Hi Jan,
There have been a few studies comparing CIS strategies with so-called “analog” strategies (which are no more analog than CIS, an electrical signal being an electrical signal). These analog strategies (which are no longer used clinically to my knowledge) provide each electrode with a compressed electrical version of the band-limited sound waveform. They allow simultaneous stimulation of all electrodes (no interleaving). A famous study comparing these two types of strategies is that by Wilson et al. (1991) which showed better speech scores with CIS than with analog, although one caveat was that the two strategies had several other differences than the interleaving, most importantly that CIS used 6 electrodes while analog used 4. Follow-up studies were not that clearcut with some subjects showing preference and/or better scores for one strategy or the other (Zwolan et al., 2005; Stupak et al., 2018).
Assuming we constrain ourselves to using pulsatile stimulation on all electrodes, I think interleaving greatly simplifies the fitting of the device and provides more consistent percepts when stimulating a certain electrode at a certain current level in the context of multichannel stim. This is because there can be important loudness differences when you present two pulses on two electrodes depending on whether these two pulses are non-simultaneous, simultaneous with the same polarity (which will sound much louder) or simultaneously with opposite polarity (which will sound much softer; e.g. de Balthasar et al., 2003). So it would be tricky (though maybe not impossible?) to allow each electrode to fire at any arbitrary time whilst controlling for the overall loudness of the multi-electrode stimulus. There have been a few strategies that fire a pulse on any given electrode every time there is an envelope maximum in the corresponding band-limited signal, still constraining the device to avoid simultaneous stimulation (e.g., FAST strategy by Zach Smith). So you end up in this case with pulse trains having an irregular rate on each electrode and pulses better synchronized to the envelope. But I think the main reported advantage of such strategies was that they reduced power consumption.
Finally, most contemporary devices stimulate at pretty high rates (around 1000 pps per electrode) and it has been shown that when loudness is equated, subjects have trouble even discriminating between pulse trains having different rates for rates above 300 pps (800 pps in the best cases), meaning that the sampling at 1000 Hz for each electrode is probably already enough for most of them. So I think it is unlikely that subjects would benefit from a more precise timing on individual channels than what is currently achieved in CIS-like stimulation strategies.
hope this helps,
Olivier
de Balthasar C, Boëx C, Cosendai G, Valentini G, Sigrist A, Pelizzone M. Channel interactions with high-rate biphasic electrical stimulation in cochlear implant subjects. Hear Res. 2003 Aug;182(1-2):77-87. doi: 10.1016/s0378-5955(03)00174-6. PMID: 12948604.
Stupak N, Padilla M, Morse RP, Landsberger DM. Perceptual Differences Between Low-Frequency Analog and Pulsatile Stimulation as Shown by Single- and Multidimensional Scaling. Trends Hear. 2018 Jan-Dec;22:2331216518807535.
Zwolan TA, Kileny PR, Smith S, Waltzman S, Chute P, Domico E, Firszt J, Hodges A, Mills D, Whearty M, Osberger MJ, Fisher L. Comparison of continuous interleaved sampling and simultaneous analog stimulation speech processing strategies in newly implanted adults with a Clarion 1.2 cochlear implant. Otol Neurotol. 2005 May;26(3):455-65. doi: 10.1097/01.mao.0000169794.76072.16.
Wilson, B., Finley, C., Lawson, D. et al. Better speech recognition with cochlear implants. Nature 352, 236–238 (1991).Dear List,
it is my understanding that the vast majority of CI sound processors in use today are still based or inspired more or less on some variant of the Continuous Interleaved Sampling algorithm, and that one of the key assumptions / design features of the algorithms in use is that having more than one electrode channel active in any one ear at the same time is to be avoided. Hence "interleaved" sampling: channels take turns to ensure they aren't active at once. What I am curious to know is: quite how bad would it be if this assumption was violated? Is it necessarily always very bad? And how certain can we be about how bad it is? Have people run head-to-head comparisons of strategies with and without strict interleaving?Intuitively, while I see that having multiple channels active at once may exacerbate problems with the already relatively poor channel isolation, I also think that forcing channels to fire "in turn" constrains the timing of pulses in a manner that may preclude independent temporal coding on different channels. Has this potential trade-off been considered? Is there a well founded consensus that the downsides of having the potential of temporal collisions of pulses in different channels will necessarily outweigh potential upsides from having richer temporal patterning across channels?
I would be grateful for references / papers / views / perspectives relevant to this topic.
Many thanks!
Jan
---------------------------------------Prof Jan Schnupp
City University of Hong Kong
Dept. of Neuroscience