform layer, granular layer, and fibers connect-ng to the olfactory tract). Looking at histological OB sec-ions regarding the density of cell layers (Bhatnagar et al.,987) suggest this hypothesis. In addition, the results ofur study corroborate that the frequent finding of CBSseen in T2 weighted sequences is not due to large cysticBVs. Taking into account the complex OB cytoarchitec-

ure, future imaging studies should consider a systematicomparison of different high-resolution T1- and/or T2-eighted sequences imaging the OB in vivo. Thinner slice

hicknesses (if applicable), the use of surface coils, and—ift is feasible—ROI measurement for quantitative SI analy-is should be used to evaluate OB microstructures. Inddition, imaging studies of human ex vivo OBs with highereld strengths (e.g. 7 Tesla or above) and a comparisonith dedicated histological sections could be useful to

dentify the appearance of human OB cellular layers usingRI.

Although the presence of an OBV or CSF in the OBas not been proven to be a basic prerequisite for migra-

ion of neuroblasts from the SVZ, we hypothesize that thenconsistent prevalence and configuration of the CBSs indults could be associated with the inter-individual variabil-

ty of neoneurogenesis. One possible explanation for theew observed adult large cystic OBVs might be abnormalrenatal developmental neuronal migration, regarding the

ncidence of OBVs, experimental studies in rodents re-ealed that changes in OB cytoarchitecture due to anbnormal regulation of OB interneuron differentiation led ton enlargement of OBVs (Long et al., 2007). Therefore,bnormal processes of interneuron differentiation in hu-ans (e.g. in neurodegenerative disorders) might effect annlargement of OBV remnants. Due to this enlargement,BV remnants in humans could become visible again

nd—as a future application—high resolution MRI could

ecome a suitable tool to evaluate accordant neurodegen-rative disorders. Furthermore—if OBVs ought not to beetected—a detection of changes of the CBS-outer rim-atio using MRI could be useful as a screening method forging or neuropsychological diseases (e.g. schizophrenia,lzheimer’s disease, or depression).

CONCLUSION

eferring to the question whether in vivo OBVs are a mythr not: High-resolution 3 Tesla MRI did not verify theypothesis of large cystic CSF-filled OBVs as a frequentnding, although evidence is growing that the hyperintenseignal in the center of OBs might be associated with inter-titial or finely dispersed CSF/fluid or with tiny, histologi-ally detectable remnants of OBVs.

cknowledgments—All authors exclude any actual or potentialonflicts of interest including any financial, personal or other rela-ionships with other people or organizations within three years ofeginning the submitted work that could inappropriately influence,r be perceived to influence, this work. No funding was receivedor this investigation.

REFERENCES

bolmaali N, Gudziol V, Hummel T (2008) Pathology of the olfactorynerve. Neuroimaging Clin N Am 18:233–242.

bolmaali N, Hummel T, Damm M (2009) [Two- and three-dimen-sional, morphologic and functional MR-imaging in smelling disor-ders]. Laryngorhinootologie 88:10–16.

ickin M, Gensler H (1996) Adjusting for multiple testing when report-ing research results: the Bonferroni vs Holm methods. Am J PublicHealth 86:726–728.

ndrews PI, Hulette CM (1993) An infant with macrocephaly, abnor-mal neuronal migration and persistent olfactory ventricles. ClinNeuropathol 12:13–18.

édard A, Parent A (2004) Evidence of newly generated neurons inthe human olfactory bulb. Dev Brain Res 151:159–168.

hatnagar KP, Kennedy RC, Baron G, Greenberg RA (1987) Numberof mitral cells and the bulb volume in the aging human olfactorybulb: a quantitative morphological study. Anat Rec 218:73–87.

lustajn J, Kirsch CFE, Panigrahy A, Netchine I (2008) Olfactoryanomalies in CHARGE syndrome: imaging findings of a potentialmajor diagnostic criterion. AJNR Am J Neuroradiol 29:1266–1269.

asselman JW, Kuhweide R, Deimling M, Ampe W, Dehaene I, MeeusL (1993) Constructive interference in steady state-3DFT MR imag-ing of the inner ear and cerebellopontine angle. AJNR Am JNeuroradiol 14:47–57.

urtis MA, Kam M, Nannmark U, Anderson MF, Axell MZ, Wikkelso C,Holtas S, van Roon-Mom WMC, Bjork-Eriksson T, Nordborg C,Frisen J, Dragunow M, Faull RLM, Eriksson PS (2007) Humanneuroblasts migrate to the olfactory bulb via a lateral ventricularextension. Science 315:1243–1249.

ietrich O, Reiser MF, Schoenberg SO (2008) Artifacts in 3-T MRI:physical background and reduction strategies. Eur J Radiol 65:29 –35.

uprez TP, Rombaux P (2010) Imaging the olfactory tract (cranialnerve #1). Eur J Radiol 74:288–298.

ottfried JA (2006) Smell: central nervous processing. Adv Otorhino-laryngol 63:44–69.

udziol V, Buschhuter D, Abolmaali N, Gerber J, Rombaux P, Hum-mel T (2009) Increasing olfactory bulb volume due to treatment ofchronic rhinosinusitis—a longitudinal study. Brain 132:3096–3101.

ohnston M, Zakharov A, Papaiconomou C, Salmasi G, Armstrong D

(2004) Evidence of connections between cerebrospinal fluid and

K

K

L

L

M

M

N

R

R

R

R

R

R

S

S

S

S

S

S

T

W

Y

Y

Y

Y

Y

H. P. Burmeister et al. / Neuroscience 172 (2011) 547–553 553

nasal lymphatic vessels in humans, non-human primates and othermammalian species. Cerebrospinal Fluid Res 1:2.

obal G, Klimek L, Wolfensberger M, Gudziol H, Temmel A, OwenCM, Seeber H, Pauli E, Hummel T (2000) Multicenter investigationof 1,036 subjects using a standardized method for the assessmentof olfactory function combining tests of odor identification, odordiscrimination, and olfactory thresholds. Eur Arch Otorhinolaryngol257:205–211.

wee RM, Kwee TC (2007) Virchow-Robin spaces at MR imaging.Radiographics 27:1071–1086.

eblanc A (2001) Olfactory nerves. In: Encephalo-peripheral nervoussystem: vascularisation, anatomy, imaging (Leblanc A, ed), pp1–19. Berlin Heidelberg New York: Springer-Verlag.

ong JE, Garel S, Alvarez-Dolado M, Yoshikawa K, Osumi N, Alvarez-Buylla A, Rubenstein JLR (2007) Dlx-dependent and -independentregulation of olfactory bulb interneuron differentiation. J Neurosci27:3230–3243.

ueller A, Rodewald A, Reden J, Gerber J, von Kummer R, HummelT (2005) Reduced olfactory bulb volume in post-traumatic andpost-infectious olfactory dysfunction. Neuroreport 16:475–478.

uller F, O’Rahilly R (2004) Olfactory structures in staged humanembryos. Cells Tissues Organs 178:93–9116.

egoias S, Croy I, Gerber J, Puschmann S, Petrowski K, Joraschky P,Hummel T (2010) Reduced olfactory bulb volume and olfactorysensitivity in patients with acute major depression. Neuroscience169:415–421.

ae AS (1994) Nodules of cellular proliferation in sheep olfactoryventricle. Clin Neuropathol 13:17–18.

ombaux P, Duprez T, Hummel T (2009a) Olfactory bulb volume inthe clinical assessment of olfactory dysfunction. Rhinology 47:3–9.

ombaux P, Grandin C, Duprez T (2009b) How to measure olfactorybulb volume and olfactory sulcus depth? B-ENT 5 (Suppl 13):53– 60.

ombaux P, Potier H, Bertrand B, Duprez T, Hummel T (2008) Olfac-tory bulb volume in patients with sinonasal disease. Am J Rhinol22:598–601.

ombaux P, Potier H, Markessis E, Duprez T, Hummel T (2010)Olfactory bulb volume and depth of olfactory sulcus in patients withidiopathic olfactory loss. Eur Arch Otorhinolaryngol 267:1551–1556.

oy EP, Frost JL, Schochet SS (1987) Persistent olfactory bulb ven-

tricle. Clin Neuropathol 6:86–87.

anai N, Berger MS, Garcia-Verdugo JM, Alvarez-Buylla A (2007)Comment on “Human neuroblasts migrate to the olfactory bulb viaa lateral ventricular extension.” Science 318:393.

awamoto K, Wichterle H, Gonzalez-Perez O, Cholfin JA, Yamada M,Spassky N, Murcia NS, Garcia-Verdugo JM, Marin O, RubensteinJLR, Tessier-Lavigne M, Okano H, Alvarez-Buylla A (2006) Newneurons follow the flow of cerebrospinal fluid in the adult brain.Science 311:629–632.

chmitt F, Grosu D, Mohr C, Purdy D, Salem K, Scott KT, Stoeckel B(2004) [3 Tesla MRI: successful results with higher field strengths].Radiologe 44:31–47.

chneider JF, Floemer F (2009) Maturation of the olfactory bulbs: MRimaging findings. AJNR Am J Neuroradiol 30:1149–1152.

mitka M, Abolmaali N, Witt M, Gerber JC, Neuhuber W, BuschhueterD, Puschmann S, Hummel T (2009) Olfactory bulb ventricles as afrequent finding in magnetic resonance imaging studies of theolfactory system. Neuroscience 162:482–485.

uzuki M, Takashima T, Kadoya M, Takahashi S, Miyayama S, TairaS (1989) MR imaging of olfactory bulbs and tracts. AJNR Am JNeuroradiol 10:955–957.

uretsky BI, Moberg PJ, Yousem DM, Doty RL, Arnold SE, Gur RE(2000) Reduced olfactory bulb volume in patients with schizophre-nia. Am J Psychiatry 157:828–830.

alter BA, Valera VA, Takahashi S, Ushiki T (2006) The olfactoryroute for cerebrospinal fluid drainage into the peripheral lymphaticsystem. Neuropathol Appl Neurobiol 32:388–396.

ousem DM, Geckle RJ, Bilker WB, Doty RL (1998) Olfactory bulb andtract and temporal lobe volumes. Normative data across decades.Ann N Y Acad Sci 855:546–555.

ousem DM, Geckle RJ, Bilker W, McKeown DA, Doty RL (1996a) MRevaluation of patients with congenital hyposmia or anosmia. AJRAm J Roentgenol 166:439–443.

ousem DM, Geckle RJ, Bilker WB, McKeown DA, Doty RL (1996b)Posttraumatic olfactory dysfunction: MR and clinical evaluation.AJNR Am J Neuroradiol 17:1171–1179.

ousem DM, Geckle RJ, Doty RL, Bilker WB (1997) Reproducibilityand reliability of volumetric measurements of olfactory eloquentstructures. Acad Radiol 4:264–269.

ousem DM, Turner WJ, Li C, Snyder PJ, Doty RL (1993) Kallmannsyndrome: MR evaluation of olfactory system. AJNR Am J Neuro-

radiol 14:839–843.

(Accepted 25 October 2010)(Available online 31 October 2010)