Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online before print September 24, 2001, 10.1148/radiol.2212010275
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2212010275v1
221/2/543    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Whiting, B. R.
Right arrow Articles by Skinner, M. W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Whiting, B. R.
Right arrow Articles by Skinner, M. W.
(Radiology. 2001;221:543-549.)
© RSNA, 2001

Technical Developments

Cochlear Implants: Three-dimensional Localization by Means of Coregistration of CT and Conventional Radiographs1

Bruce R. Whiting, PhD, Kyongtae T. Bae, MD, PhD and Margaret W. Skinner, PhD

1 From the Mallinckrodt Institute of Radiology (B.R.W., K.T.B.) and Department of Otolaryngology–Head and Neck Surgery (M.W.S.), Washington University School of Medicine, 510 S Kingshighway Blvd, St Louis, MO 63110. Received January 10, 2001; revision requested March 5; revision received April 3; accepted May 1. Supported in part by the Mallinckrodt Institute of Radiology and Electronic Radiology Lab, National Institute on Deafness and Other Communication Disorders (grant R01-DC00581), and the Whitaker Foundation (Biomedical Engineering Program). Address correspondence to B.R.W. (e-mail: whitingb@mir.wustl.edu).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 Materials and Methods
Results
 Discussion
 APPENDIX
 REFERENCES
 
With use of radiopaque implanted objects as internal fiducial markers, the authors developed and evaluated a technique for coregistering computed tomographic (CT) and computed radiographic images to help determine three-dimensional location information for implant electrodes in the cochlea in phantoms and patients. Three-dimensional positional data from CT were assigned on a radiograph, which permitted identification of individual cochlear electrode locations that were not depicted at CT.

Index terms: Computed tomography (CT), image processing, 21.12115, 21.12117, 21.1215 • Computed tomography (CT), technology, 21.12115, 21.12117, 21.1215 • Computed tomography (CT), three-dimensional, 21.12117 • Ear, CT, 21.12115, 21.12117, 21.1215 • Ear, labyrinth, 21.12115, 21.12117, 21.1215 • Phantoms • Radiography, digital, 21.1215


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
Materials and Methods
Results
 Discussion
 APPENDIX
 REFERENCES
 
Many patients with severe or profound hearing impairment who use cochlear implants experience dramatically improved hearing. Before implantation, the average recipient can understand only a few words through hearing aids. With the implant, however, the recipient can understand 73%–85% of words in sentences on the basis of sound alone (1–3). Despite this remarkable success, there is substantial potential for increased patient benefit by reduction of the large variation in speech recognition among individuals through development of more effective ways to stimulate an individual’s residual auditory neurons (4,5). To accomplish this goal, it is essential to gain a greater understanding of the fundamental mechanisms of electrical stimulation that produce hearing for these recipients on the basis of knowledge of the position of each implanted electrode relative to auditory anatomy (6,7). Radiologic imaging could play a prominent role in this effort.

Either computed tomography (CT) or computed radiography (CR) can be used to determine the position of cochlear implants in an individual’s ear, but neither is sufficient to locate implants to submillimeter accuracy in three-dimensional (3D) space. In fact, the ideal imaging technique would have the CT attributes of 3D location information and sensitivity to low-contrast objects along with the high spatial resolution of CR. The purpose of our study was to evaluate a method that coregisters CT and CR images with use of commonly available diagnostic equipment.


   Materials and Methods
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results
 Discussion
 APPENDIX
 REFERENCES
 
Algorithm
The problem we sought to solve is stated as follows: Given a 3D image (CT volume) of an object and a perspectively projected two-dimensional (2D) transmission mapping of the same object (CR image), find the mathematic transformation that relates the points on the two images. The difficulty is that there is no knowledge of a specific point on one image that corresponds to a specific point on the other image; only the physical object is the same on each image. If a few known corresponding points could be uniquely identified on both images, the problem could be trivially solved. The concept of our method was to use the high-contrast opaque implant objects as an ensemble of points that allow the determination of a perspective projection transformation that links CT and CR image sets. We used an iterative scheme to compute transformation parameters by minimizing the ensemble distance between a trial projection of the CT data onto the CR image. The mathematic basis of this algorithm is explained in the Appendix. The required input data consisted of the coordinates of points corresponding to metal objects on CT and CR images. Routines were written with software (MATLAB; Mathworks, Natick, Mass) to perform various image-processing steps to prepare these data.

For CT data, we found that a threshold value of 2,500 HU reliably selected voxels associated with metal objects. An automatic search for all corresponding voxels was made to form a list of the x-y coordinates and section number for metal points. This list was manually edited to remove points associated with the interior of the receiver-stimulator package and to identify an initial estimate of the position of the electrode tip and the entry point of the lead wires into this package. The accuracy requirements for these estimates were very modest; varying the inputs by several millimeters (tens of voxels) resulted in the same final parameter values. Typical numbers of metal voxel points used were 50,000 points for 0.1-mm voxels. Because the CT gantry was tilted for most patients by some angle, the position indexes (i,j,k) corresponded to a nonorthogonal coordinate system and had to be converted into x-y-z coordinates in real space with an appropriate trigonometric transformation as follows:

where {px, py, px} are the voxel sizes, t is the gantry tilt angle, and {x,y,z} are the physical coordinates in a CT frame of reference.

For the CR data (Fig 1), metal electrode shadows corresponded to substantial negative modulation over a 5 x 5-pixel (0.5-mm2) area on a very nonuniform background (due to anatomic structures), which precluded simple thresholding. Therefore, the image was processed by applying a high pass filter of variable kernel sizes and manually identifying the resulting matched-filter points associated with electrodes (Fig 2). Again, the transformation calculation was not sensitive to the selection of individual points. Typically, a list of 5,000 pixels was generated.



View larger version (109K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. CR image of implanted electrode array. High-contrast metal objects, including implant electrodes, can be easily detected, but 3D relationships of objects cannot be determined. Solid arrows point to unique positions of the tip of the electrode (lower arrow) and the entry point of lead wires to the stimulator receiver (upper arrow). Also note the platinum spheres (dotted arrows); these were part of an unsuccessful attempt to use external fiducial markers to determine perspective transformation parameters.

 


View larger version (55K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. Cochlear implant points extracted from a CR image. Individual electrodes (arrows) are clearly depicted. Approximately 5,000 pixels are labeled in this binary representation.

 
The data were entered into our analysis program, and transformation parameters were determined. A typical fitted superposition of the CT and CR data is shown in Figure 3. Note that the CT data encompass two to three times the area of the CR data but are generally symmetrically centered over the CR points. Once a 3D-to-2D transformation was determined, CT anatomic structures could be projected onto radiographic images (Fig 3). To determine anatomic features in the patients with the implants, we used a public domain program (NIH IMAGE; National Institutes of Health, available at ) to view the CT images and select a list of voxel indexes that correspond to important landmarks. Radiographs were prepared with anatomy highlighted for use in clinical planning.



View larger version (59K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3. Superimposition of CT and CR images of the cochlear implant by means of coregistration. CT representation (gray) of a cochlear implant was projected by means of transformation onto CR representation (white). Since the scanned CT volume did not include all of the lead wires, CT representation was absent in the vicinity of the loop (right arrow). Left arrow denotes the entry point in Figure 2 projected onto the 2D radiograph, indicating that 22 electrodes and one supporting ring have been inserted into the cochlea.

 
Phantom Construction
To validate the accuracy of our technique, a phantom was constructed to simulate cochlear implant objects with precisely known locations. Two polymethyl methacrylate (acrylic) matched hemispheres with diameters of 12.5 mm were created. On the plane surface of one of them, a 5-mm-diameter circular groove, 1 mm wide and 1 mm deep, was milled (total outside diameter, 6 mm). A straight groove, 1 mm wide and 1 mm deep, was milled tangent to the edge of the circle and extending 6 mm to the edge of the hemisphere. An implant array (Nucleus-22; Cochlear, Lane Cove, New South Wales, Australia) that did not meet manufacturing tolerances was cemented into the groove, filling a 270° turn on the circle and the tangent straight section, leaving a 30-mm loose section of electrodes and lead wires. In the other hemisphere, a 1-mm-diameter hole was drilled through its "north pole." An acrylic 25-mm-diameter split cylinder holder was formed with hemispheric cavities and a recessed slot, into which the electrode-containing sphere and wires were placed. This arrangement allowed an electrode array with precisely known geometry to be placed in an arbitrary orientation for scanning. CT and CR images of the phantom were generated under conditions close to those used in clinical patients.

Patient Data Acquisition
Data sets of CT and CR images were collected in 12 patients (six men and six women; age range, 30–76 years; mean age, 57.8 years) following cochlear implantation surgery (8). As part of the clinical care, an electrode array approved by the U.S. Food and Drug Administration for clinical use (Nucleus 22) was implanted in each patient. The institutional human studies committee approved the participation of the patients in the present research study, and informed consent was obtained. Both data sets were acquired within 2 hours on the same day, with the CR examination followed by the CT examination. The equipment we used is typical of that available in general hospitals.

Imaging Protocols
CT scans were obtained with helical CT scanners (Somatom Plus S or Somatom Plus 4; Siemens Medical Systems, Iselin, NJ) with use of a high-spatial-resolution head protocol (120 kVp, 215 mA, 1-mm collimation, 1 mm/sec table speed, 1 second per rotation) with submillimeter reconstruction (0.1-mm isotropic voxels, extended Hounsfield scale). The scanner gantry was tilted between -7° and +15° to align image sections with the modiolar axis of the cochlea in each patient. Image sections were reconstructed; images and raw projection data were saved on an optical disk and were transmitted over the network to research workstations.

Digital radiographic images were collected by using a protocol defined by Cohen et al (9). A head position similar to that in a modified Stenver view was used; that is, the head was positioned such that the central ray traveled parallel to the modiolar axis of the cochlea. An x-ray neurologic unit (Neurostand; Siemens Medical Systems) was used, with a 915-mm focal spot–to-detector distance, 0.5-mm focal spot, Bucky tray with a moving 9:1 grid, and table-to-cassette distance of 1.75 inches (4.4 cm). Set-up images were obtained with either screen-film film (Lanex Regular, T-MAT G; Eastman Kodak, Rochester, NY) or computed radiographic film (Ektascan storage phosphor reader model 400 with an 18 x 24-cm storage phosphor high-resolution cassette; Eastman Kodak). When the patient was positioned correctly, the electrode array appeared as a nominal circular pattern, and each electrode was distinctly separate from the other electrodes. The exposure parameters for data collection with the storage phosphor plate were 81 kVp and 80 mAs. CR images were used as final radiographs for analysis.


   Results
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results
Discussion
 APPENDIX
 REFERENCES
 
For the phantom object, the average distance of a projected CT point to the nearest CR point was 0.180 mm, with an SD of 0.176 mm. The point where the electrode array entered the sphere object was estimated on the CT image sections. This point was projected onto the CR image, and the number of electrodes contained in the sphere was counted: 22 electrodes plus one supporting ring inside and nine supporting rings outside the sphere. The center of the circular array, as marked by the base of the hole drilled in the upper hemisphere, was estimated on CT images. The projection of this point onto the CR image was then compared with the center of an ellipse that was fitted to electrode points on the CR image, which resulted in an estimated separation error of 0.75 mm.

For the 12 patients, the mean of all individual average distances from projected CT points to the nearest CR point was 0.184 mm, with an SD of 0.044 mm. Anatomic landmarks (eg, modiolar axis, apical turn) on the CT image sets were projected onto the CR image. The images were then analyzed for the implant insertion length, number of electrodes in the cochlea, and presence of kinks or compressions of the array. This information was valuable for programming individual patients’ speech processors to optimize speech recognition. For example, the assignment of frequency boundaries of incoming sound to electrodes needs to be an appropriate compromise between matching the estimated characteristic frequency of neurons near each electrode and the number of electrodes that provide different pitch percepts to deliver acoustic cues in each frequency range (eg, first and second formant speech energy) (5,10,11). For less than full insertion of the electrodes into the inner ear, it is important to know postoperatively how many electrodes are inside the inner ear so that only those will be stimulated. Because there is a modest correlation between word recognition and the 3D functional insertion depth of the electrode array as a percentage of the total cochlear length (5), knowledge of this depth will provide insight into a patient’s speech recognition ability as a basis for auditory training. Electrodes that are kinked cause an unpleasant sound when stimulated, and they need to be removed from the speech processor program. This information is valuable for programming in children.


   Discussion
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results
 Discussion
APPENDIX
 REFERENCES
 
The 3D position of an electrode array within the anatomy of the inner ear presents a profound challenge for existing radiology technologies in terms of spatial resolution and localization of 3D structures. The size of relevant anatomic structures and electrodes involved with cochlear imaging is 1 mm or less (12,13). For example, the cochlear canal is a tube that includes the sensory and supporting cells of the organ of Corti and the fluid-containing spaces of the scala tympani, scala vestibuli, and scala media. The scala tympani, into which the electrode array is placed, is approximately 1 mm in diameter (12,13). The diameters of the basal, middle, and apical turns are approximately 8.0, 4.5, and 1.5 mm, respectively, and the canal spirals are at an inclination of approximately 1 mm per turn (14). Implanted electrode arrays are x-ray–opaque metals (ie, platinum), which consist of rings, balls, or plates of nominal 0.5-mm diameter, to each of which are connected 0.01-mm-thick lead wires (15–17).

With spatial resolution of four line pairs per millimeter, CR can readily depict these high-contrast metal objects, but overlying shadows often obscure low-contrast anatomy and structures, with the result that 3D relationships of objects cannot be detected on radiographs. On the other hand, CT provides excellent information about the 3D location of low-contrast objects but, unfortunately, with limited spatial detail (Fig 4). The spatial resolution at CT is governed by the acquisition apertures: minimum longitudinal collimation is usually set to 1 mm, while in-plane detector arrays have an effective aperture of typically greater than 0.5 mm (eg, 768 detectors subtending a field of view of 500 mm gives an effective aperture of 0.68 mm at the isocenter of our scanner). With standard filtered back-projection reconstruction, the cochlear implant electrode signals result in images blurred 0.5 mm or more, with significant blooming or spread of the metal objects into surrounding imagery. While the centroids of symmetric objects can be determined with accuracies that approach 0.3 mm, attenuation numbers (Hounsfield units) are severely distorted in the vicinity of electrode metal for more than 1 mm. (In fact, individual electrodes are not depicted at CT.)



View larger version (88K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4a. CT volumetric representation of cochlear implant with use of (a) transverse multiplanar reformat and (b) shaded surface display images. In a, arrow points to entry of electrode into cochlea (location provided by Darlene Ketten, PhD). In b, note that individual electrodes are not depicted on CT images, but they are depicted on CR images.

 


View larger version (48K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4b. CT volumetric representation of cochlear implant with use of (a) transverse multiplanar reformat and (b) shaded surface display images. In a, arrow points to entry of electrode into cochlea (location provided by Darlene Ketten, PhD). In b, note that individual electrodes are not depicted on CT images, but they are depicted on CR images.

 
Despite these imaging limitations, several investigators (9,14,16,18–20) report radiologic techniques for imaging cochlear implants. Other investigators (9,18,19) used radiographs acquired with a head position similar to that of the modified Stenver view. Their goal was to image the individual’s electrode array so that it was nominally perpendicular to the modiolar axis of the cochlear spiral, and the landmarks of the superior semicircular canal and the vestibule of the vestibular part of the inner ear are well defined. A template spiral was devised to characterize the position of the electrode array in relation to the modiolar axis, the inner and outer walls of the cochlear canal, and the round window. With this technique, an insertion depth (measured in degrees of rotation) was estimated. Ketten et al (14) and Skinner et al (8) used CT image volumes to determine the total length of the cochlear canal, the 3D functional depth at which the most apical electrode is inserted into the cochlea (measured in millimeters and percentage of total cochlear length), and the 3D trajectory of the electrode array within the cochlear canal. Wang et al (20) developed an unwrapping procedure based on CT findings for estimating the depth of array insertion by iteratively determining the center of the array from its insertion point into the cochlea to the tip of the electrode. All these methods depend on a high degree of subjective manipulation by operators, however, and they do not provide 3D localization of cochlear implant electrodes.

Initially, we unsuccessfully attempted to determine the perspective transformation linking x-ray and CT data coordinate systems by using external fiducial markers attached to the patients’ skin. The markers consisted of six 0.5-mm-diameter platinum spheres, mounted several millimeters apart on a thin plastic base, that were positioned on the skin over a patient’s zygomatic arch on the side of the ear with the cochlear implant. The markers remained attached during the two postoperative imaging examinations. Patients were at different positions for the examinations, however, being upright for the x-ray examination and supine for the CT data acquisition. Because the skin is somewhat mobile and does not provide a solid anchor point for physical attachment, the fiducial markers moved slightly. Originally, we planned to uniquely identify six reference points on each image and determine the parameters of the perspective transformation, by using the 12 equations for the measured radiograph coordinates to solve for nine unknowns. Because of the poor conditioning of the inversion problem in the perspective geometry, small imprecision in coordinate values yielded large errors in the transformation estimate, which precluded robust transformation determination. For the method to work, it was clear that reference fiducials must be attached in an immovable fashion, extend over several centimeters to provide a good "leverage arm" for the location determination equations, and consist of multiple points to achieve better statistical averaging. Our key insight was that the implant array itself was well suited as a fiducial marker system.

The electrode array and its receiver-stimulator package possess enough asymmetry and spatial extension to allow unambiguous transformations. The electrodes and receiver hardware are surgically implanted in a patient’s skull and usually remain immobile thereafter. The titanium package that surrounds the receiver-stimulator electronics is implanted above and behind the middle and inner ear at a distance of several centimeters. Lead wires extend from this package through the middle ear and travel inside the electrodes in the array, which extends as much as 25 mm inside the cochlea. In the implant array, for example, the centers of the 22 electrodes and 10 supporting rings are manufactured to be 0.75 mm apart, and the platinum-iridium wires are encircled by the ring electrodes. Given the 0.30-mm electrode length, the distance between electrode edges is 0.45 mm, and the total length of the array is roughly 25 mm (15,16). Thus, a fixed set of extended reference objects is present on the images for each modality.

With use of implants as internal markers, our method combines the high spatial resolution of radiographic data with 3D structures from CT data to help locate electrode arrays in the cochlea. Anatomic features can be selected in CT data and projected onto radiographic images that are familiar to radiologists. Our method potentially can be used to provide positional information to model the electrical characteristics, to determine the 3D orientation of an individual’s inner ear with external landmarks, and to facilitate a correlative evaluation between patient performance and electrode position. In addition to the cochlear implant application, our technique might be extended to other medical applications that require the coregistration of 3D and 2D imaging data, such as CT angiography, oral maxillary surgery, conformal radiation therapy, and imaging-guided surgical navigation.



View larger version (20K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5. Perspective transformation between 3D CT and 2D CR coordinate systems. In the x-ray frame of reference, the x-ray focal spot projects a point P {xCT, yCT, zCT} onto the detector plane point x (xCR, yCR). The position P in the CT coordinate system can be found from the translation vector {x0CT, y0CT, z0CT} and the rotation required to align the CT axes with the x-ray axes.

 



View larger version (124K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6. Cost function for registration determination. The distance L between the projected 3D point P and the nearest metal point on the radiograph was used as an error metric that was minimized as a function of perspective projection parameters.

 



View larger version (31K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 7. Search pattern for determining the shortest distance between a projected CT data point and a CR data point. A spiral search pattern was used to find the closest CR metal pixel to the projected CT pixel. The projected CT point resides in pixel 1; cells are checked for the presence of a CR metal point in the order listed. Only the first 25 locations are shown. When a match is found, the distance is accumulated in the total error.

 


   APPENDIX
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results
 Discussion
 APPENDIX
REFERENCES
 
To describe our algorithm, we first define the projective transformations involved and then explain the iterative method that was used.

Perspective Transformations
To coregister CT and CR images, a mathematic function is needed that will relate the 3D coordinates of a point in the CT volume to the coordinates of the same point on a CR image. Inverse mapping is also possible (ie, construction of a perspective ray, or line) in the 3D CT space that connects a point on the CR image to the x-ray focal spot. A radiographic image is formed from a 3D object by means of a perspective projection that is characterized by nine parameters: three coordinates {x0CT, y0CT, z0CT} for the center of the CT coordinate system relative to the x-ray coordinate system, three coordinates {xFS, yFS, zFS} of the x-ray focal spot in the x-ray coordinate systems, and three Euler angles {{phi}, {theta}, {psi}} (21) that rotate the CT coordinate axes into alignment with the x-ray coordinate axes (Fig 5). Operationally, CT coordinates must first be transformed into the physical 3D coordinate system of the x-ray source or detector through a series of rotations and translations, expressed as the following:



Then, the points are perspectively projected onto the 2D detector plane by the equations:


The effects of these nine parameters on image appearance are strongly interrelated (eg, moving the x position of the central x-ray beam is, to first order, the equivalent of rotating the CT data points about the z axis, and only small second-order differences in magnification result from the two operations). Indeed, it is not feasible to manually vary parameters and coregister images with even an approximate optimization for the parameter values. Therefore, an automated process for coregistration was sought that quantitatively determined a goodness of fit.

Algorithm for Determining Transformation Parameters
If the coordinate values of four nonplanar noncolinear corresponding points on CT and CR images were known exactly, Equations (A1)–(A5) could be solved for the nine parameters needed. Unfortunately, there are limited unique corresponding points that can be identified between the images. The closest candidates are the tip of the electrode and the junction of the lead wires with the receiver-stimulator package (Fig 1). To overcome this, an algorithm was devised that selected projection parameters based on minimization of the ensemble distance of all projected CT electrode points to the nearest electrode point on the CR image, without requiring a specific one-to-one correspondence between any individual points.

Specifically, the cost function or error metric that we seek to minimize is the sum over all CT points of the minimum distance between the projection of each CT data point on the CR image to the nearest metal point on the CR image (Fig 6). This minimum distance was determined by using a computer spiral search pattern (Fig 7), checking neighboring CR pixels for the presence of a metal point, and terminating the search when one point was located. This approach has the characteristic that as the alignment becomes better and search distances become smaller, the convergence rate speeds up. A MATLAB function fminsearch, which incorporates a Nelder-Mead unconstrained nonlinear gradient-search algorithm to minimize the cost function (22), was used to determine the best-fit condition. It was found that convergence of the search routine depended on initial values of parameters; in fact, the algorithm could converge to an incorrect local minimum under some conditions. Therefore, one of the authors (B.R.W.) selected initial parameters to visually place the electrode tip and lead-to-receiver connection point in an approximate alignment between the CR and projected CT data. Convergence typically took less than 5 minutes. After convergence to a transformation set, typical distances to the nearest CR metal point were less than 2 pixels (0.2 mm) per voxel.


   ACKNOWLEDGMENTS
 
The authors gratefully acknowledge discussions on image processing with Barry S. Brunsden, BS, and his skillful rendering of 3D images, including Figure 4b. Darlene Ketten, PhD, provided expert determination of anatomic features in patient image sets and helpful comments on the manuscript for this article.


   FOOTNOTES
 
Abbreviations: CR = computed radiography, 3D = three-dimensional, 2D = two-dimensional

Author contributions: Guarantors of integrity of entire study, B.R.W., K.T.B., M.W.S.; study concepts and design, B.R.W., K.T.B., M.W.S.; literature research, B.R.W., K.T.B., M.W.S.; clinical studies, B.R.W., K.T.B., M.W.S.; experimental studies, B.R.W., K.T.B., M.W.S.; data acquisition, data analysis/interpretation, B.R.W., K.T.B., M.W.S.; statistical analysis, B.R.W., K.T.B., M.W.S.; manuscript preparation, manuscript definition of intellectual content, manuscript editing, manuscript revision/review, manuscript final version approval, B.R.W., K.T.B., M.W.S..


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results
 Discussion
 APPENDIX
 REFERENCES
 

  1. Helms J, Muller J, Schon F, et al. Evaluation of performance with the COMBI40 cochlear implant in adults: a multicentric clinical study. ORL J Otorhinolaryngol Relat Spec 1997; 59:23-35.
  2. Osberger MJ, Fisher L. SAS-CIS preference study in postlingually deafened adults implanted with the CLARION cochlear implant. Ann Otol Rhinol Laryngol Suppl 1999; 177:74-79.
  3. Skinner MW, Arndt PL, Staller SJ. Nucleus 24 advanced encoder conversion study: performance versus preference. Ear Hear; (in press).
  4. Pfingst BE, Zwolan TA, Holloway LA. Effects of stimulus configuration on psychophysical operating levels and on speech recognition with cochlear implants. Hear Res 1997; 112:247-260.
  5. Fu QJ, Shannon RV. Effects of electrode configuration and frequency allocation on vowel recognition with the Nucleus-22 cochlear implant. Ear Hear 1999; 20:332-344.
  6. Pfingst BE. Auditory prostheses. In: Chapin JK, Moxon KA, eds. Neural prostheses for restoration of sensory and motor function. Boca Raton, Fla: CRC, 2000; 3-43.
  7. Cohen LT, Busby PA, Clark GM. Cochlear implant place psychophysics. II. Comparison of forward masking and pitch estimation data. Audiol Neurootol 1996; 1:278-292.
  8. Skinner MW, Ketten DR, Holden LK, et al. CT derived estimation of cochlear morphology and electrode array position in relation to word recognition in Nucleus-22 recipients. Ann Otol Rhinol Laryngol; (in press).
  9. Cohen LT, Xu J, Xu SA, Clark GM. Improved and simplified methods for specifying positions of the electrode bands of a cochlear implant array. Am J Otol 1996; 17:859-865.
  10. Skinner MW, Holden LK, Holden TA. Effect of frequency boundary assignment on speech recognition with the speak speech-coding strategy. Ann Otol Rhinol Laryngol Suppl 1995; 166:307-311.
  11. Skinner MW, Fourakis MS, Holden TA, Holden LK, Demorest ME. Identification of speech by cochlear implant recipients with the Multipeak (MPEAK) and Spectral Peak (SPEAK) speech coding strategies. I. Vowels. Ear Hear 1996; 17:182-197.
  12. Zrunek M, Lischka M, Hochmair-Desoyer I, Burian K. Dimensions of the scala tympani in relation to the diameters of multichannel electrodes. Arch Otorhinolaryngol 1980; 229:159-165.
  13. Hatsushika S, Shepherd RK, Tong YC, Clark GM, Funasaka S. Dimensions of the scala tympani in the human and cat with reference to cochlear implants. Ann Otol Rhinol Laryngol 1990; 99:871-876.
  14. Ketten DR, Skinner MW, Wang G, et al. In vivo measures of cochlear length and insertion depth of Nucleus cochlear implant electrode arrays. Ann Otol Rhinol Laryngol Suppl 1998; 175:1-16.
  15. Clark GM, Blamey PJ, Brown AM, et al. The University of Melbourne—nucleus multi-electrode cochlear implant. Adv Otorhinolaryngol 1987; 38:1-181.
  16. Skinner MW, Ketten DR, Vannier MW, et al. Determination of the position of Nucleus cochlear implant electrodes in the inner ear. Am J Otol 1994; 15:644-651.
  17. Bredberg G, Lindstrom B, Lopponen H, et al. Electrodes for ossified cochleas. Am J Otol 1997; 18(suppl 6):S42-S43.
  18. Marsh MA, Xu J, Blamey PJ, et al. Radiologic evaluation of multichannel intracochlear implant insertion depth. Am J Otol 1993; 14:386-391[Erratum: Am J Otol 1993; 14:627.].
  19. Xu J, Xu SA, Cohen LT, Clark GM. Cochlear view: postoperative radiography for cochlear implantation. Am J Otol 2000; 21:49-56.
  20. Wang G, Vannier MW, Skinner MW, et al. Unwrapping cochlear implants by spiral CT. IEEE Trans Biomed Eng 1996; 43:891-900.
  21. Zwillinger D, ed. CRC standard mathematical tables and formulae 30th ed. Boca Raton, Fla: CRC, 1996; 300.
  22. Press W, Teukolsky S, Vetterling W, Flannery B. Numerical recipes in C 2nd ed. New York, NY: Cambridge University Press, 1992; 408-412.



This article has been cited by other articles:


Home page
 Am. J. Neuroradiol.Home page
B.M. Verbist, R.M.S. Joemai, W.M. Teeuwisse, W.J.H. Veldkamp, J. Geleijns, and J.H.M. Frijns
Evaluation of 4 Multisection CT Systems in Postoperative Imaging of a Cochlear Implant: A Human Cadaver and Phantom Study
AJNR Am. J. Neuroradiol., August 1, 2008; 29(7): 1382 - 1388.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Neuroradiol.Home page
B. M. Verbist, J. H. M. Frijns, J. Geleijns, and M. A. van Buchem
Multisection CT as a Valuable Tool in the Postoperative Assessment of Cochlear Implant Patients
AJNR Am. J. Neuroradiol., February 1, 2005; 26(2): 424 - 429.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Neuroradiol.Home page
M. A. Mecca, W. Wagle, A. Lupinetti, and S. Parnes
Complication of Cochlear Implantation Surgery
AJNR Am. J. Neuroradiol., November 1, 2003; 24(10): 2089 - 2091.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2212010275v1
221/2/543    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Whiting, B. R.
Right arrow Articles by Skinner, M. W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Whiting, B. R.
Right arrow Articles by Skinner, M. W.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE