Nine fresh cadaver heads underwent CT scanning with the Volume Zoom® and Sensation® 64 scanners (Siemens Medical Solutions, Forchheim, Germany) using a custom cochlear implant protocol with the following parameters: 120 KVp, 315 mA, 0.5 mm collimation and 120 KVp, 360 mA, 0.6 collimation respectively. The scans were obtained in spiral mode with a 0.9 pitch and then reconstructed using a U70u kernel, a 51 mm field of view and a slice thickness of 0.1 mm resulting in 0.1 mm isotropic voxels. In the pre-operative condition heads were positioned as they would be clinically, with chins tilted backwards to obtain images in a modified Stenver’s angle. Heads were situated such that the scan plane was parallel to a line that traversed the inferior orbital rim and petrous apex, and were secured in place using surgical tape. CTs were obtain with both scanners in order validate previous work acquired with the Volume Zoom® and support current data acquired with the Sensation 64®. The use of donated cadaver material was exempt from review by our institutional review board.
Bilateral cochlear implant surgeries were performed by two experienced otologists (TEH and RAC). A standard transmastoid facial recess approach under microscopic guidance was employed for all surgeries. While “soft surgeries” were performed in some specimens, intentional trauma was introduced during array insertion in others to produce a variety of outcomes. For example, in some specimens the array was inserted beyond the point of first resistance. The 18 cochleae were implanted in approximate equal numbers with the Advanced Bionics HIfocus® 1J and Cochlear Nucleus® 24 Contour Advance electrode arrays which account for the majority of the devices in our CI patient population. Arrays were fixed in place with cyanoacrylate glue and cut approximately one centimeter outside the cochleostomy. Incision flaps were sutured and heads underwent post-operative CT scanning. During post-operative scanning heads were positioned with chins tilted downward, mimicking the clinical positioning that avoids having the receiver-stimulator in the scan plane. The same scanning protocol used for the pre-operative CT was used for the post-operative CT with the addition of enabling extended data range so that the higher Hounsfield values of the electrode array would not be truncated. Air trapped within the calvarium was flushed with a large-bore saline syringe prior to scanning.
For each ear, the pre and post-operative clinical CT data sets were then analyzed using the 2007 Skinner volume registration method and the resulting composite volume used to create a 3D reconstruction of the implanted cochlea and array (Fig. 1)(17 (link)). The post-operative scan image A in Fig 1 exemplifies the large metallic bloom artifact of the electrode array which must be eroded by intensity thresholding to identify only the higher Hounsfield values at the center of the bloom corresponding to the metallic center of the mass of the lead wires and electrode contacts of the array. The intensity threshold value is adjusted to erode the bloom to a size approximately equal to the array diameter specified by the manufacturer. For arrays in which the electrode contacts are far enough apart, it’s possible to erode the bloom to individual objects such as the red spheres in Fig 2. When individual contacts are too close to be resolved by intensity thresholding alone, a summed voxel projection of the CT volume (Fig 1 Post-operative scan image B) is used in conjunction with manufacturer specifications of contact location to segment the eroded bloom into individual contact objects such as the red discs in Fig 1. To determine the scalar position for each electrode contact of the array, the 3D reconstruction volume was then manually registered and scaled to a best fit to the cochlear atlas and a series of mid-modiolar images of the combined volume were generated (Fig 1 overlay on OPFOS atlas image). The cochlear canal was divided by a line projecting along the osseous spiral lamina of the cochlear atlas and if 75% or greater of the electrode contact object was below or above that line it was assigned a positron of ST or SV respectively. For all other intermediate positions, the contact was assigned an M designation. A single observer (TAH) performed this analysis and was blinded to the histology analysis. Analyze™ software (Robb, 2001 (link)), a Windows-based 3D biomedical image visualization and analysis program, was used to import the 2D image series from each of the imaging modalities used in this study, create 3D volumes reformatted to a mid-modiolar orientation, and perform the image manipulation and analysis(28 (link)).