Abstract
Objective:
This study aimed to construct three-dimensional (3D) inner ear models in common animal models using micro-computed tomography (micro-CT) and a highly semi-automatic centerline-based method and to preliminarily assess the morphological similarities of their bony labyrinths to those in humans.
Methods:
Temporal bones from three adult mice, three guinea pigs, three minipigs, and one human were scanned using micro-CT. After segmentation, 3D inner ear models were reconstructed. A highly semi-automated centerline method was established to estimate 3D properties and calculate length, volume, and angular parameters across species.
Results:
Detailed 3D models of the cochlea, vestibule, and semicircular canals were successfully constructed. The mean lengths and volumes of the inner ear and semicircular canals generally increased with body size, with the cochlear volume and length of minipigs demonstrating proximity to human data. Notably, numerical assessments have shown that the semicircular canals in both mice and minipigs closely resemble human structures.
Conclusion:
Highly semi-automatic centerline-based measurements are effective tools for assessing 3D inner ear morphology. The bony labyrinths of minipigs demonstrated proximity to human inner ear parameters in terms of length, volume, and semicircular canal angles. However, mice exhibited an even closer resemblance to humans, specifically in semicircular canal angles. These findings provide preliminary guidance for the selection of animal models for various otological research fields.
Keywords: Cochlea, semicircular canals, micro-ct, morphometry, three-dimensional imaging
KEY MESSAGES
-
(1)
Developed a highly semi-automated, micro-CT-based centerline method for precise 3D inner ear characterization
-
(2)
Achieved the first high-precision 3D visualization and geometric morphometric analysis of the minipig inner ear
-
(3)
Interspecies comparisons confirm that minipigs exhibit a higher anatomical similarity to humans in the length and volume of the cochlea, while mice are more similar to humans in terms of the angles of the semicircular canals.
-
(4)
Provides a mathematical basis for the study of inner ear mechanics.
INTRODUCTION
The osseous labyrinth, situated within the petrous portion of the temporal bone, comprises the vestibule, semicircular canals, and cochlea. The intricate geometry of these small inner ear components and their spatial relationships are critical for both surgical intervention and fundamental research on auditory mechanics.[1,2,3] The precise geometric characterization of the inner ear is essential for elucidating its physiological functions. Consequently, various imaging modalities have been used to acquire anatomical data, including traditional histology, computed tomography (CT), magnetic resonance imaging (MRI), and synchrotron radiation-based micro-computed tomography (micro-CT).[1,4,5,6,7,8] Advanced post-processing and 3D visualization enable precise quantification of inner ear structures. The ability to manipulate these 3D models from any spatial orientation facilitates comprehensive visualization of anatomical features, identification of subtle structural variations, and enhanced evaluation of pathological conditions.[3,8,9] Considering the density resolution and operational procedures, CT appears to be the optimal method for obtaining inner ear topography.[10] In particular, micro-CT offers a significantly higher resolution, meeting the stringent requirements for the 3D reconstruction of fine inner ear structures.[11]
Mammalian models are indispensable in otological research for visualizing temporal bone anatomy, evaluating surgical trauma, and establishing hearing loss models.[12,13,14,15] The accurate quantification of inner ear dimensions is crucial for mechanical modeling and the analysis of acoustic stress distribution.[16] A comprehensive characterization of the morphological features of these animal models is essential for advancing biomechanical insights into the similarities and differences between animal models and the human inner ear.[11,14,17] However, the quantitative analysis of these 3D models is highly dependent on elaborate preparation, image resolution, and iterative measurements, which may cause radiation damage or inevitable errors and eventually affect the accuracy and validity of the results.[9] Furthermore, the lack of quantitative cross-species comparisons of the inner ear hinders evidence-based selection of animal models for diverse otological research paradigms.
Despite the inherent complexity of the inner ear, the cochlea has an approximate spiral geometry, whereas the semicircular canals resemble quasi-circular arcs. Recently, the centerline method, which has been successfully implemented in the segmentation of the vasculature[18], vocal tract[19,20], and cochlea[12], has demonstrated robust performance in path tracking. This approach facilitates the precise morphological characterization of intricate anatomical structures such as the inner ear. In contrast to conventional slice-by-slice segmentation, this method simplifies the process by requiring only key landmark identification. By calculating the optimal path using 3D grayscale gradients, the algorithm automatically constructs the centerline and executes segmentation, thereby significantly reducing manual intervention.[21,22]
Accordingly, this study used the centerline method to overcome the limitations of traditional 3D imaging, thereby enabling the highly semi-automated extraction of morphological parameters across different mammals. This highly semi-automated approach offers a robust platform for extensive structural and morphometric analyses. The present study delivers comprehensive 3D visualizations of the inner ear across various species while providing a mathematical characterization of the anatomical divergences between animal models and the human labyrinth.
MATERIALS AND METHODS
Preparation of Mammal Specimens
To investigate the differences between animals, healthy adult mice (9 weeks, weight 22–25 g), guinea pigs (15 weeks, average weight 350 g), and minipigs (10 months, average weight 15 kg; n = 3 per group) were included in this study. All animal procedures were approved by the Ethics Committee of the Chinese PLA General Hospital (Approval No. S2020-465-01). Additionally, a left adult human temporal bone specimen obtained from a cadaveric donor without pathological changes was analyzed to compare anatomical variations between animals and humans. The collection and use of the human specimen were approved by the Ethics Committee of the Chinese PLA General Hospital (Approval No. S2020-466-10).
To obtain an image of the intact inner ear, animal samples were prepared as follows. After intraperitoneal injection of a lethal dose of sodium pentobarbital, mice, guinea pigs, and minipigs were sacrificed. Following decapitation, the heads of the mice and guinea pigs were divided, and both the left and right bullae were removed. The left and right temporal bones of the minipigs, including the complete middle and inner ears, were cut immediately after decapitation. Because the human temporal bone was obtained from the left side, and considering the potential differences between the left and right temporal bones, only left-sided specimens from all animals were used for subsequent CT scanning. Iterative procedures for sample preparation, which are known to cause unavoidable sample shrinkage[8], were not required in our study because we used micro-CT. Therefore, fixation, dehydration, or decalcification steps were not performed.
Micro-CT Scanning Procedure
All the prepared samples were scanned using a high-speed micro-CT system (Quantum GX, PerkinElmer Inc., Norwalk, CT, USA). A 5 × 5 mm field of view was used for the mouse samples, a 20 × 20 mm field of view for guinea pigs and minipigs, and a 30 × 30 mm field of view for the human temporal bone. The scan was performed at a voltage of 90 kV and a current of 120 µA. A fast scan time of 4.50 min was set, and 512 slices of the image with voxel sizes of 9, 39, and 59 µm were acquired. The specific parameters are listed in Table 1. The dataset was exported to a disk in the DICOM format after workstation processing. No defects or infections were observed in any of the samples.
Table 1.
Micro-CT scanning parameters for different species
| Parameter | Mice | Guinea pigs/Minipigs | Human temporal bone |
|---|---|---|---|
| Scanner model | Quantum GX Micro-CT (PerkinElmer, USA) | ||
| Tube voltage (kV) | 90 | 90 | 90 |
| Tube current (µA) | 120 | 120 | 120 |
| Field of view (mm) | 5 × 5 | 20 × 20 | 30 × 30 |
| Rotation step | 0.5° (total 360° rotation) | ||
| Voxel size (µm) | 9.0 | 39.0 | 59.0 |
| Scan time (min) | 4.5 | 4.5 | 4.5 |
| Number of slices | 512 | 512 | 512 |
| Reconstruction method | Filtered back projection (FBP) | ||
| Data format | DICOM | DICOM | DICOM |
3D Reconstruction
The micro-CT DICOM datasets were imported into MIMICS 17.0 (Materialize, Leuven, Belgium) for image segmentation and 3D volumetric reconstruction. The interpolation function in MIMICS allows for highly semi-automated segmentation between slices.[23] The following pipeline was implemented to ensure high-fidelity modeling:
Segmentation and Thresholding
The initial segmentation was performed using a global thresholding technique. Based on the Hounsfield unit (HU) distribution, a specific gray-value range (1500–4000 HU) was defined to differentiate the high-density temporal bone from the background and air-filled cavities.
Region Growing and Refinement
To isolate the inner ear labyrinth, a region-growing algorithm was applied to ensure the connectivity of the mask and eliminate disconnected noise artifacts. To capture the intricate and interconnected morphology of the inner ear accurately, a 26-connectivity algorithm was used during the region-growing process. For areas where the bone density was low (e.g., the semicircular canal walls in mice), manual segmentation and interpolation functions were used across consecutive slices (transverse, coronal, and sagittal) to ensure anatomical continuity and prevent “pseudo-holes.”
3D Rendering and Post-processing
The 3D volume of the labyrinth was rendered using the “Calculate 3D” function with high-quality settings. To optimize the mesh for further analysis, a Laplacian smoothing filter (iterations: 3; smoothing factor: 0.1) was applied to reduce staircase artifacts while preserving the essential morphometric features of the bony labyrinth [Figure 1]. This final 3D model provided an anatomical basis for the subsequent centerline extraction and morphometric measurements.
Figure 1.
Volume render of the left temporal bone and inner ear in minipig model. The temporal bone is rendered translucent.
Automatic Measurements of 3D Models
Calculate Centerline
Centerline extraction was performed using the MedCAD module in Mimics 17.0. Given the microscopic dimensions of the specimens, the “Fit Centerline” tool was executed with a refined control point density of 0.05 mm (for humans and minipigs) or 0.02 mm (for mice and guinea pigs) to ensure that the trajectory accurately captured the curvature of the cochlea and semicircular canals. To ensure the accuracy and smoothness of the generated trajectory, a smoothing factor of 0.5 with 10 iterations was applied during the fitting process to eliminate staircase artifacts from the voxels.
After the automated calculation, the centerlines were manually inspected, and refinement was limited to a few control points that clearly deviated from the center to ensure that they remained equidistant from the canal walls, particularly in areas of high curvature, such as the common crus (CC) and ampullae. Specific anatomical landmarks were manually identified, including the round window niche, apex of the cochlea, centroids of the semicircular canal ampullae, and midpoint of CC. These landmarks were subsequently anchored to the centerline to establish their precise spatial relationship with the labyrinthine trajectory. The accuracy of landmark placement was verified by cross-referencing with orthogonal slices (transverse, coronal, and sagittal) to ensure colocalization with the centerline. Finally, the three-dimensional spatial coordinates of the landmarks and corresponding centerline nodes were exported for morphometric calculations. The spatial coordinates of these points and their corresponding centerlines are shown in Figures 2 and 3.
Figure 2.
Centerlines in 3D model of inner ear in minipig.
Figure 3.
Centerlines and landmarks of inner ear in minipig.
Automatic Measurements of Inner Ear Parameters in Various Samples
Distances between landmarks were used to automatically calculate the length. The length of the cochlea was automatically measured from its apex to the round window niche (bifurcation) along the centerline. The lengths of the anterior and posterior semicircular canals were measured from feature points representing the ampullae to the CC. The length of the lateral semicircular canal was measured between two footpoints. The CC length was measured using two footpoints.
The 3D labyrinth model was segmented into six anatomical components (cochlea [CO], vestibule [VL], anterior semicircular canal [ASC], lateral semicircular canal [LSC], posterior semicircular canal [PSC], and CC) using a standardized protocol in the MedCAD module of Mimics 17.0. To ensure reproducibility, segmentation was performed by defining the cutting planes perpendicular to the generated centerlines at specific anatomical landmarks: (1) the junction of the round window niche and the basal turn for the CO-VL boundary; (2) the transition from the ampulla to the vestibule for each canal; and (3) the bifurcation of the CC. Each component was isolated as a separate 3D object. Finally, the volume of each part was automatically calculated based on a voxel-based triangulation mesh using the “Properties” tool in Mimics. To minimize inter-observer bias, this process was repeated three times, and the mean values were recorded.
The extracted centerlines from Mimics 17.0 were exported in the Initial Graphics Exchange Specification (IGS) format and imported into Geomagic Studio 2012 for spatial orientation analysis. Initially, the “Mesh Doctor” module was used to automatically detect and rectify common mesh defects, including spike removal to eliminate erroneous protrusions caused by noise or partial volume effects and hole filling to seal any surface discontinuities. Subsequently, a “Global Remesh” operation was executed with a target edge length of 0.05 mm. This step standardized the triangulation density across the entire mesh to ensure a uniform distribution of geometric information.
To determine the orientation of each semicircular canal (ASC, LSC, and PSC), a reference plane of best fit was calculated for each centerline using the least squares fitting algorithm within the “Feature” module. To ensure the accuracy of the fitting, only points corresponding to the arc portion of each canal (excluding ampullary expansion to avoid morphological bias) were selected. To ensure computational rigor, the convergence tolerance was set to 10−4 mm. The quality of fit was evaluated using the root mean square error, which was maintained below 0.05 mm in all cases, ensuring that the fitted planes accurately represented the spatial trajectory of the canal centerlines. The reference planes in the corresponding semicircular canals are shown in Figure 4. Taking the anterior and lateral SC as examples, as shown in Figure 5, the method of obtaining the plane-included angle is as follows: 
= (a1, a2, a3) is the vector of the anterior SC reference plane, 
= (b1, b2, b3) is the vector of the lateral SC reference plane, and the angle between the two intersection planes is described in Eq. (1).[24,25]
Figure 4.
Illustration of the best fit plane in minipig model. ASC is the reference plane of anterior semicircular canal, LSC is the reference plane of lateral semicircular canal, and PSC is the reference plane of posterior semicircular canal.
Figure 5.
Illustration of the angle measurement in the minipig model. a: axis of the plane of the anterior semicircular canal; b: axis of the plane of the lateral semicircular canal; and α: included angle between a and b.
The algorithm used to calculate the angles was implemented using MATLAB R2012b (The MathWorks, Inc., Natick, MA, USA). All calculations were performed in the Cartesian coordinate system, and the final angles were expressed in degrees.
RESULTS
Geometric Reconstruction of the Inner Ear Renders Exquisite Structures of CO, VL, and SC
The intricate and elaborate structure of the inner ear model was rendered as a complex geometric solid composed of the cochlea, vestibule, and semicircular canals, as shown in Figure 6. 3D visualization allowed us to observe the surface texture and anatomical details of the inner ear, such as the apical and basal turns and the round and oval windows. The spatial location and concrete form of the semicircular canals were embodied in 3D models. The three semicircular canals were similar in length and appeared to be mutually perpendicular to each other. The common crus and ampulla were easily identified in the models. To the best of our knowledge, this is the first detailed 3D visualization of a minipig inner ear.
Figure 6.
Illustration of inner ear structure of different mammals. (A: mouse; B: guinea pig; C: minipig; and D: human). Abbreviations: ASC, anterior semicircular canal; CC, common crus; CO, cochlea; LSC, lateral semicircular canal; and PSC, posterior semicircular canal.
The Total Length of Inner Ear Increases with the Body Size
Figure 7 shows the morphologies of the centerlines and landmarks in three animal models and human model. The centerlines and landmarks form a framework for the morphological characterization of the inner ear. Landmarks along the centerlines correspond to the anatomical features in the 3D models, such as the oval window, ampullae, and common crus. The number of cochlear turns varies among species. In particular, mice have 2 turns, guinea pigs have 4 turns, minipigs have 3.5 turns, and humans have fewer than 3 turns [Figure 8].
Figure 7.
Centerlines and landmarks in various mammal models (A, B: mouse inner ear model [A] and its centerlines and landmarks [B]; C, D: guinea pig inner ear model [C] and its centerlines and landmarks [D]; E, F: minipig inner ear model [E] and its centerline and landmarks [F]; and G, H: human inner ear model [G] and its centerline and landmarks [H].
Figure 8.
Turns of different mammals’ cochlea (A: mouse cochlea; B: guinea pig cochlea; C: minipig cochlea; and D: human cochlea).
The length measurements of the inner ear models are listed in Table 2, including the cochlea, lateral semicircular canal, anterior semicircular canal, posterior semicircular canal, common crus, and total length. Within the same species, the cochlear length was the greatest, whereas the three semicircular canals exhibited comparable lengths and were shorter than the cochlea. The mean cochlear length in minipigs was closest to that in humans. Among the four species examined, the total length of the inner ear was positively correlated with body size.
Table 2.
Length measurements of inner ear (n = 3)
| Length(mm) | CO | LSC | ASC | PSC | CC | Total length |
|---|---|---|---|---|---|---|
| Mouse | 6.14 ± 0.36 | 3.54 ± 0.35 | 4.28 ± 0.33 | 3.90 ± 0.35 | 1.33 ± 0.13 | 19.19 ± 0.71 |
| Guinea pig | 17.65 ± 0.53 | 8.14 ± 0.36 | 7.60 ± 0.28 | 7.52 ± 0.69 | 2.80 ± 0.09 | 43.71 ± 0.99 |
| Minipig | 26.09 ± 1.22 | 9.72 ± 0.33 | 9.87 ± 0.67 | 9.36 ± 0.67 | 3.98 ± 0.34 | 59.02 ± 1.62 |
| Human | 24.95 | 13.26 | 16.24 | 16.93 | 4.15 | 75.53 |
Note: n = number of independent ears. The value represents mean ± SD for the group of mice, guinea pigs, and minipigs. Abbreviations: ASC, anterior semicircular canal; CC, common crus; CO, cochlea; LSC, lateral semicircular canal; PSC, posterior semicircular canal; and VL, vestibule.
To provide a preliminary inference of the differences in the inner ear parameters between various experimental animals and humans, the relative deviation (%) was calculated using Eq. (2).[26] The results of the relative deviation in the length of each substructure are presented in Table 3. Considering the relative deviation in the lengths of the inner ear substructures between the experimental animals and their human counterparts, minipigs demonstrated the closest overall resemblance to humans. Specifically, cochlear length exhibited the highest degree of similarity.
Table 3.
Relative deviation of inner ear substructure lengths (n = 3)
| Relative deviation(%) | CO | LSC | ASC | PSC | CC | Total length |
|---|---|---|---|---|---|---|
| Mouse | 75.39 | 73.30 | 73.65 | 76.63 | 67.95 | 74.59 |
| Guinea pig | 29.26 | 38.61 | 53.20 | 55.58 | 32.53 | 42.13 |
| Minipig | 4.57 | 30.10 | 39.22 | 44.71 | 4.10 | 21.86 |
Note: n = the number of independent ears. Abbreviations: ASC, anterior semicircular canal; CC, common crus; CO, cochlea; LSC, lateral semicircular canal; and PSC, posterior semicircular canal.
The Volume Parameters of the Inner Ear Increase with the Body Size of Animals
Figure 6 shows 3D models of the inner ears of different species, with segmented and color-coded components: cochlea (yellow), vestibule (green), anterior semicircular canal (red), lateral semicircular canal (gray), posterior semicircular canal (blue), and common crus (purple). The volume of each component was calculated using the Mimics software. Generally, the inner ear volumes of mice, guinea pigs, and minipigs are smaller than those of humans and increase with body size. In each species, the cochlea constituted the largest proportion, followed by the vestibule and semicircular canals, with the common crus being the smallest. Detailed data are presented in Table 4.
Table 4.
Volume measurements of inner ear (n = 3)
| Volume(mm3) | CO | VL | LSC | ASC | PSC | CC |
|---|---|---|---|---|---|---|
| Mouse | 1.54 ± 0.27 | 0.55 ± 0.05 | 0.18 ± 0.05 | 0.15 ± 0.01 | 0.14 ± 0.03 | 0.07 ± 0.01 |
| Guinea pigs | 13.68 ± 1.51 | 7.11 ± 0.89 | 1.52 ± 0.29 | 0.79 ± 0.14 | 0.80 ± 0.20 | 0.63 ± 0.26 |
| Minipigs | 40.97 ± 5.41 | 20.53 ± 2.13 | 3.13 ± 0.92 | 2.78 ± 0.66 | 2.41 ± 0.69 | 1.65 ± 0.30 |
| Human | 59.37 | 46.65 | 13.70 | 9.85 | 12.25 | 4.28 |
Note: n = the number of independent ears. The value represents mean ± SD for the group of mice, guinea pigs, and mini pigs. Abbreviations: ASC, anterior semicircular canal; CC, common crus; CO, cochlea; LSC, lateral semicircular canal; PSC, posterior semicircular canal; and VL, vestibule.
Relative deviation calculations, detailed in Eq. (2), were also performed to assess the volumetric similarity of inner ear substructures of each species to their human counterparts. Compared with rodents, the minipig inner ear substructure volumes demonstrated greater proximity to the human inner ear. Specifically, within the minipig inner ear, the cochlear volume exhibited the highest degree of similarity to that in humans. The relative deviation of each substructure is listed in Table 5.
Table 5.
Relative deviation of inner ear substructure volume (n = 3)
| Relative deviation(%) | CO | VL | LSC | ASC | PSC | CC |
|---|---|---|---|---|---|---|
| Mouse | 97.41 | 98.82 | 98.69 | 98.48 | 98.86 | 98.36 |
| Guinea pig | 76.96 | 84.76 | 88.91 | 91.98 | 93.47 | 85.28 |
| Minipig | 30.99 | 55.99 | 77.15 | 71.78 | 80.33 | 61.45 |
Note: n = the number of independent ears. Abbreviations: ASC, anterior semicircular canal; CC, common crus; CO, cochlea; LSC, lateral semicircular canal; PSC, posterior semicircular canal; and VL, vestibule.
Angular Parameters of the Inner Ear: Minipigs and Mice Exhibit Greater Human Similarity
The spatial relationships between the semicircular canals are presented in Table 6. Although the three reference planes appeared mutually perpendicular, the angular parameters differed among species. Our study also compared angular discrepancies between animal models and human data using Euclidean distance. It integrates the angular parameters of the lateral and anterior (L–A), lateral and posterior (L–P), and anterior and posterior (A–P) semicircular canals into a single measure of overall similarity to the human semicircular canal configuration, with smaller values indicating greater similarity. The calculation method is described by Eq. (3).[27] The results of Euclidean distance indicated that the angular parameters of mice were closest to those of humans in measurements, the minipigs were also very similar, and guinea pigs were the least similar; namely, the spatial relationships of semicircular canals were more similar to those of humans in mice and minipigs than in guinea pigs.
Table 6.
Angle measurements between the semicircular canals (n = 3)
| L–A SC(°) | L–P SC(°) | A–P SC(°) | Euclidean Distance | |
|---|---|---|---|---|
| Mouse | 95.84 ± 0.33 | 97.46 ± 0.80 | 91.54 ± 2.31 | 16.10 |
| Guinea pig | 100.77 ± 4.18 | 83.93 ± 1.88 | 87.43 ± 2.43 | 23.13 |
| minipig | 71.37 ± 4.01 | 83.45 ± 1.07 | 86.15 ± 2.15 | 16.33 |
| Human | 81.40 | 96.24 | 84.53 | 0.00 |
Note: n = the number of independent ears. Angle parameters for mouse, guinea pig, and minipig are expressed as mean ± SD. Abbreviations: L–A SC, angles between the lateral and anterior semicircular canals; L–P SC, angles between the lateral and posterior semicircular canals; and A–P SC, angles between the anterior and posterior semicircular canals.
DISCUSSION
In this study, micro-CT scanning and the centerline method were used to perform 3D modeling and precise quantification of various inner ear substructures (including length, volume, and semicircular canal angles). Notably, this is the first instance of 3D visualization and precise quantification of the inner ear of a minipig. The objective of this study was to provide a preliminary assessment of the morphological similarities between the bony labyrinths of different experimental animals and those of humans, thereby providing a foundation for future biomechanical studies on the inner ear membranous labyrinth.
In terms of the length and volume of various inner ear bony labyrinth substructures, particularly cochlear length and volume, which have greater clinical reference value for surgical procedures, minipigs demonstrate a higher degree of similarity to humans than rodent models (e.g., mice and guinea pigs). Although the anatomical similarity between the minipig middle ear and that of humans has been established[28], our findings extend this by providing a high-precision evaluation of the cochlear microstructure. Moreover, our analysis of the inner ear substructure lengths and volumes revealed that all parameters, except cochlear length, exhibited a positive correlation with animal body size. Notably, human cochlear length presented an exception, being marginally shorter than that of the minipig. We attribute this observation to the differing number of cochlear turns, with the human cochlea possessing approximately 2.5 turns compared to the minipig’s approximately 3.5 turns. A previous study reported a minipig cochlear length of approximately 39.48 mm.[28] Another study measured the cochlear length to be 32.89 mm.[29] Although these two values show a considerable discrepancy, both are larger than the values we obtained using the centerline method (26.09 ± 1.22 mm). Similarly, our measured human cochlear length of 24.95 mm fell below the range of 27.44–35.91 mm typically reported in the literature.[30] We attribute this difference to the measurement methodology. Specifically, we used the centerline method (measuring the central axis of the bony cochlear duct), whereas previous studies typically measured the outer contour or shell of the bony cochlea. This disparity became more pronounced with increasing cochlear diameter. Consequently, for the mouse, which possesses a smaller cochlear diameter, our measured mean cochlear length was 6.14 mm, closely aligning with previous data ranging from 5.81 to 6.67 mm.[31,32] The mean cochlear length of guinea pigs in our study was 17.65 mm, which is also consistent with other reported values ranging from 15.70 to 22.00 mm.[14,33] Regarding cochlear volume, the mean minipig cochlear volume measured using the centerline method in our study was 40.97 ± 5.41 mm3, which closely approximated the previously reported value of 39.96 mm3.[34] A similar consistency was observed in our measurements of mouse cochlear volume.[8] For human cochlear volume, although our study was limited to a single human fresh-frozen temporal bone specimen, our measured value of 59.37 mm3 also closely aligned with findings from previous research.[35] Therefore, we preliminarily suggest that the minipig cochlea, in terms of both volume and length, exhibits greater similarity to human dimensions than other rodent models, making it a promising animal model for clinical cochlear implantation and other surgical procedures.
Semicircular canals are sensory organs that are sensitive to angular acceleration of the head.[36] Understanding the planar orientation of semicircular canals is crucial for advancing fundamental research into postural and balance control and forms the basis for the diagnosis and treatment of patients experiencing dizziness.[36,37] Although no published studies have reported the angular dimensions of the inner ear semicircular canals in mice, guinea pigs, or minipigs to date, our preliminary findings obtained using the centerline method provide a novel contribution by addressing this critical research gap. Contrary to the common anatomical depictions of mutually orthogonal semicircular canals, our results indicate that their planes are not perfectly perpendicular. This observation aligns with prior investigations into the inter-canal angles in humans, using the method of a histologic section. Hashimoto et al. reconstructed 3D models of semicircular canals and reported that the average angle for each pair of semicircular canals was approximately 90°.[36] Specifically, recent MRI-based soft tissue modeling studies have calculated mean inter-canal angles for the L–A SC at 84.67 ± 5.76°, A–P SC at 94.21 ± 3.81°, and L–P SC at 91.79 ± 5.22°.[38] Our measured angles for the L–A and L–P SCs were in good agreement with the MRI-based soft tissue modeling studies. However, the anterior–posterior (A–P) semicircular canal angle in our study was slightly lower than that previously reported.[38] This observed discrepancy in the A–P SC angle could potentially be attributed to differences in modeling methodology. Specifically, the anterior and posterior semicircular canals share a common crus, and the centerline of the bony common crus reconstructed via micro-CT may not precisely coincide with the centerline of the soft tissue common crus reconstructed from MRI data, thus contributing to measurement variation. Although our preliminary study used only a single ex vivo temporal bone specimen, which could introduce potential anatomical variations, the overall consistency of our other measured parameters with those in the existing literature suggests that methodological differences in 3D reconstruction and angular measurement are a more plausible explanation for the observed deviation. Collectively, our findings, along with those of previous studies, suggest that mice and minipigs exhibit a higher degree of human anatomical similarity in semicircular canal angular parameters than guinea pigs. Consequently, owing to their economic advantages and high reproductive rates, mice may serve as suitable models for fundamental research on semicircular canal function and pathology. Conversely, for surgery-related investigations, where the anatomical scale is paramount, minipigs, with more human-like inner ear dimensions, could offer a superior translational model.
However, the observed similarity in cochlear volume and length, as well as semicircular canal angles, cannot be directly extrapolated to imply comparable characteristics of the membranous labyrinth (such as the organ of Corti, utricle, and saccule).[39] Consequently, the overall utility of minipigs and mice for comprehensive investigations of inner ear pathophysiology remains to be fully established. Nevertheless, our current study provides a centerline-based 3D reconstruction and calculation method, as well as a foundational framework for future research on inner ear mechanics.
Our study has some limitations. First, despite the relatively high degree of automation offered by the centerline method, manual identification and delineation of key anatomical landmarks, such as the round window niche and common crus, are still necessary. Second, our study used micro-CT scanning of ex vivo inner ear specimens from various species. This approach allowed for direct scanning, thus circumventing potential measurement inaccuracies caused by tissue shrinkage during post-processing (which can lead to up to 3% shrinkage in bone scaning[8,40]. However, micro-CT technology has some limitations. Specifically, the coil size restricts its application to large specimens, precluding in vivo scanning of living organisms. Furthermore, current image post-processing capabilities cannot guarantee a consistent small voxel size (9 μm in our study) across all analyzed specimens. Third, our study was conducted using a relatively small sample size, which consequently provided only preliminary data. However, this limitation will be addressed and rectified in subsequent inner ear mechanics studies once an optimal animal model is established.
CONCLUSION
In summary, we developed a highly semi-automated quantification method for the mammalian inner ear using a micro-CT-based centerline method. Our comparative morphometric analysis provides preliminary evidence that minipigs are promising candidate models for inner ear surgical experimental studies, owing to their inner ear length and volume being comparable to those of humans. Conversely, mice exhibit semicircular canal angles similar to those in humans, suggesting that they could be superior animal models for research on balance perception. However, it is crucial to acknowledge that the morphological similarity of the bony labyrinth alone is not a sufficient basis to fully establish the overall translational relevance of animal models to human inner ear research. Nevertheless, our study establishes a foundation for subsequent detailed similarity investigations and inner ear mechanics.
Availability of Data and Materials
All relevant data are included within the article. Any additional materials (raw recordings, analysis scripts, or large imaging datasets) are available from the corresponding author upon reasonable request and in accordance with the journal’s data-sharing policy.
Author Contributions
Lei Shi and Yi Du contributed equally to this work. Lei Shi: Methodology, Investigation, Formal analysis, and Writing—original draft. Yi Du: Visualization, Validation, and Investigation. Cong Wei: Supervision. Wenjie Huang: Formal analysis. Handai Qin: Visualization. Chen Liu: Methodology. Da Liu: Visualization. Shuolong Yuan: Conceptualization. Qingqing Jiang: Resources. Jun Yang: Supervision. Gang Li: Methodology. Honglan Xie: Validation. Jianguo Zhao: Investigation. Ning Yu: Writing—review & editing, Conceptualization, and Funding acquisition. Shiming Yang: Writing—review & editing, Project administration, Funding acquisition, and Conceptualization. All authors read and approved the final version.
Ethics Approval and Consent to Participate
This animal study was approved by the Ethics Committee of the Chinese PLA General Hospital (No. S2020-465-01). Human specimen collection was approved by the Ethics Committee of the Chinese PLA General Hospital (No. S2020-466-10).
Conflicts of Interest
The authors declare that they have no competing financial interests. This work was previously shared as a preprint on BIORXIV (DOI: https://doi.org/10.1101/2020.09.30.317941; available at: https://www.biorxiv.org/content/10.1101/2020.09.30.317941v2). The manuscript has been updated from the preprint.
Acknowledgment
The authors would like to thank the staff at the imaging facility for their technical assistance with the Micro-CT scanning and our colleagues for their valuable suggestions on the data analysis.
Funding Statement
This study was supported by the State Key Laboratory of Hearing and Balance Science, the National Key Research and Development Program of China (No. 2020YFC2004002), and the Noise Research Fund (No. BWS14B080 and JDZYY20132).
REFERENCES
- 1.Christensen GE, He J, Dill JA, Rubinstein JT, Vannier MW, Wang G. Automatic measurement of the labyrinth using image registration and a deformable inner ear atlas. Acad Radiol. 2003;10:988–999. doi: 10.1016/s1076-6332(03)00121-1. [DOI] [PubMed] [Google Scholar]
- 2.Du Y, Ren LL, Jiang QQ, Liu XJ, Ji F, Zhang Y, et al. Degeneration of saccular hair cells caused by MITF gene mutation. Neural Dev. 2019;14:1. doi: 10.1186/s13064-019-0126-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Skrzat J, Wróbel A, Walocha J. A preliminary study of three-dimensional reconstruction of the human osseous labyrinth from micro-computed tomography scans. Folia Morphol (Warsz) 2013;72:17–21. doi: 10.5603/fm.2013.0003. [DOI] [PubMed] [Google Scholar]
- 4.Avci E, Nauwelaers T, Lenarz T, Hamacher V, Kral A. Variations in microanatomy of the human cochlea. J Comp Neurol. 2014;522:3245–3261. doi: 10.1002/cne.23594. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Rau C, Hwang M, Lee WK, Richter CP. Quantitative X-ray tomography of the mouse cochlea. PLoS One. 2012;7:e33568. doi: 10.1371/journal.pone.0033568. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Müller M, von Hünerbein K, Hoidis S, Smolders JW. A physiological place-frequency map of the cochlea in the CBA/J mouse. Hear Res. 2005;202:63–73. doi: 10.1016/j.heares.2004.08.011. [DOI] [PubMed] [Google Scholar]
- 7.Melhem ER, Shakir H, Bakthavachalam S, MacDonald CB, Gira J, Caruthers SD, et al. Inner ear volumetric measurements using high-resolution 3D T2-weighted fast spin-echo MR imaging: initial experience in healthy subjects. Am J Neuroradiol. 1998;19:1819–1822. [PMC free article] [PubMed] [Google Scholar]
- 8.Buytaert JA, Johnson SB, Dierick M, Salih WH, Santi PA. MicroCT versus sTSLIM 3D imaging of the mouse cochlea. J Histochem Cytochem. 2013;61:382–395. doi: 10.1369/0022155413478613. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Shin KJ, Lee JY, Kim JN, Yoo JY, Shin C, Song WC, et al. Quantitative analysis of the cochlea using three-dimensional reconstruction based on microcomputed tomographic images. Anat Rec (Hoboken. 2013;296:1083–1088. doi: 10.1002/ar.22714. [DOI] [PubMed] [Google Scholar]
- 10.Buytaert JA, Salih WH, Dierick M, Jacobs P, Dirckx JJ. Realistic 3D computer model of the gerbil middle ear, featuring accurate morphology of bone and soft tissue structures. J Assoc Res Otolaryngol. 2011;12:681–696. doi: 10.1007/s10162-011-0281-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Poznyakovskiy AA, Zahnert T, Kalaidzidis Y, Schmidt R, Fischer B, Baumgart J, et al. The creation of geometric three-dimensional models of the inner ear based on micro computer tomography data. Hear Res. 2008;243:95–104. doi: 10.1016/j.heares.2008.06.008. [DOI] [PubMed] [Google Scholar]
- 12.Poznyakovskiy AA, Zahnert T, Kalaidzidis Y, Lazurashvili N, Schmidt R, Hardtke HJ, et al. A segmentation method to obtain a complete geometry model of the hearing organ. Hear Res. 2011;282:25–34. doi: 10.1016/j.heares.2011.06.009. [DOI] [PubMed] [Google Scholar]
- 13.Honda K, Noguchi Y, Kawashima Y, Takahashi M, Nishio A, Kitamura K. Ex vivo visualization of the mouse otoconial layer compared with micro-computed tomography. Otol Neurotol. . 2015;36:311–317. doi: 10.1097/MAO.0000000000000376. [DOI] [PubMed] [Google Scholar]
- 14.Lee CF, Li GJ, Wan SY, Lee WJ, Tzen KY, Chen CH, et al. Registration of micro-computed tomography and histological images of the guinea pig cochlea to construct an ear model using an iterative closest point algorithm. Ann Biomed Eng. 2010;38:1719–1727. doi: 10.1007/s10439-010-9961-1. [DOI] [PubMed] [Google Scholar]
- 15.Escudé B, James C, Deguine O, Cochard N, Eter E, Fraysse B. The size of the cochlea and predictions of insertion depth angles for cochlear implant electrodes. Audiol Neurootol. 2006;11:27–33. doi: 10.1159/000095611. [DOI] [PubMed] [Google Scholar]
- 16.Zhao F, Koike T, Wang J, Sienz H, Meredith R. Finite element analysis of the middle ear transfer functions and related pathologies. Med Eng Phys. 2009;31:907–916. doi: 10.1016/j.medengphy.2009.06.009. [DOI] [PubMed] [Google Scholar]
- 17.Ifediba MA, Rajguru SM, Hullar TE, Rabbitt RD. The role of 3-canal biomechanics in angular motion transduction by the human vestibular labyrinth. Ann Biomed Eng. 2007;35:1247–1263. doi: 10.1007/s10439-007-9277-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Lesage D, Angelini ED, Bloch I, Funka-Lea G. A review of 3D vessel lumen segmentation techniques: models, features and extraction schemes. Med Image Anal. 2009;13:819–845. doi: 10.1016/j.media.2009.07.011. [DOI] [PubMed] [Google Scholar]
- 19.Poznyakovskiy AA, Mainka A, Platzek I, Mürbe D. A fast semiautomatic algorithm for centerline-based vocal tract segmentation. Biomed Res Int. 2015;2015:906356. doi: 10.1155/2015/906356. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Mainka A, Poznyakovskiy A, Platzek I, Fleischer M, Sundberg J, Mürbe D. Lower vocal tract morphologic adjustments are relevant for voice timbre in singing. PLoS One. 2015;10:e0132241. doi: 10.1371/journal.pone.0132241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Zhao F, Sun F, Hou Y, Chen Y, Chen D, Cao X, et al. A monocentric centerline extraction method for ring-like blood vessels. Med Biol Eng Comput. 2018;56:695–707. doi: 10.1007/s11517-017-1717-8. [DOI] [PubMed] [Google Scholar]
- 22.Sironi A, Turetken E, Lepetit V, Fua P. Multiscale Centerline Detection. IEEE Trans Pattern Anal Mach Intell. 2016;38:1327–1341. doi: 10.1109/TPAMI.2015.2462363. [DOI] [PubMed] [Google Scholar]
- 23.Zhao C, Duan Y, Yang D. Contour interpolation by deep learning approach. J Med Imaging (Bellingham) 2022;9:064003. doi: 10.1117/1.JMI.9.6.064003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Bradshaw AP, Curthoys IS, Todd MJ, Magnussen JS, Taubman DS, Aw ST, et al. A mathematical model of human semicircular canal geometry: a new basis for interpreting vestibular physiology. J Assoc Res Otolaryngol. 2010;11:145–159. doi: 10.1007/s10162-009-0195-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Anton H. Elementary Linear Algebra: Applications Version. 11th ed. Hoboken(NJ): John Wiley & Sons; 2010. [Google Scholar]
- 26.Ravi P, Chepelev LL, Stichweh GV, Jones BS, Rybicki FJ. Medical 3D printing dimensional accuracy for multi-pathological anatomical models 3D printed using material extrusion. J Digit Imaging. 2022;35:613–622. doi: 10.1007/s10278-022-00614-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Okumura S, Goudo M, Hiwa S, Yasuda T, Kitae H, Yasuda Y, et al. Demarcation line determination for diagnosis of gastric cancer disease range using unsupervised machine learning in magnifying narrow-band imaging. Diagnostics (Basel) 2022;12:2491. doi: 10.3390/diagnostics12102491. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Yi HJ, Guo W, Wu N, Li JN, Liu HZ, Ren LL, et al. The temporal bone microdissection of miniature pigs as a useful large animal model for otologic research. Acta Otolaryngol. 2014;134:26–33. doi: 10.3109/00016489.2013.835866. [DOI] [PubMed] [Google Scholar]
- 29.Sharma V, Das KN, Jangra A, Tiwari S, Khera P, Soni K, et al. Dependability of electrode to modiolus distance in patients specific electrode selection: A cadaveric model study. Laryngoscope. 2024;134:4736–4744. doi: 10.1002/lary.31565. [DOI] [PubMed] [Google Scholar]
- 30.Alnafjan FF, Allan SM, McMahon CM, da Cruz MJ. Assessing cochlear length using cone beam computed tomography in adults with cochlear implants. Otol Neurotol. 2018;39:e757–e764. doi: 10.1097/MAO.0000000000001934. [DOI] [PubMed] [Google Scholar]
- 31.Keiler S, Richter CP. Cochlear dimensions obtained in hemicochleae of four different strains of mice: CBA/CaJ, 129/CD1, 129/SvEv and C57BL/6J. Hear Res. 2001;162:91–104. doi: 10.1016/s0378-5955(01)00374-4. [DOI] [PubMed] [Google Scholar]
- 32.Burda H, Branis M. Postnatal development of the organ of Corti in the wild house mouse, laboratory mouse, and their hybrid. Hear Res. 1988;36:97–105. doi: 10.1016/0378-5955(88)90140-2. [DOI] [PubMed] [Google Scholar]
- 33.Salt AN, Henson MM, Gewalt SL, Keating AW, DeMott JE, Henson OW., Jr. Detection and quantification of endolymphatic hydrops in the guinea pig cochlea by magnetic resonance microscopy. Hear Res. 1995;88:79–86. doi: 10.1016/0378-5955(95)00103-b. [DOI] [PubMed] [Google Scholar]
- 34.Yildiz E, Gerlitz M, Gadenstaetter AJ, Landegger LD, Nieratschker M, Schum D, et al. Single-incision cochlear implantation and hearing evaluation in piglets and minipigs. Hear Res. 2022;426:108644. doi: 10.1016/j.heares.2022.108644. [DOI] [PubMed] [Google Scholar]
- 35.Jun B, Song S. Analysis of cochlear morphology for cochlear implantation using three-dimensional reconstruction of computed tomography images. Audiol Neurootol. 2024;29:207–215. doi: 10.1159/000534739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Hashimoto S, Naganuma H, Tokumasu K, Itoh A, Okamoto M. Three-dimensional reconstruction of the human semicircular canals and measurement of each membranous canal plane defined by Reid’s stereotactic coordinates. Ann Otol Rhinol Laryngol. 2005;114:934–938. doi: 10.1177/000348940511401207. [DOI] [PubMed] [Google Scholar]
- 37.Aoki S, Takei Y, Suzuki K, Masukawa A, Arai Y. Planer orientation of the bilateral semicircular canals in dizzy patients. Auris Nasus Larynx. 2012;39:451–454. doi: 10.1016/j.anl.2011.07.020. [DOI] [PubMed] [Google Scholar]
- 38.Liu W, Chen G, Xie J, Liang T, Zhang C, Liao X, et al. A new coordinate system for magnetic resonance imaging of the vestibular system. Front Neurol. 2021;12:789887. doi: 10.3389/fneur.2021.789887. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Iversen MM, Rabbitt RD. Wave mechanics of the vestibular semicircular canals. Biophys J . 2017;113:1133–1149. doi: 10.1016/j.bpj.2017.08.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Wu C, Yang L, Hua C, Wang K, Zhang T, Dai P. Geometrical and volume changes of the membranous vestibular labyrinth in guinea pigs with endolymphatic hydrops. ORL J Otorhinolaryngol Relat Spec. 2013;75:108–116. doi: 10.1159/000350569. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All relevant data are included within the article. Any additional materials (raw recordings, analysis scripts, or large imaging datasets) are available from the corresponding author upon reasonable request and in accordance with the journal’s data-sharing policy.