Bone conduction hearing implants were installed in polyurethane artificial bone blocks (13 cm × 8.8 cm × 4 cm) with different densities (Sawbones, USA) (Figure 1A). Two high-density (50 pounds per cubic foot) and one low-density (40 pounds per cubic foot) artificial bone blocks were created. Installation of 4 mm wide implants (Oticon Medical AB, Askim, Sweden) mounted with 14 mm abutments (Oticon Medical AB, Askim, Sweden) were carried out using the surgical instrument designed for MIPS (5). Implants were fully inserted with 4.5 rotations or partially inserted with 3.5 rotations (Figure 1B).

Virtual insertion depth measurements on the CBCT and direct manual caliper measurements were completed for 18 implants placed in three bone blocks (Figure 1C). According to specifications provided by the manufacturer, the distance from the top of the abutment (14 mm) and the implant rim should range between 14.43 and 14.59 mm when fully inserted. The middle (14.51 mm) was used as a comparative reference value for full insertion in the analysis. In 3D Slicer, virtual markings were placed indicating the highest point of the abutment top, the bottom of the implant rim, and bone surface in four quadrants (Figure 1D). Per quadrant, the virtual bone surface to abutment top distance and the virtual bone surface to implant rim distance were calculated. Manual caliper measurements were obtained at every quadrant from the abutment top to the level of the artificial bone and thereafter averaged per implant. The distance of the implant rim to artificial bone was estimated by subtracting the abutment length from the caliper measurements.

The Osstell ISQ (Ostell, Gothenburg, Sweden) was used to measure ISQ values by mounting a Smartpeg Type 55 on the abutments. Two perpendicular ISQ measurements (ISQ horizontal, ISQ vertical) were obtained for all implants (12). To test the influence of abutment length, ISQ values were obtained for each implant mounted consecutively with 9, 12, and 14 mm abutments, resulting in six measurements per implant. Twelve implants were tested for the high-density bone configuration, half of them fully seated and half partially seated. For the low-density bone configuration, six implants were placed, half fully seated, and half partially seated (Figure 1C). For the high-density bone configurations, one fully and one partially seated implant moved during abutment replacement. Consequently, in high-density bone, 10 BCHIs were evaluated (n = 5 fully seated, n = 5 partially seated) (Figure 1C). Due to gross implant mobility during abutment replacement, it was not possible to change the abutment in the low-density artificial bone block that was implanted with partially inserted BCHIs, hence only fully inserted BCHIs were tested (n = 3).

Analyses were used for in vitro validation of CBCT imaging before they could be applied in vivo. For direct caliper measurements and CBCT measurements, mean (M) and SD for full insertion and partial insertion were calculated for abutment top to bone surface distances and implant rim to bone surface distances. Mean distances for abutment top to bone surface of CBCT measurements and caliper measurements for full and partial insertion were compared to each other and to the theoretical average for full insertion. Regression analysis in which the CBCT abutment top to bone surface distance and CBCT implant rim to bone surface distance were regressed to insertion depth (full or partial) was performed. 95% Prediction interval for fully inserted and partially inserted BCHIs were determined. Discriminant analysis with equal prior probabilities was applied to determine the optimal cutoff point for classification of BCHIs as fully or partially inserted.

For ISQ measurements, linear mixed model analyses with implant as random factor and abutment length (9, 12, and 14 mm), artificial bone density (high, low), and insertion depth (full, partial) as fixed factors and a compound symmetry covariance matrix of the residuals. Because in the low-density artificial bone block, only fully inserted BCHIs could be measured, the effect of insertion depth could only be determined at high bone density. Similarly, the effect of bone density could only be determined for fully inserted BCHIs. When statistically significant effects were identified, Bonferroni adjusted post hoc comparisons between the different factor levels were additionally performed.

Direct caliper measurements from abutment top to the bone surface of fully inserted BCHIs (M = 14.62, SD = 0.06, n = 9) were 0.11 mm greater than the theoretical average for full insertion in vitro (Figure 2A). CBCT measurements of fully inserted BCHIs (M = 14.83, SD = 0.08, n = 9) were on average 0.32 mm greater than the theoretical average for full insertion in vitro. For the partially inserted implants, direct caliper measurements (M = 15.06, SD = 0.09, n = 9) and CBCT measurements (M = 15.27, SD = 0.10, n = 9) were both 0.44 mm greater compared to the fully inserted BCHIs.

The implant rim to artificial bone surface distances as measured on CBCT were 0.47 mm larger in partially inserted BCHIs (M = 0.73, SD = 0.07, n = 9) compared to fully inserted BCHIs (M = 0.26, SD = 0.09, n = 9) (Figure 2B).

Regression analyses revealed that the type of artificial bone was a significant predictor for abutment top to bone surface distance (p < 0.015). 95% Prediction intervals for high density, low density, and the combined group are presented in Table 1. For implant rim to bone surface no significant difference between high and low density was found. By means of discriminant analysis, cutoff points were obtained which could potentially be used to classify BCHI’s either as fully or partially inserted (Table 1). Mean abutment top to bone surface distance for fully inserted BCHIs in high-density bone was 14.85 mm (SD = 0.06, n = 6) compared to 14.78 mm (SD = 0.10, n = 3) for fully inserted BCHIs in low-density bone resulting in a difference of 0.07 mm. Mean abutment top to bone surface distance for partially inserted BCHIs in high-density bone was 15.32 mm (SD = 0.09, n = 6) compared to 15.18 mm (SD = 0.05, n = 3) for partially inserted BCHIs in low-density bone resulting in a difference of 0.14 mm.

Evaluation of the in vitro 3D models allowed for a detailed qualitative assessment of the implant, artificial bone, and the implant-artificial bone interface. BCHI insertion and angulation could be evaluated as well in the axial, coronal, or sagittal plane allowing to visually distinguish between partially and fully inserted BCHIs (Figure 3). Softening effects were visible directly under the BCHI in the artificial bone models (Figure 3C).

Implant stability quotient measurements are shown in Figure 4, and the models are presented in Table 2. In a split plot ANOVA analysis, the ISQ in high-density artificial bone revealed a significant negative relationship with abutment length (p < 0.001) and insertion depth (p = 0.001), whereas no interaction between abutment length and insertion depth was found (p = 0.378) (Table 2). To evaluate the effect of bone density, a linear mixed model analysis was performed using the ISQ data of the fully inserted BCHIs. After removing the non-significant interaction (p = 0.153) between abutment length and bone density, abutment length (p < 0.001) and bone density (p = 0.024) were both significant predictors of ISQ (Table 2).

Four subjects were selected 1-week post-surgery at the ENT outpatient department of Maastricht University Medical Center to evaluate BCHI with a clinical indication. Subject characteristics are summarized in Table 3.

Cone beam CT measurements are described in Table 4. Distances between the implant rim and bone surface were consistent with a full insertion (0.49 mm) for subjects 1, 2, and 3. For subject 4, the measurements per quadrant were 1.0, 0.09, 0.07, and 0.37 mm, respectively, resulting in a mean distance of 0.38 mm. The mean distance was consistent with full insertion, but the distance of 1.0 mm for one quadrant was greater than the cutoff value of 0.49 which could indicate an angulated insertion (See 3D models).

We assumed that the abutment top to bone surface cutoff distances for the 14 mm abutments could be adjusted for 9 and 12 mm abutments. The cutoff value for 14 mm abutments (15.05 mm) was therefore adapted for 9 mm abutments (10.05 mm) and 12 mm abutments (13.05 mm). Distances between abutment top and bone were consistent with full insertion in all subjects.

Evaluation of the in vivo 3D models allowed for qualitative assessment of the BCHI, bone, and soft tissue, allowing visual appraisal of implant seating and angulation (Figure 3). In the in vitro 3D models, the complete abutment could be evaluated, while softening effects around the abutment top made it difficult to evaluate the location of the abutment top in the in vivo 3D models. Qualitative evaluation of the post-surgery scans for subjects 1, 2, and 3 revealed a full non-angulated BCHI insertion (Figure 3D). In the postsurgery scan of subject 4, the 3D model revealed an angulated insertion of the BCHI (Figure 3E), which is consistent with the differences observed per quadrant. Unexpectedly, the 3D patient model also facilitated evaluation of the skin next to the implant. In subject 1, the presence of skin sagging, which entails excess skin superior to the abutment, was noticeable (Figure 3D).