Channel Interaction During Infrared Light Stimulation in the Cochlea

Abstract

Background and objectives: The number of perceptually independent channels to encode acoustic information is limited in contemporary cochlear implants (CIs) because of the current spread in the tissue. It has been suggested that neighboring electrodes have to be separated in humans by a distance of more than 2 mm to eliminate significant overlap of the electric current fields and subsequent interaction between the channels. It has also been argued that an increase in the number of independent channels could improve CI user performance in challenging listening environments, such as speech in noise, tonal languages, or music perception. Optical stimulation has been suggested as an alternative modality for neural stimulation because it is spatially selective. This study reports the results of experiments designed to quantify the interaction between neighboring optical sources in the cochlea during stimulation with infrared radiation.

Study design/materials and methods: In seven adult albino guinea pigs, a forward masking method was used to quantify the interaction between two neighboring optical sources during stimulation. Two optical fibers were placed through cochleostomies into the scala tympani of the basal cochlear turn. The radiation beams were directed towards different neuron populations along the spiral ganglion. Optically evoked compound action potentials were recorded for different radiant energies and distances between the optical fibers. The outcome measure was the radiant energy of a masker pulse delivered 3 milliseconds before a probe pulse to reduce the response evoked by the probe pulse by 3 dB. Results were compared for different distances between the fibers placed along the cochlea.

Results: The energy required to reduce the probe's response by 3 dB increased by 20.4 dB/mm and by 26.0 dB/octave. The inhibition was symmetrical for the masker placed basal to the probe (base-to-apex) and the masker placed apical to the probe (apex-to-base).

Conclusion: The interaction between neighboring optical sources during infrared laser stimulation is less than the interaction between neighboring electrical contacts during electrical stimulation. Previously published data for electrical stimulation reported an average current spread in human and cat cochleae of 2.8 dB/mm. With the increased number of independent channels for optical stimulation, it is anticipated that speech and music performance will improve. Lasers Surg. Med. © 2020 Wiley Periodicals LLC.

Keywords: channel interaction; cochlear implant; hearing; infrared stimulation; laser.

Fig. 1.

The sketch shows the basal turn of a guinea pig cochlea. Two cochleostomies have been created and two optical fibers, which are connected to two independent diode lasers, are inserted through the holes. The distance between the fibers is determined from the images captured during the experiments. Moreover, a silver wire is placed on the round window membrane to record auditory compound action potentials.

Fig. 2.

(A) Compound action potentials (CAPs) to optical pulses delivered by laser-1 (L1) at 5 milliseconds and laser-2 (L2) at 1 through 9 milliseconds (top to bottom trace) at fixed radiant exposures (~80 μJ/pulse). (B) The peak-to-peak amplitude of the CAP is plotted for L1 and L2. When the time at which L1 delivers a pulse is larger than the time L2 delivers a pulse, L1 is the probe and L2 the masker; if L1 delivers a pulse before L2, L1 is the masker and L2 the probe. The masking effect can be seen by the reduction in the response of the probe. (C and D) show the CAP responses for each laser at different radiant energies. The energy is increasing from the bottom to the top trace. Values can be seen in (E and F), which show the corresponding peak-to-peak CAP amplitudes versus the radiant energy. (G and H) show the CAP amplitudes (ordinate) evoked by the probe and the masker. From the bottom to the top trace, the radiant energy of the masker was 4, 19.1, 35.6, 50.6, 65.0, and 79.5 μJ/pulse in panel (G), and 5.3, 20.5, 35.5, 50.5, 64.5, and 80.4 μJ/ pulse in (H), respectively. The distance between the optical fibers was constant. (I and J) show the probe peak-to-peak CAP amplitudes for each masker-level (from G and H). Each probe-level pair is plotted and is shown as a contour plot (I and J). The color bars give the amplitude in μV.

Fig. 3.

Pure tone thresholds of the guinea pigs used in the study. Particular animals shown by the black filled markers had elevated thresholds.

Fig. 4.

Forward masking: the masker is delivered by one optical fiber followed 2 milliseconds later by the probe delivered via a second fiber. Shown is the compound action potential amplitude evoked by the probe. The distance between the optical fiber was 500 μm for (A) and (B) and 250 μm for (C) and (D), respectively. Based on the selection of the fiber to deliver the masking or probe stimulus, base-to-apex interactions (A and C), or apex-to-base interactions (B and D) are plotted. Although for (A) and (B) the timing of the pulse delivery was changed, for (C) and (D) the laser source was switched. The colors denote the CAP amplitude in μV. Along the abscissa, the radiant energy of the masker is plotted and along the ordinate, the radiant energy of the probe. Lines that are parallel lines to the x-axis indicate no interaction.

Fig. 5.

Parallel lines to the x-axis give the changes in probe's response amplitude at a fixed radiant energy for the probe and increasing energy for the masker (left to right), (A) and (B). To better quantify the masking effects, for each probe level the responses to different masker levels were normalized by dividing the results of the masked response by the results of their unmasked response (C) and (D). The intercept of the purple and the red dashed line shows the decrease of the probe's unmasked response by 3 dB. The probe's radiant energy was in (A) from the bottom to the top trace: 5.3, 20.5, 35.5, 50.5, 64.5, and 80.4 μJ/pulse and was in (B) from the bottom to the top trace: 4, 19.1, 35.6, 50.6, 65.0, and 79.5 μJ/pulse.

Fig. 6.

Shown are the masker radiant energies that are required to reduce the responses evoked by the probe by 3 dB. Base-to-apex configuration is shown by the black circles, the apex-to-base configuration by the red diamonds. The energies are plotted along a logarithmic axis versus the corresponding distances between the probe and the masker. For the entire data set, the space constant was 17.1 dB/octave (red broken line). If only values were considered below which the masking criteria could not be reached (green circled data points were used to fit the exponential function), the space constant was 26.0 dB/octave (blue broken line). The direction of the masker relative to the probe has little effect on the masking ability.

Fig. 7.

(A) The figure shows the change in frequency resolution in mm/octaves (ordinate) along the cochlea. The abscissa shows the frequency range (top) and the distance from the apex (bottom). Three train service has been added to demonstrate the spacing of spectral channels. Black stars are acoustic channels, Blue circles represent the distribution of spectral channels in the cochlear implant, and the red diamonds show the possible distribution of optical channels along the cochlea. (B) The three traces show again the spectral patterns for acoustic, electric, and optic stimulation. Corresponding filter shapes are shown by the black (acoustic), the blue (electric), and the red lines. Between the center off to frequency bands the signal would have dropped by 7.5 dB (black circle). Projecting this point towards optical and electrical stimulation shows that stimulation sources must be separated further to be considered as independent. The factor is 5.9 for electric stimulation and 2.7 for optical stimulation. For the length of a contemporary CI electrode the number of independent channels is about eight for electrical stimulation and about 18 for optical stimulation.