Activation Region Overlap Visualization for Image-Guided Cochlear Implant Programming

Abstract

Objective:

The cochlear implant is a neural prosthesis designed to directly stimulate auditory nerve fibers to induce the sensation of hearing in those experiencing severe-to-profound hearing loss. After surgical implantation, audiologists program the implant’s external processor with settings intended to produce optimal hearing outcomes. The likelihood of achieving optimal outcomes increases when audiologists have access to tools that objectively present information related to the patient’s own anatomy and surgical outcomes. This includes visualizations like the one presented here, termed the activation region overlap image, which is designed to decrease subjectivity when determining amounts of overlapping stimulation between implant electrodes.

Approach:

This visualization uses estimates of electric field strength to indicate spread of neural excitation due to each electrode. Unlike prior visualizations, this method explicitly defines regions of nerves receiving substantial stimulation from each electrode to help clinicians assess the presence of significant overlapping stimulation. A multi-reviewer study compared this and an existing technique on the consistency, efficiency, and optimality of plans generated from each method. Statistical significance was evaluated using the two-sided Wilcoxon rank sum test.

Main results:

The study showed statistically significant improvements in consistency (p < 10⁻¹²), efficiency (p < 10⁻¹⁵), and optimality (p < 10⁻⁵) when generating plans using the proposed method versus the existing method.

Significance:

This visualization addresses subjectivity in assessing overlapping stimulation between implant electrodes, which currently relies on reviewer estimates. The results of the evaluation indicate the provision of such objective information during programming sessions would likely benefit clinicians in making programming decisions.

1. Introduction

Over the last three decades, the cochlear implant (CI) has become the standard-of-care intervention for individuals with severe-to-profound sensorineural hearing loss (SNHL), with an estimated 736,900 devices implanted worldwide as of 2019, including approximately 118,100 and 65,000 in American adults and children, respectively (National Institute on Deafness and Other Communication Disorders (NIDCD), 2021). SNHL is attributed to loss of function in the inner ear or cochlear nerve, resulting in insufficient stimulation of the auditory nerve fibers (ANFs) composing the cochlear nerve housed within the modiolus (shown in green in figure 1(a)) to produce the typical range of human hearing. In unaided acoustic hearing, vibrations traveling through the outer and middle ear stimulate hair cells in the inner ear that activate ANFs to produce the sensation of sound (NIDCD, 2015). To address SNHL, the CI bypasses the mechanisms of acoustic hearing by directly stimulating ANFs using an electrode array surgically inserted into the cochlea (NIDCD, 2021). This array is connected to internal protective circuitry and a receiver/stimulator with an RF coil placed between the scalp and skull, which is then magnetically connected to a transmitting coil attached to the removable sound processor worn on the exterior of the scalp or on the ear, similar to a behind-the-ear hearing aid (Naples & Ruckenstein, 2020). Sounds detected by a microphone in the processor are decomposed into their frequency components, which the processor uses to determine how power is distributed to the electrodes of the array according to a set of processor instructions, also referred to as a patient’s map. An audiologist seeks to optimize a patient’s hearing outcomes by modifying map parameters including but not limited to the set of activated electrodes, the electrical dynamic range of each electrode, and the frequency range assigned to each electrode (Shapiro & Bradham, 2012).

Figure 1.

(a) A 3-D representation of the cochlea (red) and the inserted electrode array (gray). The modiolus containing the spiral ganglion (SG) cells is shown in green. (b) Estimated spread of excitation for each electrode in the array shown against the modiolus. Electrodes are numbered 1–12 from most apical to most basal.

The differences in neural activation in acoustic versus electrically-induced hearing can pose significant challenges for programming CIs. The auditory nerve is composed of approximately 30,000 ANFs, each of which has an associated characteristic frequency, typically in the range of 20 Hz to 20 kHz. The ANFs are arranged tonotopically along the length of the cochlea, meaning different regions of the cochlea are associated with different frequencies of sound. The highest characteristic frequencies are associated with ANFs at the entrance, or basal end, of the cochlea, and the lowest are associated with the deepest, or most apical, nerve sites (see figure 1(a)). Each ANF is activated by hair cell stimulation only when the acoustic signal contains energy at the highly selective characteristic frequency associated with the ANF. However, CIs provide far fewer discrete frequency channels than in acoustic hearing from which audiologists must reconstruct the range of human hearing. Implants from FDA-approved CI manufacturers have 12, 16, 20, or 22 electrodes (MED-EL, n.d.; Advanced Bionics, n.d.; Guevara et al., 2010; Cochlear Limited, n.d.). Stimulation strategies also primarily use monopolar stimulation, in which current is supplied to a single contact and returned through a distant electrode, resulting in broad current spread. Because of these factors, each electrode stimulates a broad range of overlapping ANFs, and thus individual ANFs are recruited by multiple electrodes. Since each electrode is assigned a different frequency band, each ANF can be recruited for multiple frequencies bands rather than just its characteristic frequency. This overlapping stimulation can result in spectral smearing artifacts and channel interaction artifacts, where recruitment by an electrode is not possible due to recent stimulation by another electrode due to refractoriness of the ANFs, both of which negatively impact audiologic outcomes (Baer & Moore, 1994).

Multiple studies examining the relationship between speech recognition scores and the number of active channels in monopolar stimulation strategies show that many CI users receive only marginal benefits from using more than 8 active channels (Fishman et al., 1997; Friesen et al., 2001; Shannon et al., 2011; Berg et al., 2019), with the most likely cause being channel interaction due to distance between the electrodes and the nerves they stimulate. For monopolar stimulation, the method of increasing channel independence, in general, is to deactivate certain channels to improve independence in the remaining channels, rather than attempting to increase the number of independent channels. Other strategies, however, attempt to increase the number of independent channels by shaping intracochlear electric fields through simultaneous stimulation of two or more electrodes. Two such strategies are current steering and current focusing. Current steering strategies stimulate two electrodes simultaneously to shift the peak of stimulation to a location between the contacts, creating “virtual channels” beyond those provided by the physical electrode array. Alternatively, current focusing uses multipolar electrode configurations, such as tripolar (TP) and partial tripolar (pTP) stimulation, to limit the spread of excitation due to a stimulus, thereby increasing the independence of the existing channels. Both TP and pTP strategies activate three electrodes simultaneously: one primary electrode and its two adjacent electrodes. The two adjacent electrodes receive stimuli that are of opposite polarity and a fraction of the amplitude of the stimulus delivered to the primary electrode. For basic TP, the two contacts receive stimuli with half the amplitude of the primary stimulus, while pTP allows the amplitude of the opposing stimuli to be further reduced. However, both current steering and focusing have their own limitations. In general, the multipolar configurations used in current focusing increase power consumption and require significant increases in stimulus levels to attain adequate loudness, which often becomes impossible within the constraints of compliance and safety regulations (Berenstein et al., 2008; Bonham & Litvak, 2008). Current steering, on the other hand, does not address the issue of broad stimulation and can even exacerbate the issue of channel interactions due to the additive nature of electric fields produced by simultaneous stimulation (Reynolds & Gifford, 2019). For these reasons, as well as the predominance of monopolar stimulation over multipolar configurations, we have opted to focus solely on the former in this study, although we are planning to explore extensions of this work that implement these alternative strategies in the future.

To determine if adjacent electrodes have excessive overlapping stimulation, we can first define the nerve site most likely to be stimulated by an electrode as its peak activation region, a subset of the larger activation region, which consists of all nerve sites likely to receive significant stimulation from that electrode, i.e., its spread of excitation. Figure 1(b) uses colored lines to demonstrate estimates of each electrode’s activation region. If we define an electrode’s activation region as the region containing ANFs that should only be substantially activated by that electrode, then one approach to define when deleterious channel interaction occurs is to detect any case in which the peak activation region of one electrode falls within the activation region of another electrode.

A seemingly obvious solution to the issue of channel interaction is to eliminate overlap by decreasing the spread of excitation for each electrode. However, modifying the width of current spread is complicated by the relationship between the amount of current required for adequate stimulation and the distance between an electrode and the modiolus. As the distance between an electrode and the modiolus increases, more current is needed to achieve the same comfortable levels of perception in that region as a closer electrode. The width of the spread of excitation is directly related to the amount of current delivered to the electrode, so greater current results in a wider activation region (Cohen, 2009; Davis et al., 2016; Saunders et al., 2002). Therefore, significantly reducing channel interaction is not as simple as reducing current levels until there is no overlapping stimulation, as doing so could also have significant impacts on a patient’s hearing.

An alternative approach aimed at increasing channel independence involves deactivation of at least one electrode in a set of electrodes with significant overlap in stimulation. To accurately determine these sets of electrodes, an audiologist would need to know the location of each electrode in the array. Without the resources to obtain this information, an audiologist must assume optimal placement of the array, despite the reality that with current surgical techniques, the array is inserted blindly, often resulting in sub-optimal placement (Holden et al., 2013; Chakravorti et al., 2019). Because prior research supports the conclusion that many aspects of array location significantly impact hearing outcomes (Holden et al., 2013; Chakravorti et al., 2019; Finley & Skinner, 2008; Aschendorff et al., 2005; Wanna et al., 2014; Wanna et al., 2011; O’Connell et al., 2016), it is imperative to develop tools that can provide this information to clinicians.

One solution is to create three-dimensional spatial models of a patient’s implanted cochlea. This process typically uses pre- and post-operative imaging to construct models of the cochlear anatomy and implanted electrode array, although the exact implementation may vary. The discrepancy between the small size of the cochlea and the limited resolution of in vivo imaging methods, such as conventional computed tomography (CT), often prohibits direct creation of these models from patient images. Instead, many methods first create a base model using manual or semiautomatic segmentation methods to obtain the cochlear structures from images of cadaver cochleae acquired using high-resolution modalities that cannot be applied in vivo, such as micro-CT (μCT) (Dang et al., 2015; Mangado et al., 2018; Noble et al., 2013; Bai et al., 2019) and photomicrography (Rattay et al., 2001; Nogueira et al., 2015; Kalkman et al., 2015). These models are then automatically or semiautomatically adapted to a patient CT using various methods, including using statistical shape models (SSMs) (Mangado et al., 2018; Noble et al., 2012), and fitting the base model to parameters defined from patient images (Nogueira et al., 2015). Electrode array models are typically created from manufacturer specifications for the specific implant and are placed within the anatomical model using segmentation techniques applied to post-operative images (Dang et al., 2015; Noble & Dawant, 2015; Nogueira et al., 2015; Zhao et al., 2016; Zhao et al., 2019).

In this research, we use a subset of the above methods to create an automated pipeline for image-guided CI programming (IGCIP) that produces segmentations with sub-voxel accuracy (Noble et al., 2012; Zhao et al., 2019). We first obtain the segmentation of a patient’s cochlear anatomy by fitting a high-resolution SSM created from μCT images of cadaver cochleae to the patient’s pre-operative CT (Noble et al., 2013; Noble et al., 2012). The electrode array is then localized using an automatic graph-based method (Zhao et al., 2019; Noble & Dawant, 2015). Examples of these segmentations are shown in figure 1. These techniques enable audiologists to evaluate multiple spatial features, including distance of the array to the modiolus, insertion depth of the array, and any undesirable placements, e.g., translocation from the scala tympani to the scala vestibuli or extracochlear electrodes.

We use the spatial information provided by the IGCIP approach to estimate an electrode’s spread of excitation as the distance from that electrode to nerve sites. The original IGCIP work proposes visualizing spread of excitation in a format termed distance-vs.-frequency, or DVF, curves (Noble et al., 2013), an example of which is shown in figure 2. In a set of DVF curves, each curve corresponds to one electrode in the array, where each point along a curve represents the distance from the associated electrode to different nerve sites. The vertical axis indicates this distance in millimeters, and the horizontal axis spans the frequency range associated with the nerve sites of the cochlea, displayed on a log scale. This plot shows the place frequencies of the nerve sites most likely to be stimulated by an electrode, and by extension, the nerve sites most likely to receive overlapping stimulation from multiple electrodes. When using the DVF curve format, the peak activation region is easy to identify, as it is assumed to be the closest nerve site to an electrode, i.e., the minimum of that electrode’s DVF curve. However, it is more difficult to precisely identify the entirety of the activation region because its width is not static across the array, with more distant electrodes typically having wider activation regions than closer electrodes. Since it is not quantitatively estimated and displayed, it is also subject to the reviewer’s estimation. Learning to read DVF curves and utilizing them to accurately and consistently select optimal electrode array deactivation plans can be difficult and time-consuming. In this work, we propose an alternative to the DVF curves that removes much of this subjectivity by visualizing channel overlap using a model of estimated electric field strength. We have termed this new visualization the activation region overlap (ARO) image because it permits visualizing the relationship between the activation regions of each electrode in the array. In an evaluation of the DVF curves and the ARO image, our results will show that plans generated using the ARO image were created more consistently and rated optimal more frequently than those generated using the DVF curves, demonstrating the utility of our proposed approach.

Figure 2.

An example of DVF curves for a 12-electrode array, showing one activation region and one peak activation region. Activated electrodes are indicated with a solid blue line, and deactivated electrodes are indicated with a dotted gray line.

2. Methods

2.1. Defining the Activation Region

Research on electro-anatomical models (EAMs) of the implanted cochlea has shown modeling a CI electrode as a simple point charge in a homogeneous medium yields similar results to more sophisticated finite-element models with assumed tissue resistivity values falling within the range of known tissue resistivity variability (Rattay et al., 2001). Therefore, a simple model we propose is to use a point charge model to estimate the strength of an electric field due to an electrode $\overrightarrow{c}$ at a specified nerve site$\overrightarrow{x}$, which can be calculated according to Coulomb’s law. The field strength at this nerve site, $\left|E\left(\overrightarrow{x}\right)\right|$, is inversely proportional to the squared distance between the nerve site and the electrode, i.e.,

$$\left|E\left(\overrightarrow{x}\right)\right|\propto \frac{1}{{‖\overrightarrow{x}-\overrightarrow{c}‖}^2}.$$ (1)

According to this model, an electrode close to a nerve site requires a relatively small amount of current with a relatively small spread of excitation to activate that site compared to a more distant electrode. The much larger spread of excitation associated with a distant electrode may result in channel interactions between this and other nearby electrodes that are also distant to nerve sites. We can also see the field strength will be greatest at an electrode’s peak activation region. Using this model, we can define the activation region of an electrode as the set of nerve sites that satisfy

$$\frac{\left|E\left(\overrightarrow{x}\right)\right|}{\left|E\left(PAR\right)\right|}=\frac{{‖PAR-\overrightarrow{c}‖}^2}{{‖\overrightarrow{x}-\overrightarrow{c}‖}^2}>\tau .$$ (2)

Using this relationship, a nerve site is in an electrode’s activation region if the electric field strength at that site exceeds some fraction τ of the field strength at the peak activation region, i.e., the ratio of the field strength at the nerve site to that at the peak activation region exceeds some threshold τ. Large values for τ indicate a greater tolerance for overlapping stimulation between electrodes, producing a narrower activation region, while small values indicate less tolerance, resulting in a larger activation region. In this paper, we use a default value of τ = 0.5, as this value produces similar rates of activation as those reported in studies of the relationship between electrode-to-modiolus distance and the number of effectively independent channels in a CI (Fishman et al., 1997; Friesen et al., 2001; Shannon et al., 2011; Berg et al., 2019). Because it is distance-based, this technique can be used to estimate the activation region directly from the DVF curves, but there is currently no visual indication of this region on each curve. Thus, with DVF curves alone, identification of the activation region is subject to a reviewer’s estimation, which may be inconsistently defined across reviewers and across multiple viewings of the curves by the same reviewer over time.

2.2. The Activation Region Overlap Image

To overcome the limitations of the DVF curves, we propose a more direct visualization of the activation region using the ARO image, an example of which is given in figure 3. The vertical axis represents each electrode of the array, with each horizontal, colored bar associated with one electrode. The most apical electrode is always located at the bottom of the image, and the most basal electrode is always located at the top. The horizontal axis is the same as the log-scale representation of the characteristic frequencies of the ANFs used in the DVF curve visualization. The ARO image uses a similar GUI to other 2D visualization approaches for intra-cochlear electrode location, which show electrode number vs. depth within the cochlea on orthogonal axes, such as that proposed by Ceresa et al. (2014), but with different content, given our use of clinically relevant programming metrics.

Figure 3.

Example ARO image, showing the same case as figure 2 with a modified deactivation plan. Green: activated electrodes. Gray: deactivated electrodes. Orange: activated electrodes with too much overlap with another electrode. Lavender: deactivated electrodes that could be activated without having significant overlap with another electrode.

This visualization implements equation 2 to define the range of frequencies associated with the nerve sites in the activation region of each electrode. The width of the activation region for an electrode is represented by the width of the bar associated with that electrode, as shown in figure 3. The peak activation region of each electrode is indicated using a vertical black line. With this visualization, it is easy to identify when the activation region of one electrode overlaps with the peak activation region of another, an indicator of problematic channel interactions. Additionally, electrodes that either are extracochlear or are intracochlear but very near to the entrance of the cochlea tend to be insufficiently able to stimulate ANFs in a way that improves hearing outcomes and may even cause interference that negatively impacts hearing (Holder et al., 2018). Therefore, in addition to electrodes that do not satisfy the relationship in equation 2, any electrodes with a peak activation region with characteristic frequency greater than 15 kHz are recommended for deactivation.

In addition to explicitly marking the activation region and peak activation region of each electrode, this visualization uses color coding to make violations of our constraints easier to see. The chosen color palette was selected to accommodate users with color blindness. As shown in figure 3, electrodes that violate our constraints, i.e., those with too much overlapping stimulation or whose peak activation region has a place frequency greater than 15 kHz, have bars that are orange in color. When an activated electrode does not violate either of these constraints, the bar for that electrode is green. Deactivated electrodes are indicated with gray bars, unless it could be activated without violating any constraints, in which case its bar is lavender. Finally, we indicate the 15 kHz cut-off frequency with a green line that extends across the entire vertical span of the image.

2.3. User Interface

To create new deactivation plans using the ARO image, the visualization is incorporated into an interactive user interface, shown in figure 3. Changing any of the available options in the interface will automatically trigger the visualization to reassess constraint violations and update the color for each electrode accordingly. The check boxes to the left of the ARO image control which electrodes are activated or deactivated, where a checked box indicates a deactivated electrode. The user can also modify the threshold value used to determine the activation region for each electrode to increase or decrease the tolerance for overlapping stimulation. Finally, if the user wishes to see the frequency range covered by the activated electrodes, they can check the option titled “Collapse Deactivated” at the bottom of the controls on the left to show only activated electrodes.

2.4. Study Methodology

To evaluate the ARO image against the DVF curves, we designed a multi-part study to determine repeatability of plans generated using each method and optimality of plans generated using each method. In this study, we asked two reviewers to evaluate each visualization method over 15 cases generated by processing CT scans acquired from research subjects at our center under IRB approval #090155. Both reviewers are expert audiologists who were not involved in the design of the proposed ARO image. They both were previously familiar with reading DVF curves and received a one-hour training session on using the ARO image.

2.4.1. Experiment 1: Intra-subject variability and time efficiency

In the first part of the study, reviewers were presented with a set of DVF curves and were asked to generate an electrode deactivation plan for the given case. After all cases were evaluated, the reviewer was asked to repeat this plan generation on the same set of cases, presented in a different random order from the first round. The reviewers then completed a third round of the same evaluation, with the cases once again presented in a different random order. After completing the evaluation of the DVF curves, reviewers repeated this three-round evaluation using the ARO image. There was no mandatory waiting period between each round. Each evaluation was timed to assess the speed with which reviewers developed plans. In this experiment, we quantified reviewer consistency using the number of plans that differed across each round, the number of differences in those plans, and the time taken to produce plans. We also recorded the threshold value τ for each case to evaluate the amount of deviation from the default value of 0.5.

To measure the number of differences between plans for a single case, we use a modified version of the hamming distance (Zhao et al., 2016), abbreviated as MHD. This modified version penalizes comparisons of certain configuration patterns less harshly compared to the standard hamming distance. As an example, two plans may both have every other electrode activated, i.e., on-off-on-off, but one plan begins with the first electrode activated while the second has the first deactivated. The standard hamming distance would be large in this example, despite the plans likely having highly similar stimulation patterns. Instead, the MHD assigns greater values to plans with more distant mismatches in electrode activation status, which likely correspond to plans with greater variations in stimulation patterns. Two examples of calculating the MHD are shown in figure 4. When calculating the MHD, if the activation of an electrode k in the first plan does not match that of the corresponding electrode in the second plan, the distance is reported as the number of electrodes between position k and the nearest electrode in the second plan whose activation status matches that of electrode k. This produces an array of distances, from which we can identify and sum the local maxima to get the value for the MHD. To account for the varying number of electrodes in implants, we then normalize the MHD by dividing by the number of electrodes in the array for that case.

Figure 4.

Two examples of the modified hamming distance metric detailed in (Zhao et al., 2016). “A” indicates an activated electrode, and “D” indicates a deactivated electrode. Distances are calculated as the difference from plan P₁ or P₃ to plan P₂, where P₂ is the same in both examples. Using the MHD metric, P₃ is assigned a larger distance than P₁ when comparing both to P₂.

2.4.2. Experiment 2: Plan optimality

In the second part of the study, reviewers were shown deactivation plans one at a time and asked to judge each as optimal or suboptimal, where rating a plan as “optimal” indicates the reviewer would not adjust anything about the plan. Each plan was visualized as both a set of DVF curves and as an ARO image, displayed side-by-side using the interface shown in figure 5. The reviewer’s evaluation was based solely on the information provided by the two visualizations. However, the reviewer did have the option to adjust the threshold value for individual plans if they felt the default value of 0.5 had either too much or too little tolerance for overlap, based solely on the information provided in the visualizations. Changes to this threshold were recorded during the experiment.

Figure 5.

An example of the GUI for experiment 2, showing one of the randomly selected plans for the same case shown in the previous figures. The same deactivation plan is shown side-by-side on both the DVF curves and the ARO image. The reviewer can manipulate the threshold value to control the tolerance for stimulation overlap.

In this portion of the study, each reviewer evaluated three plans for each of the 15 cases, for a total of 45 evaluations. Thirty of the forty-five plans were drawn from those created in the previous experiment. For each case, we randomly selected one of the three plans a reviewer created using the DVF curves, giving fifteen plans created using the DVF curves. We repeated the same process to obtain fifteen plans created using the ARO images. The fifteen remaining plans were control plans created by a third expert DVF curve reviewer. This reviewer was asked to create plans with subtle sub-optimalities in the deactivation patterns with variations such as having an extra electrode deactivated that could be activated while introducing tolerable channel overlap or containing a region of active electrodes with slightly more than tolerable channel interaction that could be alleviated by deactivation of an electrode. The inclusion of this plan evaluates a reviewer’s bias toward accepting all plans. For example, if a reviewer accepts a large number of control plans, that reviewer likely has a bias toward accepting all plans. These 45 plans were presented one at a time in random order to the reviewers, with the origin of the plan masked. It should be noted that, with the exception of the control plans, each reviewer was presented only with plans they had created, not those of the other reviewer, meaning the two reviewers did not necessarily evaluate the same deactivation plans for each case.

3. Results

A summary of the results from our first experiment for the DVF curves and the ARO image is shown in table 1. The number of differences between plans for a single case is reported in terms of the normalized MHD introduced in the previous section. From these preliminary results, we see that plan selection using the ARO image is more consistent across cases, and when differences do occur, the number of differences between two plans is lower compared to the DVF curves. Additionally, the median time taken to generate a plan is lower for the ARO image than that for the DVF curves. Two-sided Wilcoxon rank sum tests indicated the differences in the time taken to generate a plan and the consistency of the plans created for a case when using the DVF curves versus the ARO image were both highly statistically significant (p = 2.9 × 10⁻¹⁶ and p = 8.4 × 10⁻¹⁴, respectively). While it is possible the reviewers might have remembered their selections from previous rounds in the 3 round plan creation process across the 15 cases, the randomized ordering of presented plans in each round was implemented to alleviate this effect, and the reviewers did not report perceiving repeating plans by memory. The data in table 1 support this as the mean normalized MHD for the DVF curve planning was relatively high for both reviewers. We found an average value of τ = 0.534 was selected across both reviewers.

Table 1.

A summary of results for the plans generated using DVF curves and the ARO image in part 1 of the study.

	Reviewer	# of Varied Plans	Mean Normalized MHD	Median Time (s)
DVF	1 2	29 21	0.156 0.170	99 51
ARO	1 2	3 6	0.009 0.0133	31 17

The second part of our study was evaluated on the total number of plans from each method rated as optimal. A summary of these results is given in table 2. We see that ARO image plans are rated as optimal at a greater rate than DVF curve plans, with an acceptance rate of 93.3% and 66.7%, respectively. The acceptance of zero control plans by both reviewers indicates a low likelihood of bias toward accepting all plans. We again used the Wilcoxon rank sum test to assess the accuracy of each method of plan generation versus the others. We found the difference in acceptance rates for the plans created using DVF curves and the plans created using ARO images was statistically significant (p = 1.2 × 10⁻⁶). The differences in acceptance rates for both the DVF curves and the ARO images compared to acceptance rates for the control plans were also statistically significant, with p = 2.0 × 10⁻² and p = 3.8 × 10⁻¹⁰, respectively.

Table 2.

A summary of the results for the evaluation of plans generated by each method in part 2 of the study (Reproduced with permission from Bratu et al., 2021).

Reviewer	DVF Plans Accepted	ARO Plans Accepted	Control Plans Accepted
1	1	10	0
2	4	14	0

4. Conclusion

In this paper, we present a visualization method that utilizes patient-specific spatial information of intracochlear anatomy and electrode array positioning to determine the activation region of an electrode using electric field strength estimates and displays this information in an easy-to-read format. This visualization removes the need to mentally estimate the activation region and peak activation region of each electrode required when using DVF curves, decreasing subjectivity of plan generation. These results indicate that the ARO image outperforms the DVF curves in repeatability, acceptability, and time taken to generate plans. Other works have aimed to develop automatic methods for selecting deactivation plans (Bratu et al., 2020). In a future study, we will evaluate the use of this visualization technique to review automatic plans created using such methods. We are also using this visualization method to explore the effectiveness of our default threshold value of τ = 0.5 for generating deactivation plans and the resulting impact on various hearing assessment scores.