high resolution imaging of the mouse inner ear by microtomography: a new tool in inner ear research

High Resolution Imaging of theMouse Inner Ear by

Microtomography: A New Tool inInner Ear Research

M.P. VAN SPAENDONCK,1,2 K. CRYNS,3 P.H. VAN DE HEYNING,2

D.W. SCHEUERMANN,1 G. VAN CAMP,3 AND J.-P. TIMMERMANS1*1Laboratory of Cell Biology and Histology, University of Antwerp,

B-2020 Antwerpen, Belgium2Dept. of Otolaryngology, University of Antwerp, B-2020 Antwerpen, Belgium

3Dept. of Medical Genetics, University of Antwerp, B-2020 Antwerpen, Belgium

ABSTRACTA newly developed desktop microtomograph was used to evaluate

whether it is suitable for visualizing the three-dimensional (3D) morphologyof the mouse inner ear (at a micrometer level) and whether it is applicableas a fast screening tool to detect hereditary abnormalities in this organ. Tothis end, the epistatic circler, a mutant mouse showing abnormal circlingbehaviour, was used as a model. The inner ears were dissected out, form-aldehyde-fixed, and scanned at maximal resolution along the longitudinalaxis. After segmentation, stacks of tomographic images were used for 3Dreconstruction of the bony labyrinth. Finally, the obtained data were cor-related with subsequent conventional histological examination. The spatialresolution (8 mm) achieved by this instrument, was found to be far superiorto that obtained by conventional computer tomography (CT) and magneticresonance (MR)-imaging equipment. The technique provides detailed tomo-graphic images of the bony labyrinths and enables an adequate 3D recon-struction of the inner ear structures in this small mammal. In addition, itallows a screening for pathologic specimens prior to the more time- andlabour-consuming histological techniques, which are still essential togather information at a (sub)cellular level. This imaging technique can beregarded as a valuable tool in future research on hereditary inner earabnormalities. Anat Rec 259:229–236, 2000. © 2000 Wiley-Liss, Inc.

Key words: computer tomography; inner ear; vestibular develop-ment; three-dimensional reconstruction; epistatic cir-cler; mouse

Mouse models play a key role in today’s otovestibulargenetic research, either as existing mutants or as trans-genic or knockout models. In order to adequately examinewhether their otovestibular labyrinths show a normalmorphology, three-dimensional (3D) reconstruction hasproven to be extremely useful.

To this end, a number of medical imaging techniques,such as computer tomography (CT) and magnetic reso-nance (MR), which nowadays are commonly used to assesshuman inner ear morphology and clinical diagnosis, areapplied (Reisser et al., 1996; Isono et al., 1997). However,these instruments still suffer from a limited resolution,which becomes a severe drawback when extremely small

structures like murine inner ears have to be investigated.Therefore, the classical way to obtain 3D reconstructionsof the latter structures has mainly been based on theconventional labour-intensive histological serial section-ing (Harada et al., 1990). Apart from the typical limita-

*Correspondence to: J-P Timmermans, Laboratory of Cell Bi-ology & Histology, University of Antwerp (RUCA), Groenenborg-erlaan 171, B-2020 Antwerpen, Belgium.E-mail: jptimmer@ruca.ua.ac.be

Received 25 May 1999; Accepted 24 February 2000

THE ANATOMICAL RECORD 259:229–236 (2000)

© 2000 WILEY-LISS, INC.

tions of this technique regarding uneven slice thickness,loss or damage of consecutive sections and shrinkage, anadditional difficulty lies in a proper orientation and align-ment of consecutive sections, which is essential to obtainaccurate 3D reconstructions (Henson et al., 1994). Re-cently, new imaging technologies producing a muchhigher resolution have become available for laboratorypurposes. A newly developed desktop CT scanner (Sasovand Van Dyck, 1998) was used to study the inner earstructure of the epistatic circler mouse, a pre-existingmutant with an unknown otovestibular morphology. Pos-sible harmful effects on tissue morphology were evaluatedat the light microscopical level by comparison with innerears that were not scanned.

MATERIALS AND METHODSEpistatic Circler Mice

A small proportion of the F2 generation in the offspringof the cross between C57L/J and SWR/J mice shows ab-normal circling behaviour (Doolittle, 1963; Taylor, 1976).This abnormal phenotype is believed to be due to homozy-gosity for two genes, viz. Ecl and Ecs. The circling F2individuals are referred to as epistatic circlers. The innerears of three circling F2 individuals were scanned by themicrotomograph, as were those of three normally behav-ing SWR/J mice serving as controls. All scanned innerears were afterwards further processed for histology (in-cluding those of one epistatic circler and one SWR/J mousethat were not scanned).

Dissection and Tissue HandlingAll animal handling was done as prescribed by Euro-

pean Community Directive 86/609/EEC. The mice weresacrificed by cervical dislocation after which the temporalbones were quickly removed. The bony labyrinths wereimmediately dissected out of the temporal bones by sepa-rating them from the middle ear along a suture line.Ice-cold 4% buffered paraformaldehyde was gently per-fused through the round and oval windows. Followingfurther immersion fixation in the same fixative for 48 hr at4°C, the specimens were transferred to phosphate-buff-ered saline (0.01M, pH: 7.4) in which the inner ears were

stripped of excessive bone and adherent structures suchas muscle and paraflocculus. Subsequently, all inner earswere visualized under the stereomicroscope using transil-lumination, a technique which renders the bony capsuleslightly transparent.

MicrotomographyThis newly developed desktop X-ray microtomograph

(Skyscant, Aartselaar, Belgium) consists of a compact mi-crofocus X-ray tube, a slow scan CCD camera, and a Pen-tium computer processor (Sasov and Van Dyck, 1998). It isbased on the same working principles as a conventionalCT scanner, but uses instead of parallel X-rays a conicalX-ray beam generated by an X-ray tube with a very smallfocus at an energy level of 10–100 keV (maximal current100 mA, maximal voltage 80 kV). The object is transversedby this conical X-ray beam and the resulting signals arerecorded by a two-dimensional CCD camera (512 3 512pixels), producing an enlarged radiograph of the object.The magnification can be altered by changing the distancebetween the source, object, and camera (Fig. 1). Usingsmall samples, a spatial resolution as high as 6 mm can beachieved. Optimal contrast is obtained by tuning the en-ergy of the X-ray tube to the density of the tissue, char-acterized by the attenuation length.

Unlike conventional CT scanners the object to bescanned is rotated instead of the X-ray source. In this way,radiographic projections from different viewing angles areobtained, recorded, and stored. A 3D reconstruction iscalculated from these projections using the filtered back-projection algorithm. From this three-dimensional data-set, tomographic images are generated. The techniqueworks under ambient conditions, does not require anyspecimen preparation nor staining, and no harmful heat-ing of the specimen occurs.

Before being scanned, the inner ears were covered withVaselinet to prevent damage by dehydration and weremounted on the stage of the scanner using Plasticinet.Each inner ear (dimensions 3.5 3 2 3 2.5 mm) was posi-tioned in a longitudinal orientation and scanned at max-imal resolution, depending on the sample size, i.e., theenlarged radiograph should maximally cover but at thesame time be kept within the scope of the CCD camera inall viewing angles. Total scanning time was 30 min. Afterscanning, the bony labyrinths were further processed forhistology.

Morphometric Analysis of the Dimensions ofthe Semicircular Canals

Based on the calibrated tomographic images of the leftand right inner ears of three control (SRW/J) and threeepistatic circler mice, the diameter of each of the semicir-cular canals was determined. In all cases the point formeasurement was taken at the top of each semicircularcanal opposite to the vestibule. For statistical analysis, anon-parametric Mann Whitney-U test was performed.

3D ReconstructionSegmentation was performed on the outer limits of the

bony labyrinth thus revealing after 3D reconstruction thecontours of the fluid-filled space inside the bony capsulecontaining both the perilymphatic and endolymphaticstructures. Segmentation and subsequent 3D reconstruc-tions based on these tomographic images were performedusing commercially available software (Surfdriver 2.5.5t).

Fig. 1. Schematic drawing of the principle design of the desktopX-ray-microtomograph.

230 VAN SPAENDONCK ET AL.

Histological ProcessingEvaluations of inner ear morphology usually are based

on sections parallel to the modiolus, i.e., orthogonal to theplane of the images generated by the microtomograph.The temporal bones were completely decalcified in 5%EDTA (1 week), dehydrated in a graded alcohol series, andembedded in paraffin. Serial 10 mm-thick sections werecut parallel to the modiolus. All sections were stained withhematoxylin-eosin.

RESULTSSerious abnormalities of the bony labyrinth such as

severe distortion of the lateral semicircular canal weredetected by means of transillumination. However, lessconspicious deformations such as a slight narrowing of thecanal could not be discerned in sufficient detail. In gen-eral, scanning of a mouse inner ear at a slice thickness of8 mm resulted in about 440 tomographic images oriented

orthogonally to the axis of the modiolus. Under thesecircumstances, the spatial resolution (8 mm) is 100 timeshigher than that obtained by currently used clinical CTscanners. Both the bony cochlear structures (the osseousspiral lamina, the thin wall of the modiolus and the sec-ondary osseous spiral lamina at the basal portion of thecochlear duct) and the bony vestibular structures (thevestibule with its recesses, the slight widening of theampullae of the semicircular canals and the common crus)can be studied in sufficient detail (Fig. 2a–h). The softmembraneous tissues such as the stria vascularis and thelimbus are visible but not so sharply delineated sincecontrast was such that bony structures were easily distin-guishable from the surrounding soft tissues and fluidswhich greatly facilitated the segmentation process. Thecontours of the bony labyrinth (the fluid-filled space insidethe bony capsule containing both the perilymphatic andendolymphatic structures) were delineated in the com-

Fig. 2. a–h: A selection from a stack of 442 tomographic imagesfrom the left inner ear of an epistatic circler mouse showing a distinctnarrowing at the level of its lateral semicircular canal. Image-level isindicated on the scout view at the left upper corner. Cochlear structures(a–c), vestibular structures (d), semicircular canals (e–h, overleaf). Theseverely narrowed aspect of the lateral semicircular canal (CSL) is indi-cated by an open arrow in f (cross section at level of maximum reached

diameter). The arrows in g and h show the normal maximum diameter ofthe anterior semicircular canal (CSA) and the posterior semicircular canal(CSP). AOA, ampulla ossea anterior; AOL, ampulla ossea lateralis; COC,crus osseum commune; CSA, canalis semicircularis anterior; CSL, ca-nalis semicircularis lateralis; CSP, canalis semicircularis posterior; FMAI,fundus meatus acusticus internus; FV, fenestra vestibuli (ovale); LSO,lamina spiralis ossea; MOD, modiolus; VEST, vestibulum.

231HIGH RESOLUTION IMAGING WITH MICROTOMOGRAPHY

plete set of consecutive images by segmentation thus re-vealing its shape after 3D reconstruction (Fig. 3a–d).

Whereas the bony labyrinths including the semicircularcanals of the normally behaving SWR/J mice showed anormal morphology, a variety of abnormalities at the levelof the lateral semicircular canals of the epistatic circlermice (ranging from slight narrowing to complete interrup-tion) were revealed on both the tomographic images and3D reconstructions.

Figures 2a–h and 3c,d show the tomographic imagesand the 3D reconstuction of the same left bony labyrinth ofan epistatic circler. The narrowing at the level of thelateral semicircular canal which is observed on the two-dimensional tomographic images (Fig. 2f) is clearly visibleon the 3D reconstructions (Fig. 3c,d).

Morphometric analysis of the anterior semicircular ca-nal revealed a mean diameter of 193 6 19 mm (SD) for thecontrol group and 177 6 15 mm (SD) for the epistaticcircler. Similar values were obtained for the posteriorsemicircular canal (control: 186 6 21 mm [SD]; epistaticcircler: 185 6 5 mm [SD]). The mean diameter for thelateral semicircular canal in the control group amountedto 179 6 14 mm (SD), whereas in the epistatic circler groupa mean value of 14 6 22 (SD) was observed. It should also

be noted that the latter value includes a number of cases(n 5 4) where this canal was completely interrupted (Fig. 4).

After microtomography and histological processing, con-ventional midmodiolar sections were cut from the decalci-fied paraffin-embedded inner ears (Fig. 5a–c). The preced-ing CT procedure did not seem to have affected the micro-scopic structure. No differences could be noted betweenhematoxylin eosin-stained paraffin sections of the decal-cified samples scanned in the tomograph compared tosamples not previously scanned. Abnormalities at the lat-eral semicircular canals of the epistatic circler mice werealso visible on the histological sections. Figure 5c showsthe same narrowing of the left lateral semicircular canalpreviously depicted in Figures 2f and 3c,d. The orientationof the serial sections is orthogonal to that of the tomo-graphic images, explaining the different appearance ofFigures 2f and 5c.

DISCUSSIONMouse models of inner ear diseases have proven a very

valuable tool in otovestibular research. Genetic and mor-phological research based on animal models with innerear abnormalities not only greatly contributes to a betterunderstanding of inner ear ontogenesis and physiology

Figure 2. (Continued.)

232 VAN SPAENDONCK ET AL.

Fig. 3. a: 3D reconstruction of the bony labyrinth (left inner ear of anepistatic circler mouse). b: 3D image (to be viewed with conventional 3Dred/green spectacles), of the same 3D reconstruction as a. c,d: 3Dreconstruction of the left inner ear of another epistatic circler. Therestricted diameter of the lateral semicircular canal (CSL) is clearly visible(red arrow). White arrows indicate normal dimensions of the anteriorsemicircular canal (CSA) and the posterior semicircular canal (CSP).

(Same specimen as depicted in Fig. 2 and 4.) AOA, ampulla osseaanterior; AOL, ampulla ossea lateralis; AOP, ampulla ossea posterior;COC, crus osseum commune; COCH, cochlea; COS, crus osseumsimplex; CSA, canalis semicircularis anterior; CSL, canalis semicircularislateralis; CSP, canalis semicircularis posterior; FC, fenestra cochleae(rotunda); FV, fenestra vestibuli (ovale); LSS, lamina spiralis secundaria;VEST, vestibulum.

233HIGH RESOLUTION IMAGING WITH MICROTOMOGRAPHY

but also provides insight into pathophysiological processesin hereditary and non-hereditary inner ear diseases inhumans (Probst et al., 1998; Vahava et al., 1998). Thesemouse models with inner ear defects (pre-existing mu-tants as well as transgenic and knockout mice) are numer-ous, and abnormalities vary from severe otovestibulardysmorphogenesis to developmental or degenerative (sub)cellular abnormalities (Deol, 1980). Anatomical as well asfunctional abnormalities caused by mutations often varyconsiderably across animals and may be asymmetric be-tween the ears in one individual animal. It is thereforenecessary to investigate the complete bony and membra-neous labyrinths of several mice before any conclusionscan be drawn (Bohne and Harding, 1997). Investigationsdefining the morphology of the overall labyrinth are thuscomplementary to other methods demonstrating (sub)cel-lular morphology or function.

Severe dysmorphogenesis can be detected by close in-spection of the temporal bone, for instance, by means oftransillumination. This technique however does not pro-vide sufficient detail and photographs obtained in thisway are difficult to interpret. For animal studies on em-bryologic development paint-filling of the membraneouslabyrinths can be performed (Martin and Swanson, 1994).This method renders the morphological features of thedeveloping labyrinth but fails to preserve the neuro-epi-thelial tissues.

Three-dimensional reconstruction is generally consid-ered to be the most indicated way to examine the completestructure of the inner ear. Minute abnormalities that can-not be detected by two-dimensional investigation alonecan be revealed because functionally relevant spatial re-lationships of the distinct inner ear components can beevaluated more accurately (Isono et al., 1997). Such adetailed 3D examination is of utmost importance in case ofdysmorphogenesis caused by gene defects.

The classical way to investigate the complete labyrinthin the mouse consists of histological serial sectioning afterdecalcification. However, 3D reconstructions based on

classical serial sections are very time-consuming and arehampered by practical difficulties such as loss of or dam-age to the slices. In addition a proper alignment of con-secutive slices is difficult and this serial sectioning obvi-ously limits the possibilities for further ultrastructuraland molecular biological research on the same tissue.

Three-dimensional reconstruction based on imagingtechniques such as CT and MR scanning constitutesa rapid and convenient way to study the morphology of thelabyrinth. Medical imaging is a rapidly developing field inwhich improving techniques and resolutions have contrib-uted to a lot of progress over the last years (Valvassori,1994; Casselman, 1996). Clinical CT and MR imagingwith 3D reconstructions of human inner ears have alreadyresulted in clinical implications (Reisser et al., 1996; Isonoet al., 1997). However, the resolution of clinically used CTand MRI equipment is insufficient to study the inner earstructure of small laboratory animals such as mice. Re-cently, new imaging technologies producing much higherresolutions have become available for laboratory pur-poses. Micro computer tomography (Micro CT; Sasov andVan Dyck, 1998), magnetic resonance microscopy (MRmicroscopy; Henson et al., 1994), micro positron emissiontomography (micro PET; Cherry et al., 1997), and microsingle photon emission computer tomography (microSPECT; Weber et al., 1994) are new tools that allow ob-servation of both functional and morphological effects ofmutations without impeding possibilities for further re-search.

In this study the applicability of such a micro-CT deviceas a rapid non-destructive screening tool to study themorphology of the murine labyrinth was tested. The innerears from SWR/J and epistatic circler mice were scannedat a resolution of 8 mm at a scanning time of 30 min, whichis still markedly less compared to micro-MR-devices (Hen-son et al., 1994).

Comparison of the diameter of both the anterior andposterior semicircular canals between the control SWR/Jmice and the epistatic circler mice revealed no significantdifferences. The diameter of the lateral canal in the con-trol group was comparable to the anterior and posteriorones. The diameter of the lateral semicircular canal of theepistatic circler mice, however, was significantly smaller(P 5 0.002).

Stacks of detailed tomographic images of the inner ear,which were further processed for subsequent 3D recon-structions, disclosed both subtle and marked deformationsat the level of the lateral semicircular canals of the epi-static circler mice. Light microscopical examination of his-tological sections of these scanned samples indicated thatthis tomographic imaging procedure does not cause addi-tional tissue artifacts and inherently does not impedefurther processing for microscopical investigations.

In the future, MR microscopy and micro CT technologywill almost certainly find widespread use for high resolu-tion morphological investigation of all kinds of organs andtissues. Other new imaging technologies such as microPET and micro-SPECT reach much lower resolutions andwill be preferentially used for functional studies. The com-bination of these non-destructive new imaging technolo-gies with other morphological and functional investigationtechniques will probably prove to be a very valuable tool ininner ear research in the near future.

In conclusion, the unprecedented high resolution of thisCT-imaging method provides quick and detailed informa-

Fig. 4. Diagram representing the mean diameters of the semicircularcanals of the control (SWR/J) and epistatic circler mice. Error bars referto standard errors. Statistically significant differences (P 5 0.002) aremarked by an asterisk.

234 VAN SPAENDONCK ET AL.

tion about the morphology of mouse inner ear structureswithout hampering subsequent investigations. This tech-nique allowed detection of subtle abnormalities at thelevel of the semicircular canals of hereditarily impairedmice i.e. the epistatic circler, and appears to be a veryvaluable tool in inner ear research especially in view oftransgenic and knockout mouse models.

ACKNOWLEDGMENTSWe thank Dr. D. Van Dyck and Dr. N. De Clerck for

giving access to the microtomograph; and Dr. W. Decrae-mer for helping with the 3D reconstructions. The technicalassistance provided by D. De Rijck, R. Van Beeck and D.Vindevogel is also greatly acknowledged. K.C. holds apre-doctoral research position (financed by an IWT grant

from the Flemish government) and G.V.C. holds a post-doctoral research position with the FWO.

LITERATURE CITEDBohne BA, Harding GW. 1997. Processing and analyzing the mouse

temporal bone to identify gross, cellular and subcellular pathology.Hear Res 109:34–45.

Casselman JW, Kuhweide R, Ampe W, D’Hont G, Offeciers EF, FaesWK, Pattyn G. 1996. Inner ear malformations in patients withsensorineural hearing loss: Detection with gradient-echo (3DFT-CISS) MRI. Neuroradiology 38:278–286.

Cherry SR, Shao Y. Silverman RW, Chatziioannou A, Meadors K,Siegel S, Boutefnouchet A, Farquhar T, Young J, Jones WF, New-port D, Moyers C, Paulus M, Binkley DM, Nutt R, Phelps ME. 1997.MicroPET: a high resolution PET scanner for imaging small ani-mals. IEEE Trans Nucl Sci 44:1161–1166.

Fig. 5. a–c: Histological sections (haematoxylin eosin). Magnifica-tion: 573. (Same specimen as Fig. 2 and 3c,d.) a: Midmodiolar sectionshowing the cochlea and the vestibule containing the macula utriculi andthe macula sacculi. b: Section through the anterior semicircular canal(CSA) (bony labyrinth) showing a normal diameter (black arrow). Theanterior semicircular duct (DSA; membranous labyrinth) is clearly visible.c: Section showing the restricted aspect of the lateral semicircular canal(CSL; bony labyrinth; open arrow). The lateral semicircular duct (DSL;

membranous labyrinth) is clearly discontinuous. The black arrow showsthe normal outlook of the posterior semicircular canal (CSP) and poste-rior semicircular duct (DSP). AOP, ampulla ossea posterior; CSA, canalissemicircularis anterior; CSL, canalis semicircularis lateralis; CSP, canalissemicircularis posterior; DSA, ductus semicircularis anterior; DSL, duc-tus semicircularis lateralis; DSP, ductus semicircularis posterior; MS,macula sacculi; MU, macula utriculi; OC, organum spirale (Corti); SM,scala media; ST, scala tympani; SV, scala vestibuli; VEST, vestibulum.

235HIGH RESOLUTION IMAGING WITH MICROTOMOGRAPHY

Deol MS. 1980. Genetic malformations of the inner ear in the mouseand in man. Birth Defects Orig Artic Ser 16:243–261.

Doolittle DP. 1963. Two-gene circling in the mouse. Genetics 48:s887.Harada T, Ishii S, Tayama N, Sugasawa M. 1990. Computer-aided

three-dimensional reconstruction of the osseous and membranouslabyrinths. Eur Arch Otorhinolaryngol 247:348–351.

Henson MM, Henson OW, Gewalt SL, Wilson JL, Johnson GA. 1994.Imaging the cochlea by magnetic resonance microscopy. Hear Res75:75–80.

Isono M, Murata K, Aiba K, Miyashita H, Tanaka H, Ishikawa M.1997. Minute findings of inner ear anomalies by three-dimensionalCT scanning. Int J Pediatr Otorhinolaryngol 42:41–53.

Martin P, Swanson GJ. 1993. Descriptive and experimental analysisof the epithelial remodellings that control semicircular canal for-mation in the developing mouse inner ear. Dev Biol 159:549–558.

Probst FJ, Fridell RA, Raphael Y, Saunders TL, Wang A, Liang Y,Morell RJ, Touchman JW, Lyons RH, Noben-Trauth K, FriedmanTB, Camper SA. 1998. Correction of deafness in shaker-2 mice by anunconventional myosin in a BAC transgene. Science 280:1444–1447.

Reisser C, Schubert O, Forsting M, Sartor K. 1996. Anatomy of thetemporal bone: Detailed three-dimensional display based on imagedata from high-resolution helical CT: A preliminary report. Am JOtol 17:473–479.

Sasov A, Van Dyck D. 1998. Desktop X-ray microscopy and microto-mography. J Microsc 191:151–158.

Taylor BA. 1976. Epistatic circler gene of C57L/J. Mouse News Lett55:17.

Vahava O, Morell R, Lynch ED, Weiss S, Kagan ME, Ahituv N,Morrow JE, Lee MK, Skvorak AB, Morton CC, Blumenfeld A, Fry-dman M, Friedman TB, King MC, Avraham KB. 1998. Mutation intranscription factor POU4F3 associated with inherited progressivehearing loss in humans. Science 79:1950–1954.

Valvassori GE. 1994. Update of computed tomography and magneticresonance in otology. Am J Otol 15:203–206.

Weber DA, Ivanovic M, Franceschi D, Strand SE, Erlandsson K,Franceschi M, Atkins HL, Coderre JA, Susskind.H, Button T,Ljunggren K. 1994. Pinhole SPECT: an approach to in vivo highresolution SPECT imaging in small laboratory animals. J Nucl Med35:342–348.

236 VAN SPAENDONCK ET AL.