Hearing Research 75 (19Y4) 7%X0

Imaging the cochlea by magnetic resonance microscopy

M.M. Henson “,*, O.W. Henson, Jr. “, S.L. Gewalt ‘, J.L. Wilson “, G.A. Johnson

‘I LX~t.siott of‘ Otolaryngolo~ /Head and Neck Surgety, CB Number 7090, Taylor Hall, The LJ’IIIIY~.S~~~ of North Ccrrolittu. (‘huprl Ifrll,

NC 27.599, USA, ” Deparftnent of Cell Biology and Anutotny, The Uttiwrsify of’ North Curolina. Chapel Ifill. NC’ 27.599, U.S.-l. ’ Center for In Viro Microscopy, Drpartmntt of Radiology, Drtkr Ckic~ertity Medicul Ccttter, Dttrhum. NC‘ 27710, I:,S.4

(Received 13 September 1993; Revision received 7 December IYW: Accepted 10 December IYYi)

Abstract

The isolated, fixed cochlea of the mustached bat was studied with three dimensional magnetic rcsonancc (MR) microscopy. The cochlea of this animal is about 4 mm in diameter and its entire volume was imaged. With the field of view and matrix size used, the volume elements (voxels) making up the volume data set were isotropic 25 x 25 x 25 pm cubes. Three dimensional (3D) MR microscopy based on isotropic voxels has many advantages over commonly used light microscopy: I) it is non destructive; 2) it is much less time consuming; 3) no dehydration is required and shrinkage is minimized; 4) the data set can hc used to create sections in any desired plane; 5) the proper alignment of sections is inherent in the 3D acquisition so that no reference points are required; 6) the entire data set can be viewed from any point of view in a volume rendered image: 7) the data is digital and features can be enhanced by computer image processing; and 8) the isotropic dimensions of the voxels make the data well-suited for structural reconstructions and measurements. Good images of the osseous spiral lamina. spiral ligament. Scala tympani. Scala vestibuli, and nerve bundles were obtained. The vestibular (Reissncr’s) membrane was easily idcntificd in the mustached bat and it appears to bulge into the Scala vestihuli. The visibility of this structure suggests that MR microscopy would be well-suited for studies of endolymphatic hydrops.

@y words: Cochlea: Magnetic resonance microscopy: MRI

1. Introduction

The mammalian cochlea is a coiled structure with complex spaces enclosed by bone and thin membranes. Full appreciation of the geometry of the spaces is of interest from many points of view, including compara-

tive anatomy, sensory physiology, cochlear modeling, hydrodynamics, electrical field analysis and certain pathological states. The classical technique of obtain- ing images for studying these aspects of cochlear mor- phology is the preparation of histological serial sec- tions. This has many problems: 1) it is a time consum- ing process of weeks to months; 2) sections or pieces of sections are frequently lost; 3) a great deal of shrinkage and distortion occurs during dehydration; 4) uneven slice thickness results in distortion of reconstructed

* Fax: (‘)I’)) Yh6-18%

images; and 5) reliable reference marks are difficult to obtain. MR microscopy is a relatively new technique

(Johnson, 1986; Aguayo, 1986; Eccles, 1986) which is rapidly becoming a powerful tool for many types of studies on living as well as fixed tissues. The technique is distinct from clinical MR imaging in terms of the

technical requirements to clearly image voxels a million times smaller and in terms of the uniqueness of the physical mechanisms which determine contrast in the microscopic image. At the present time, there are only a few centers capable of this type of imaging, hut it is to be expected that more facilities will become avail- able and that the resolution will improve far beyond that which is currently obtainable.

The purpose of this paper is to show the quality of cochlear images that can bc obtained with the tcchnol- ogy now in place and to show the unique insight the tcchniquc provides in understanding the three dimcn- sional morphology of the cochlea.

037x-5’)55/‘)4,‘$07.00 K‘! 1994 Elsevier Science B.V. All rights resewed

SSDI 037~-SYSS(Y~)Et1219-7

2. Methods

The bony cochlea of the mustached bat, Ptcronotus

p. parnellii, consists of a dense otic capsule, approxi- mately 3 mm in diameter; it is loosely attached to the skull and can be easily and quickly removed. For MR imaging, isolated cochleae were immersion fixed in 4% phosphate buffered formalin at room temperature for a minimum of 24 h. The specimen was gently agitated but neither the round nor oval windows were compro- mised. For data collection the specimen was sub- merged in buffered formalin in a thick slide with a well 6 mm in diameter and 5 mm deep, just large enough to accommodate the entire cochlea. The slide was cus- tom-made to fit into a small radiofrequency coil.

These studies were carried out at the Duke Univer- sity Center for In Vz’rlo Microscopy. The MR imaging system used was a GE Omega System with a 7.1 Tesla, 1.5 cm diameter horizontal bore superconducting mag- net, with 85 gauss/cm shielded gradients. The ra- diofrequency coil was a Helmholtz pair specially de- signed for imaging small samples (Banson et al., 1992). The data acquisition parameters for the 3D spin echo acquisition were: 1) total scan time, using 4 averages, 58 h; 2) echo time,TE, 5 ms; 3) repetition time,TR, 800 ms; 4) resolution 256 x 256 x 256; field of view 6.4 mm

Fig. 2. Representative sequential imayea ot evel) tourth WCLIOII through the plane of acquisition. Therefore. differrncea tram image to image represent changes which have occurred over ,t distance 81i 100 pm. Bar = I mm.

Fig. 1. Cochlear slice obtained from the volume data set in the plane of acquisition. AN. auditory nerve; CA, cochlear aqueduct: OSL. osseous spiral lamina: SC, spiral ganglion; SL, spiral ligament: SM. scala media: ST. Scala tympani: VM, vestibular membrane.

Fig. 3. Volume rendered image of the cochlea, a projection of the entire volume data set.

M.M. Henson ct al. /Hearing Research 75 (1044) 7.5-X0 77

isotropic. This yielded cubic voxels 25 pm in each dimension. Volume rendering and arbitrary sectioning of the volume data representing the cochlea were ac- complished with VoxelView-ULTRA software (Vital Images, Fairfield, IA) running on a Silicon Graphics VGX 320 workstation.

The care and use of animals for this study were approved by the IACUC (Institutional Animal Care and Use Committee) of the University of North Car- olina at Chapel Hill, Animal Assurance Number A3410-01.

3. Results

Construction of sections from the rdume data From the 3D cube of MR volume data, a complete

set of serial images of the cochlea was obtained. This particular set represented sectioning through one of the orthogonal planes of the volume. As shown in a labeled slice (Fig. 1) and in sequential representative sections (Fig. 21. good images of many basic cochlear structures were evident in the 25 pm sections. The bone, which has little or no signal, appears black and

Fig. 4. Four different views of the volume rendered cochlea. Sections through any plane can quickly he ohtaincd. stored and u\ed for

reconstructions and/or measurements. Bar = 1 mm.

stands in sharp contrast to the surrounding fluid and soft tissue. The soft tissue within the cochlea gives a strong signal and is brighter than the surrounding fluid. The auditory nerve, spiral ganglion and nerve bundles

are especially prominent in the images shown here. Although the mechanisms responsible for the contrast in these images were not studied in detail, differences in signal intensity probably arise from differences in proton density and local diffusion (Cho. 1988: Callaghan, 1991).

The position and changing size of the spiral liga- ment was also evident; the signal from the ligament was especially high at its junction with the bone. The vestibular (Reissner’s) membrane which is only two cell layers thick is clearly visible and the bulging nature ot this membrane in fteronotus can be seen in Figs. 1 and 2.

Volume rerldering The data representing the entire 3D volume con-

taining the cochlea can be combined to form ;I volume rendered image (Drebin ct al.. 1%4X: Ticde et al.. 1990). Such an image is a two dimensional projection containing information from all three dimensions of the data. The large isotropic 3D array generated in the MR study is particularly well suited for creating this kind of projected image. The VoxelViewULTRA visu- alization program (Argiro, 1990) is based on a com- positing algorithm which sums contributing voxel inten- sities from the volume to generate each point of the

projected image. In addition to summing the intensity of the voxels along the ray contributing to an image point, each intensity level in the data set is also as- signed an opacity value which modulates the summing process. It is possible to interactively alter these opac-

Fig. 5, Examples of sections in different planes. Serial sections can he ohtained in any plane. Bar = 1 mm.

ity settings to affect the qualities of the projection

image or “volume rendering.” By such adjustments,

the fluid surrounding the cochlea is made transparent and the bright external surface of the cochlea is made less opaque. With these settings both the surface and internal structure of the cochlea arc made visible in the volume rendering (Fig. 3). Renderings could be created in any orientation (Fig. 3). In all orientations the complex, changing nature of the cochlear spiral was evident. The spiral ligament and cochlear nerve were especially easy to identify. In Fig. 3 the stapes and even a small sesamoid bone in the tendon of the stapedius muscle (the skeletal element of Paauw) are visible.

To view oblique slices through the cochlea. the volume rendering could be oriented to help identify the desired plane. A section in that plane was ex- tracted from the volume data and consecutive slices could be immediately obtained. The ability to retro-

spectively section in any number of different planes is an advantage of isotropic sampling of the 3D volume. For slices oblique to the axes of the volume. there was minimal reduction in resolution because the volume was sampled by isotropic voxels.

Fig. 5 shows examples of slices through several different oblique planes. The individual structures that were represented in data from the original slice plane are also clearly visible in these images. In all planes the

ST

Fig. 6. Light micrograph of a I pm section through the ha\al turn of

the mustached hat’s cochlea. The cochlea was embedded in Epon.

sectioned and stained with toluidine blue. Structures can he com-

pared with those in Fig. 1. AN, nuditoty nerve fibers (dark area

within the OSL): OSL, osseous spiral lamina; SL. spiral ligament;

SM. scala media; ST. acala tympani: VM. vestibular membrane.

Bar = 100 pm.

vestibular membrane is prominent. In A, B and D the nerves are evident as small. opaque circles or lines in

the osseous spiral lamina; in C the auditory nerve can be seen in the modiolus giving off radial branches toward the organ of Corti. Both the spiral ligament and spiral limbus appear bright in these images. Fig. 6

shows a histological section through one part of the basal turn of the mustached bat’s cochlea; the same

structures labeled in Fig. I can bc identified in this section. The anatomical features seen in MR images are fully supported by numerous LM. TEM and SEM preparations which we have studied in our laboratory.

4. Discussion

MR microscopy provides a rapid. efficient means of obtaining serial sections through the cochlea of mam- mals. The resolution of MRI images is determined by

the field of view and matrix size, as well as more complex physical constraints, e.g., encoding gradient

strength and diffusion (Callaghan. IYYI). With the bat cochlea, which could be imaged in a 6.3 mm cubic field of view with a matrix size of 2% X 36 X 256, the voxels were 25 pm on each side. Imaging a smaller cochlea or use of a larger matrix size could provide greater resolution. This assumes the concomitant in- creases in signal sensitivity, gradient strength. and data handling capabilities which would be required for higher resolution imaging. The theoretical lower limit of voxel size has been estimated at I to IO pm (Cal- laghan and Eccles, lYX7: lY88: Callaghan, IYYO: Mc- Farland. lYY2; Hyslop and Lautcrbur. IYYl: Cho et al.. I%& Ahn and Cho, I%+)) and thus it is expected that the resolution of internal structures may be greatly improved over that which we have demonstrated here (see below ).

This is one of the first studies to show that the small membranes in the cochlea can be imaged with a non- destructive technique. The thickness of the vestibular membrane is not constant along its extent; it is very thin in many places but is enlarged where nuclei or other inclusions are present. It has been reported to be 1.7-5.1 pm wide in other small mammals (Duvall and Rhodes. lY67; Hunter-Duvar, lY78). and mcasurc- ments in Ptrronotus range from 1.S to 5.4 pm (Hen- son. unpublished observations). The clarity of this structure using 25 pm isotropic voxels was unexpected. It is an important structure to be able to visualize in that it separates the Scala media from the scala vcstibuli. It is now possible to dcterminc the dimen- sions of each of these spaces with confidence since it can be done without compromising their integrity.

The 25 pm resolution obtained with our tissue is not significantly better than that obtained by Voie et al. (1’902; lYY3) who have developed a technique of fixing

and clearing the cochlea and staining with a fluores- cent dye. They reconstructed the cochlea on the basis of planar fluorescence obtained with sheets of laser light. Since their procedure included dehydration and clearing of the specimen, a significant amount of dis- tortion and/or shrinkage probably occurred. In our preparation the specimens were fixed only and shrink- age should have been minimal. While there are poten- tial geometric distortions in MR images due to the

undesirable magnetic field gradients resulting from magnetic susceptibility differences in the object being imaged, these differences have been minimized by im-

mersing the cochlea in solution, and the remaining

effects minimized by using an acquisition sequence that is radiofrequency refocused (Jara et al.. 1993).

The images obtained in this study required the substantial acquisition time of 58 h. The long acquisi- tion time arose principally from the need to collect the complete data representing the 3D volume and from the need to signal average to gain sensitivity to the signal from the small voxels. Some of our more recent

data with the guinea pig cochlea have already demon- strated that good images can be obtained in 15 h (unpublished observations) using a fast spin echo tech- nique designed specifically for MR microscopy (Zhou et al., 1993). The recent demonstration of high temper- ature superconducting radiofrequency coils promises an improvement in sensitivity of 50 times (Black et al.. 1993). Such sensitivity reduces the acquisition time necessary to attain sufficient signal by an even greater factor. Therefore, it is reasonable to expect acquisition times of 1 to 3 h in the future, the time then being determined by physical contrast mechanisms and no

longer by sensitivity. One of the real advantages of MR microscopy is the

ability to obtain slices in a variety of planes and to perform three dimensional reconstruction of specific structures within the cochlea. Details concerning re- constructions of the cochlea of the bat and other species will be the subject of a subsequent paper.

There is a great deal of information to be obtained with three dimensional MR microscopy as applied to cochlear morphology and pathology. Specifically, the ability to view the vestibular membrane opens new possibilities for the study of endolymphatic hydrops. Rapid progress in MR microscopy and animal support and monitoring should lead to methods of imaging the living cochlea in an intact animal in the near future.

Acknowledgments

We would like to thank Gary Cofer for sharing imaging expertise and Art Keating, Ding-hua Xie and Elaine Fitzsimons for their help with illustrations and critical reading of the manuscript. This work was sup-

ported by NIH grants NlDCD DC 00113 (OWH) and

IP41RR05959 (SLG & GAJ).

References

Ahn, C.B. and Cho, Z.H. (1989) A generalized formulation of

diffusion effects in Frn resolution nuclear magnetic resonance

imaging. Med. Phys. 16, 22-28.

Aguayo, J.B., Blackband, S.J., Schoeniger, J., Mattingly, M.A. and

Hintermann, M. (1986) Nuclear magnetic resonance imaging of a

single cell. Nature, 322, 190.

Argiro, V.J. t 1990) Seeing in volume. Pixel 1, 35-30.

Banson, M.L., Cofer, G.P.. Black, R. and Johnson, G.A. (1992) A

probe for specimen magnetic resonance microscopy. Investigative

Radiology 27, 157- 164.

Black, R.D., Early, T.A., Roemei-. P.B., Mueller, O.M., Morgo-

Campero, A., Turner, L.C. and Johnson, G.A. (1993) A high

temperature superconducting receiver for NMR microscopy. Sci-

ence 259, 703-795.

Callaghan. P.T. (1990) Susceptibility-limited resolution in nuclear

magnetic resonance microscopy. J. Magn. Reson. 87, 304-318.

Callaghan, P.T. (1991) Principles of Nuclear Magnetic Resonance

Microscopy, Oxford, Oxford University Press.

Callaghan, P.T. and Eccles, C.D. (1987) Sensitivity and resolution in

NMR imaging. J. Magn. Reson. 71, 426-445.

Callaghan, P.T. and Eccles, C.D. (1988) Diffusion-limited resolution

in nuclear magnetic resonance microscopy. J. Magn. Reson. 78,

1-8.

Cho, Z.H., Ahn, C.B., Juh, SC., Lee, H.K., Jacobs, R.E., Lee, S.. YI,

J.H. and Jo, J.M. (1988) NMR Microscopy with 4 mm resolution _ theoretical study and experimental results. Med. Phys. 15,

X15-823.

Drebin, R.A., Carpenter, L. and Hanrahan, P. (1988) Volume ren-

dering. ACM Computer Graphics 22, 6.5-74.

Duvall, A.J. and Rhodes, V.T. (1967) Reissner’s membrane: An

ultrastructural study. Arch. Otolaryngol. 86, 143-151.

Eccles, C.D. and Callaghan, P.T. (1986) High resolution imaging: the

NMR microscope. J. Magn. Reson. 68, 393-398.

Hunter-Duvar, I.M. (1978) Electron microscopic assessment of the

cochlea: Some techniques and results. Acta Oto-Laryngol. S351.

24-32.

Hyslop, W.B. and Lauterbur, P.C. (1991) Effects of restricted diffu-

sion on microscopic NMR imaging. J. Magn. Reson. 94, 501-510.

Jara, H., Wehrli, F.W., Chung, H. and Ford, J.C. (1993) High

resolution variable angle 3D MR imaging of trabecular mi-

crostructure in vivo. Magnetic Resonance in Med. 29, 528-539.

Johnson. G.A., Thompson, M.B., Gewalt, S.L. and Hayes, C.E.

(1986) Nuclear magnetic resonance imaging at microscopic reso-

lution. J. Magn. Reson. 68, 129-137.

McFarland. E. (1992) Time independent point-spread function for

MR microscopy. Magn. Reson. Imaging 10, 269-278.

Tiede, U., Hoehne, K.H., Bomans, M., Pommert, A. and Riemer, M.

(1990) Investigation of medical 3D rendering algorithms. IEEE

Comput. Graph. Appl. 10, 41-53.

Voie, A.H., Spelman, F.A., Burns, D.H. and Sutton E. (1992) Fluo-

rescence optical sectioning: An imaging technique for the study

and analysis of the inner ear. Abstr. Assoc. Res. Otolaryngol. 288,

p. 100.

Voie. A.H., Burns, D.H. and Spelman, F.A. (1993) Orthogonal-

planefluorescence optical sectioning: Three-dimensional imaging of macroscopic biological specimens. J. Microscopy 170, 229-236.

Zhou, X., Cofer, G.P., Suddarth, S.A. and Johnson, G.A. (1993)

High field MR microscopy using fast spin echoes. Magn. Reson. Med. 30, 60-67.