In-vitro microCT validation of preoperative cochlear duct length estimation W. Wimmer¹, N. Gerber¹, A. Dhanasingh³, P. Mistrik³, C. Jolly³, B. Bell¹, S. Weber¹, M. Kompis², M. Caversaccio² ¹ ARTORG Center for Biomedical Engineering, University of Bern, Switzerland ² Department of ENT, Head and Neck Surgery, Inselspital, University of Bern, Switzerland ³ Med-El Corporation, Innsbruck, Austria Contact: wilhelm.wimmer@artorg.unibe.ch Abstract: Variation in human cochlear dimension must be considered when selecting a patient-suitable electrode array for coch- lear implantation. A promising way of cochlear duct length (CDL) prediction utilizes statistical properties of variations in cochlear morphology. In this work, CDL values estimated by statistically derived equations were validated using three-dimensional measurements in micro-CT data sets of seven human temporal bones with implanted electrode ar- rays. Further, the lateral wall length (LWL) was assessed manually and compared to the prediction of Escudé’s equa- tion. Comparison showed good congruency of the measured and predicted CDL and LWL at one turn length (basal turn). Deviations of about 5 % were observed in CDL at 1.5 turn lengths, as well in LWL at 1.5 and 2 turn lengths. Re- sults suggest that CDL prediction based on a single radiographic measurement of the cochlea could support surgeons in electrode array selection, but further investigation with increased sample size is necessary. Key words: Cochlear implantation, Insertion depth, Escudé’s equation 1 Problem Knowledge of the patient-specific cochlear duct length (CDL) is particularly important when precise intracochlear elec- trode array placement is desired. In cases with no residual hearing the surgeon aims to insert the electrode array as deep- ly as possible in order to achieve a full coverage of the sensory range. In contrast, in cases of patients with residual hearing, electrode arrays are designed to be placed only partially within the cochlea (electric-acoustic stimulation, up to 1.5 turns). Provided that a resistance-free (and atraumatic) insertion can be achieved, the usage of an electrode array with suitable length is crucial for optimal implant placement in both patients with and without residual hearing. Never- theless, an electrode array inserted the same length in two different cochleae, may result in a completely different inser- tion depth angle due to variations of the cochlear size. In this context, a preoperative estimation of the CDL could help the surgeon to choose an electrode array suitable for the patient’s anatomy and therefore increase the patient’s benefit after implantation. The utilization of statistical correlation between the length of the organ of Corti and the size of the cochlea seems to be a possible method for preoperative CDL estimation. The longest diameter of the basal turn (distance ‘A’) through the round window (RW) and the modiolar axis can be determined by single plane assessments in preoperative radiographs acquired as part of the routine clinical pro- cedure. Based on previous work on cochlear size variation, statistical equations were formulated allowing for an estima- tion of the CDL at 1, 1.5, 2 turn lengths (TL) of the cochlea [1-7]. This work aims to validate the CDL prediction equations using in-vitro three-dimensional measurements in micro- computed tomography (μCT) datasets. Moreover, measurements of the cochlear lateral wall length (LWL) will be com- pared to the determination of Escudé’s equation [8]. 2 Materials and Methods In a preceding study, 1.8 mm diameter direct cochlear access (DCA) tunnels were drilled by an image-guided robotic system in eight Thiel-embalmed human temporal bones (both sides of four heads). The robot system was specifically constructed for surgeries on the lateral skull base and has a targeting accuracy of 0.15 ± 0.08 mm at the RW [9]. Four standard electrode arrays (31.5 mm) and four Flex28 electrode arrays (28 mm) provided by Med-El Corporation (Inns- bruck, Austria) were implanted via the DCA tunnel at the RW. Electrode array insertion was stopped as soon as re- sistance was detected. Temporal bones were then excised and trimmed in order to fit in a specimen holder of 36 mm in 143 diameter. One cochlea was damaged during this preparation step. Electrode array insertion depth angles were found to be 606°±79° (n=7) [10]. Samples were scanned using a μCT device (μCT 40, SCANCO Medical AG, Brüttisellen, Switzerland) using a 70 kVp tube potential and 114 μA tube current. Determined by the size of the samples, a voxel size of 18 x 18 x 18 μm3 was ob- tained. Thereafter, the cochlea was segmented using Amira 5 visualization software (VSG, Burlington, MA, USA) start- ing with a region growing algorithm. Subsequently, manual correction of imaging artifacts caused by the implanted electrode arrays (exponential edge-gradient effect) was performed. Next, a 3D surface model was generated and a zero reference angle plane [11] intersecting the modiolar axis and the center of the RW was manually aligned as seen in fig. 1. The longest distance from the RW to the lateral wall of the ba- sal turn was then measured in this plane (distance ‘A’, fig. 1). Further, the lateral wall length (LWL) was measured manually along the surface of the cochlea, following the outermost points of the cochlear turns. Starting from the zero reference angle plane, the LWL was obtained at 1 TL (360°), 1.5 TL (540°) and 2 TL (720°), as shown in fig. 2. For CDL measurement, a 3rd order spline (500 samples per turn) was fitted in the center of the segmented electrode array (as an approximation to the position of the organ of Corti). In case of array bending in the proximal part of the basal turn, the spline was aligned with respect to the lateral wall course (fig. 3). CDL at 2 TL was determined using overlaid μCT slices in order to locate the basilar membrane position if the electrode array was not inserted deeper than 1.5 TL. Figure 1: Surface model of a left human cochlea and implanted electrode array (dark gray). The zero reference angle plane is aligned through the modiolar axis and the center of the round window (RW), perpendicular to the basal turn. The distance ‘A’ is found between the RW center point and the outermost opposite surface of the basal turn. Figure 2: Surface model of a right human cochlea and lateral wall length measurement paths for 1, 1.5 and 2 turns (TL). 144 Figure 3: Measurement of cochlear duct length in the basal turn of a right human coch- lea with implanted electrode array (EA). A spline (CD) is fitted in the center of the elec- trode array in order to approximate the position of the cochlear duct. In the proximal part of the basal turn electrode array bending lead to a deviation of the array position compared to the location of the organ of Corti. Therefore, the spline is positioned with respect to the course of the lateral wall (LW). Based on the measured distance ‘A’ (mm), the CDL (mm) was computed at 1, 1.5 and 2 TL using following set of equa- tions: ‫ܮܦܥ‬ை஼ǡଵ்௅ ൌ ʹǤͶ͵ ȉ ‫ ܣ‬െ ʹǤͶ͵ (1) ‫ܮܦܥ‬ை஼ǡଵǤହ்௅ ൌ ͵ǤͲͲ ȉ ‫ ܣ‬െ ͵ǤͲʹ (2) ‫ܮܦܥ‬ை஼ǡଶ்௅ ൌ ͵Ǥ͸ͷ ȉ ‫ ܣ‬െ ͵Ǥ͸͵ (3) Further, the LWL (mm) was estimated by applying Escudé’s equation (insertion depth angle θ in degrees) [8]: ‫ ܮܹܮ‬ൌ ʹǤ͸ʹ ȉ ‫ ܣ‬ȉ ŽሺͳǤͲ ൅ ߠȀʹ͵ͷሻ (4) 3 Results Lengths of distance ‘A’ were found to range within 8.86-9.77 mm, with a mean of 9.35±0.32 mm (n=7). At 1 TL, meas- ured values showed correspondence with the lengths estimated by eqs. (1) and (4). Measurements revealed a deviation to the estimated CDL of about 1-1.5 mm and 0.5 mm at 1.5 TL and 2 TL, respectively. LWL measurements were found to deviate about 1 mm at 1.5 TL and 1.5 mm at 2 TL (fig. 4). Fig. 4: Summary data for distance ‘A’, as well as the estimated (○) and the measured (♦) cochlear duct and lateral wall lengths. Linear regression is plotted for predicted (dotted lines) and measured (dashed lines) values at 1, 1.5 and 2 turn lengths (TL) of the samples (n=7). 145 4 Discussion In this work, three-dimensional in-vitro measurements of the CDL and LWL of human cochleae were obtained and compared to values estimated by statistically derived equations. Although samples were taken from four human heads only, a high variability of the cochlear size, reflected by ‘A’, was observed. The cochlear size variation of the investigat- ed samples (9.35±0.32 mm, n=7) are within the range reported in literature (6.8-10.3 mm, mean 8.55±0.57 mm, n=104 [6]). The CDL was more or less congruent at 1 TL and 2 TL, whereas a deviation of approximately 5% was observed at 1.5 TL. This may be a result of either the measurement method (overestimation of the CDL at 1.5 TL), the impact of ana- tomical variations due to a small sample size or a deviation within the derived equations. Deviations of about 5% were observed for the LWL at 1.5 TL and 2 TL, which may again be caused by the measurement method or anatomical varia- tion. Nevertheless, a trend of higher difference between the predicted and measured values for increasing angles can be seen. These preliminary results suggest that a preoperative estimation of the CDL based on the measurement of a single value (distance ‘A’) is a practical approach for patient-specific electrode array selection in both cases with and without residu- al hearing. According to literature, the CDL of human cochleae ranges from 25 to 36 mm, with an average value of 31.5 mm (n=95) [1,2]. Currently there is no single free-fitting electrode array available which covers the whole CDL range. A short electrode array inserted into a cochlea with a long CDL is not sufficient in stimulating low frequency regions. Conversely, intracochlear damage may occur if a long electrode array is inserted into a cochlea with short CDL. Further, the risk of occurrence of extracochlear contacts is increased when electrode arrays longer than the CDL are inserted. In this context, findings of this study may help surgeons in reliably selecting a suitable electrode array length from portfo- lios provided by different CI manufacturers and ultimately increase the patient’s benefit. In order to improve the method, further investigations with an increased sample size and statistical analysis are carried out. Moreover, the measurement accuracy of the presented method and the influence of lower imaging resolution of clinical data are evaluated. 5 Acknowledgments This work was financially supported by the European Union's Seventh Framework Programme (FP7/2007-2013) under HEAR-EU grant agreement n° 304857. Electrode arrays and the software tool for CDL estimation were provided by Med-El Corporation. 6 References [1] Hardy M: The length of the organ of Corti in man. Am J Anat 62:179-311 (1938) [2] Lee J, Nadol JB, Eddington DK: Depth of Electrode Insertion and Postoperative Performance in Humans with Cochlear Implants: A Histopathologic Study. Audiol Neurotol 15:323-331 (2010) [3] Kawano A, Seldon H, Clark G: Computer aided three dimensional reconstruction in human cochlear maps: Measurements of the lengths of organ of corti, outer wall, and rosenthal’s canal. Ann Otol Rhinol Laryngol 105:701-709 (1996) [4] Erixon E, Högstorp H, Wadin K, Rask-Andersen H: Variational Anatomy of the Human Cochlea: Implications for Cochlear Implantation. Otol Neurotol 30:14-22 (2008) [5] Adunka O, Unkelbach MH, Mack MG, Radeloff A, Gstoettner W: Predicting Basal Cochlear Length for Elec- tric-Acoustic Stimulation. Arch Otolaryngol 131: 488-492 (2005) [6] Martinez-Monedero R, Niparko JK, Aygun N: Cochlear Coiling Pattern and Orientation Differences in Cochle- ar Implant Candidates. Otol Neurotol 32: 1086-1093 (2011) [7] Dimopoulos P, Muren C: Anatomic variations of the cochlea and relations to other temporal bone structures. Acta Radiol 31:439-444 (1990) [8] Escudé B, James C, Deguine O, Cochard N, Eter E, Fraysse B: The Size of the Cochlea and Predictions of Inser- tion Depth Angles for Cochlear Implant Electrodes. Audiol Neurotol 11:27-33 (2006) [9] Bell B, Williamson T, Gerber N, Gavaghan KA, Wimmer W, Caversaccio M, Weber S: In Vitro Accuracy Evalu- ation of Image-Guided Robot System for Direct Cochlear Access. Otol Neurotol 2013;in print. [10] Wimmer W, Bell B, Huth ME, Weisstanner C, Gerber N, Kompis M, Weber S, Caversaccio M: Cone Beam and Micro Computed Tomography Validation of Manual Array Insertion for Minimally Invasive Cochlear Implanta- tion. Audiol Neurotol 2013:in review. [11] Verbist BM, Skinner MW, Cohen LT, Leake PA, James C, et. Al.: Consensus Panel on a Cochlear Coordinate System Applicable in Histologic, Physiologic, and Radiologic Studies of the Human Cochlea. Otol Neurotol 31:722-730 (2010). 146