OBJECTIVE:To develop and externally validate a deep learning segmentation network capable of automatically segmenting the inner ear in preoperative clinical computed tomography (CT) scans across various resolutions and protocols. METHODS:A deep learning-based segmentation network was developed using 100 cadaveric specimens that were scanned with synchrotron-radiation phase contrast imaging (SR-PCI) and various clinical CT scanners. Different acquisitions, protocols, and augmentations were used to create a total of 4,784 paired SR-PCI and clinical three-dimensional datasets used for deep learning training and model development. Performance and accuracy of the network were assessed on a separate unseen dataset and externally validated against manual segmentations from seven individual domain experts (otologists and radiologists), the mean expert performance, and a simultaneous truth and performance level (STAPLE) consensus segmentation. RESULTS:The network pipeline significantly outperformed each individual expert segmentation, the average of the expert segmentations, and the STAPLE consensus segmentation. Compared to the SR-PCI ground truth data, the network achieved a Dice similarity coefficient of 0.922, a maximum absolute Hausdorff distance of 0.329 mm, and an average Hausdorff distance of 0.006 mm on cone-beam CT and helical CT with resolutions as low as 625 μm. CONCLUSION:This is the first automated segmentation algorithm for the inner ear that has been shown to outperform segmentations from domain experts, establishing a new clinical gold standard. LEVEL OF EVIDENCE:N/A.

OBJECTIVE:Visualization of cochlear implant (CI) electrode contact position offers insights into electrode-cochlea biomechanics. Prior in vivo imaging studies have assessed electrode distances to a fixed modiolar axis, but electrodes' orientation within the cochlear canal and their relationship to the scala tympani's dynamic anatomical tilt have not been investigated. This study employed in vivo flat-panel computed tomography (FPCT) imaging to examine the flexibility and dynamic behavior of lateral wall arrays. METHODS:This retrospective cohort study included eight MED-EL FLEX28 CI users. FPCT images were acquired, and reconstructions were aligned to a standardized cochlear coordinate system. Vectors perpendicular to electrode contacts were used to estimate contact orientation using mid-modiolar image slices and 3D renderings. Anatomical landmarks were used to calculate scalar tilt. RESULTS:A significant correlation was observed between scalar tilt and contact orientation (r = -0.44, p < 0.0001). The scala tympani exhibited its greatest tilt relative to the X-Y plane in the hook region, then transitioned to a parallel orientation at approximately 180° angular depth before increasing tilt again between 180° and 360°. Electrode arrays demonstrated rotational alignment with the scala between 180° and 360° angular insertion depth. CONCLUSION:Flexible lateral wall electrode arrays accommodate the natural anatomical tilt of the scala tympani, with contact orientation aligning to cochlear structures for much of the insertion trajectory. This in vivo analysis provides new insights into the biomechanical interactions of electrode arrays within the cochlea and may have implications for structure preservation, spectral resolution, and CI stimulation patterns. LEVEL OF EVIDENCE:N/A.

IntroductionThe natural, tonotopic frequency distribution of the inner ear is typically described by the Greenwood function, which logarithmically projects the audible frequency spectrum onto the intracochlear basilar membrane. Recent developments in cochlear implant (CI) programming aim to improve sound quality and music perception through consideration of the frequency distribution as described by the Greenwood function when assigning frequency bands to the individual contacts of the electrode array. This approach is commonly referred to as anatomy-based fitting (ABF). However, empirical validation of the Greenwood function to accurately describe pitch as perceived by CI users is lacking.MethodsTwelve CI patients with single-sided deafness (SSD) participated in the study. A pitch matching task was conducted at four different appointments and with two different fitting maps (standard and ABF). At each test appointment, participants were asked to set the frequency of a pure tone presented through a loudspeaker to the pitch perceived when stimulated with the single contacts of the CI electrode array. The cochlear anatomy of the patients was reconstructed based on clinical imaging to derive the location of the stimulating contacts relative to the basilar membrane, allowing for the comparison of the pitch perceived by the patients to the frequency suggested by the Greenwood function for each stimulating contact.ResultsIn general, subjective pitch percepts were found to agree well with the frequency suggested by the Greenwood function independent of subject, contact, or applied fitting map. Differences between pitch matches and Greenwood were found to be not statistically significant. At least part of the outcomes of previous studies reporting a basal frequency shift can be explained by the tonotopic mapping functions applied within these studies.DiscussionThe present results suggest that the Greenwood function is well-suited for representing the tonotopic frequency distribution not only for normal hearing subjects but for CI recipients as well. Further advances in frequency mapping should also take the neural health of the cochlea into account, allowing for additional individualization of frequency mapping in CIs.

Objective Comparison of acute speech recognition for cochlear implant (CI) alone and electric‐acoustic stimulation (EAS) users listening with default maps or place‐based maps using either a spiral ganglion (SG) or a new Synchrotron Radiation‐Artificial Intelligence (SR‐AI) frequency‐to‐place function. Methods Thirteen adult CI‐alone or EAS users completed a task of speech recognition at initial device activation with maps that differed in the electric filter frequency assignments. The three map conditions were: (1) maps with the default filter settings ( default map ), (2) place‐based maps with filters aligned to cochlear SG tonotopicity using the SG function ( SG place‐based map ), and (3) place‐based maps with filters aligned to cochlear Organ of Corti (OC) tonotopicity using the SR‐AI function ( SR‐AI place‐based map ). Speech recognition was evaluated using a vowel recognition task. Performance was scored as the percent correct for formant 1 recognition due to the rationale that the maps would deviate the most in the estimated cochlear place frequency for low frequencies. Results On average, participants had better performance with the OC SR‐AI place‐based map as compared to the SG place‐based map and the default map. A larger performance benefit was observed for EAS users than for CI‐alone users. Conclusion These pilot data suggest that EAS and CI‐alone users may experience better performance with a patient‐centered mapping approach that accounts for the variability in cochlear morphology (OC SR‐AI frequency‐to‐place function) in the individualization of the electric filter frequencies (place‐based mapping procedure). Level of Evidence 3 Laryngoscope , 133:3540–3547, 2023

PURPOSE:To introduce and characterize inexpensive and easily produced 3D-printed axon-mimetic diffusion MRI phantoms in terms of pore geometry and diffusion kurtosis imaging metrics. METHODS:Phantoms were 3D-printed with a composite printing material that, after the dissolution of the polyvinyl alcohol, exhibits microscopic fibrous pores. Confocal microscopy and synchrotron phase-contrast micro-CT imaging were performed to visualize and assess the pore sizes. Diffusion MRI scans of four identical phantoms and phantoms with varying print parameters in water were performed at 9.4 T. Diffusion kurtosis imaging was fit to both data sets and used to assess the reproducibility between phantoms and effects of print parameters on diffusion kurtosis imaging metrics. Identical scans were performed 25 and 76 days later, to test their stability. RESULTS:Segmentation of pores in three microscopy images yielded a mean, median, and SD of equivalent pore diameters of 7.57 μm, 3.51 μm, and 12.13 μm, respectively. Phantoms had T1 /T2 = 2 seconds/180 ms, and those with identical parameters showed a low coefficient of variation (~10%) in mean diffusivity (1.38 × 10-3 mm2 /s) and kurtosis (0.52) metrics and radial diffusivity (1.01 × 10-3 mm2 /s) and kurtosis (1.13) metrics. Printing temperature and speed had a small effect on diffusion kurtosis imaging metrics (< 16%), whereas infill density had a larger and more variable effect (> 16%). The stability analysis showed small changes over 2.5 months (< 7%). CONCLUSION:Three-dimension-printed axon-mimetic phantoms can mimic the fibrous structure of axon bundles on a microscopic scale, serving as complex, anisotropic diffusion MRI phantoms.

The human cochlea transforms sound waves into electrical signals in the acoustic nerve fibers with high acuity. This transformation occurs via vibrating anisotropic membranes (basilar and tectorial membranes) and frequency-specific hair cell receptors. Frequency-positions can be mapped within the cochlea to create a tonotopic chart which fits an almost-exponential function with lowest frequencies positioned apically and highest frequencies positioned at the cochlear base (Bekesy 1960, Greenwood 1961). To date, models of frequency positions have been based on a two-dimensional analysis with inaccurate representations of the cochlear hook region. In the present study, the first three-dimensional frequency analysis of the cochlea using dendritic mapping to obtain accurate tonotopic maps of the human basilar membrane/organ of Corti and the spiral ganglion was performed. A novel imaging technique, synchrotron radiation phase-contrast imaging, was used and a spiral ganglion frequency function was estimated by nonlinear least squares fitting a Greenwood-like function ( F = A (10 ax − K )) to the data . The three-dimensional tonotopic data presented herein has large implications for validating electrode position and creating customized frequency maps for cochlear implant recipients.

Objective: Cochlear implants are traditionally programmed to stimulate according to a generalized frequency map, where individual anatomic variability is not considered when selecting the centre frequency of stimulation of each implant electrode. However, high variability in cochlear size and spatial frequency distributions exist among individuals. Generalized cochlear implant frequency maps can result in large pitch perception errors and reduced hearing outcomes for cochlear implant recipients. The objective of this work was to develop an individualized frequency mapping technique for the human cochlea to allow for patient-specific cochlear implant stimulation. Methods: Ten cadaveric human cochleae were scanned using synchrotron radiation phase-contrast imaging (SR-PCI) combined with computed tomography (CT). For each cochlea, ground truth angle-frequency measurements were obtained in three-dimensions using the SR-PCI CT data. Using an approach designed to minimize perceptual error in frequency estimation, an individualized frequency function was determined to relate angular depth to frequency within the cochlea. Results: The individualized frequency mapping function significantly reduced pitch errors in comparison to the current gold standard generalized approach. Conclusion and Significance: This paper presents for the first time a cochlear frequency map which can be individualized using only the angular length of cochleae. This approach can be applied in the clinical setting and has the potential to revolutionize cochlear implant programming for patients worldwide.

HYPOTHESIS:Measuring the length of the basilar membrane (BM) in the cochlear hook region will result in improved accuracy of cochlear duct length (CDL) measurements.BACKGROUND:Cochlear implant pitch mapping is generally performed in a patient independent approach, which has been shown to result in place-pitch mismatches. In order to customize cochlear implant pitch maps, accurate CDL measurements must be obtained. CDL measurements generally begin at the center of the round window (RW) and ignore the basal-most portion of the BM in the hook region. Measuring the size and morphology of the BM in the hook region can improve CDL measurements and our understanding of cochlear tonotopy.METHODS:Ten cadaveric human cochleae underwent synchrotron radiation phase-contrast imaging. The length of the BM through the hook region and CDL were measured. Two different CDL measurements were obtained for each sample, with starting points at the center of the RW (CDLRW) and the basal-most tip of the BM (CDLHR). Regression analysis was performed to relate CDLRW to CDLHR. A three-dimensional polynomial model was determined to describe the average BM hook region morphology.RESULTS:The mean CDLRW value was 33.03 ± 1.62 mm, and the mean CDLHR value was 34.68 ± 1.72 mm. The following relationship was determined between CDLRW and CDLHR: CDLHR  = 1.06(CDLRW)-0.26 (R2  = 0.99).CONCLUSION:The length and morphology of the hook region was determined. Current measurements underestimate CDL in the hook region and can be corrected using the results herein.

HYPOTHESIS Evaluating the accuracy of cochlear duct length (CDL) measurements from a published three-dimensional (3D) analytical cochlear model using Synchrotron Radiation Phase-Contrast Imaging (SR-PCI) data will help determine its clinical applicability and allow for model adjustments to increase accuracy. BACKGROUND Accurate CDL determination can aid in cochlear implant sizing for full coverage and frequency map programming, which has the potential to improve hearing outcomes in patients. To overcome problems with the currently available techniques for CDL determination, a novel 3D analytical cochlear model, dependent on four basal turn distances, was proposed in the literature. METHODS SR-PCI data from 11 cadaveric human cochleae were used to obtain reference measurements. CDL values generated by the analytical cochlear model were evaluated in two conditions: when the number of cochlear turns (NCT) were automatically predicted based on the four input distances, and when the NCT were manually specified based on SR-PCI data. RESULTS When the analytical cochlear model automatically predicted the NCT, the mean absolute error was 2.6 ± 1.6 mm, with only 27% (3/11) of the samples having an error in the clinically acceptable range of ±1.5 mm. When the NCT were manually specified based on SR-PCI data, the mean absolute error was reduced to 1.0 ± 0.6 mm, with 73% (8/11) of the samples having a clinically acceptable error. CONCLUSION The 3D analytical cochlear model introduced in the literature is effective at modeling the 3D geometry of individual cochleae, however tuning in the NCT estimation is required.

Background Despite significant anatomical variation amongst patients, cochlear implant frequency-mapping has traditionally followed a patient-independent approach. Basilar membrane (BM) length is required for patient-specific frequency-mapping, however cochlear duct length (CDL) measurements generally extend to the apical tip of the entire cochlea or have no clearly defined end-point. By characterizing the length between the end of the BM and the apical tip of the entire cochlea (helicotrema length), current CDL models can be corrected to obtain the appropriate BM length. Synchrotron radiation phase-contrast imaging has made this analysis possible due to the soft-tissue contrast through the entire cochlear apex. Methods Helicotrema linear length and helicotrema angular length measurements were performed on synchrotron radiation phase-contrast imaging data of 14 cadaveric human cochleae. On a sub-set of six samples, the CDL to the apical tip of the entire cochlea (CDL TIP ) and the BM length (CDL BM ) were determined. Regression analysis was performed to assess the relationship between CDL TIP and CDL BM . Results The mean helicotrema linear length and helicotrema angular length values were 1.6 ± 0.9 mm and 67.8 ± 37.9 degrees, respectively. Regression analysis revealed the following relationship between CDL TIP and CDL BM : CDL BM = 0.88(CDL TIP ) + 3.71 ( R 2 = 0.995). Conclusion This is the first known study to characterize the length of the helicotrema in the context of CDL measurements. It was determined that the distance between the end of the BM and the tip of the entire cochlea is clinically consequential. A relationship was determined that can predict the BM length of an individual patient based on their respective CDL measured to the apical tip of the cochlea.

Purpose To introduce and characterize inexpensive and easily produced 3D-printed axon-mimetic (3AM) diffusion MRI (dMRI) phantoms in terms of pore geometry and diffusion kurtosis imaging (DKI) metrics. Methods Phantoms were 3D-printed with a composite printing material that, after dissolution of the PVA, exhibits microscopic fibrous pores. Confocal microscopy and synchrotron phase contrast micro-CT imaging were performed to visualize and assess the pore sizes. dMRI scans of four identical phantoms and phantoms with varying print parameters in water were performed at 9.4T. DKI was fit to both datasets and used to assess the reproducibility between phantoms and effects of print parameters on DKI metrics. Identical scans were performed 25 and 76 days later to test their stability. Results Segmentation of pores in three microscopy images yielded a mean, median, and standard deviation of equivalent pore diameters of 7.57 μm, 3.51 μm, and 12.13 μm, respectively. Phantoms with identical parameters showed a low coefficient of variation (∼10%) in DKI metrics (D=1.38 ×10−3 mm2/s and K=0.52, T1= 3960 ms and T2=119 ms). Printing temperature and speed had a small effect on DKI metrics (<16%) while infill density had a larger and more variable effect (>16%). The stability analysis showed small changes over 2.5 months (<7%). Conclusion 3AM phantoms can mimic the fibrous structure of axon bundles on a microscopic scale, serving as complex, anisotropic dMRI phantoms. ### Competing Interest Statement The authors have declared no competing interest.