3Recording technique
A clean ECochG trace is the product of three deliberate choices: where the electrode sits, what stimulus you deliver, and how you filter and average what comes back. The disease pages assume you have a clean trace. This module is how you get one.
The cochlea makes signals on the order of one microvolt. The scalp around it makes signals — muscle artefact, the EEG, the heart — on the order of millivolts. Recording ECochG is therefore a problem of signal isolation, not signal generation: the response is always there, but whether you can see it depends on getting the electrode close enough, the stimulus consistent enough, and the noise floor low enough that averaging can pull a 1 µV signal out of a 50 µV background.[2020]
FWhere to put the electrode
Four placements are in clinical use. They differ in invasiveness and — directly because of that — in signal amplitude. The closer the recording electrode sits to the cochlea, the larger the recorded potential, with roughly an order-of-magnitude step at each move inward.[2017]
| Placement | Where it sits | Typical amplitude | Invasiveness |
|---|---|---|---|
| Tiptrode | Gold-foil-wrapped foam insert, in the ear canal | ~1× (≈0.5–1.5 µV CAP) | None — patient comfortable |
| TM electrode (tymptrode) | Soft tip resting against the tympanic membrane | ~3× (3–10× tiptrode) | Mild — needs cooperation |
| Transtympanic | Needle through the TM onto the cochlear promontory | ~10× tiptrode amplitude | Invasive — needs local anaesthesia |
| Intracochlear | An apical electrode of an implanted CI array | ~25× tiptrode | Only available during/after CI surgery |
The distance-to-source effect is straightforward physics — the cochlea is a tiny dipole, the recorded voltage falls off with distance — but it has practical consequences. About 71% of clinical ECochG users prefer the tiptrode or TM electrode despite the smaller amplitudes, because the gain in patient tolerance outweighs the cost of needing more averaging and tighter noise control.[2010]Transtympanic is reserved for cases where the diagnostic question is hard enough that the larger, cleaner signal earns the needle (longstanding Ménière's evaluation in some centres, intraoperative monitoring of the round window niche during CI surgery).
Published SP/AP ratios from transtympanic and extratympanic recordings are not directly comparable. Both ratios are dimensionless, so a naive reader might think a 0.45 ratio means the same thing from a tiptrode and a TT needle. It doesn't — the SP and AP scale together, but not perfectly, because the SP has slightly different generators (and therefore slightly different spatial dropoff) than the AP. Gibson noted in 1992 that confusion on this point — audiologists applying TT-derived cutoffs to ear-canal recordings — held the field back for years.[2017] When you read a sensitivity figure for ECochG, always check which electrode was used.
FTPlacement explorer
The four placements share an anatomy but differ in how far the recording site sits from the cochlear dipole. Read the schematic below from canal (1) to cochlea (4) and note how the listed amplitude scale tracks the inward step.
FChoosing a stimulus
Three stimulus families are in routine use. The choice is not aesthetic — each shapes which of the ECochG components you can read and which questions you can answer.
Click
A brief (~100 µs) electrical pulse delivered to the insert phone produces a broadband transient. It excites the entire cochlea, with the basal turn responding first (because high-frequency regions sit at the base) and the apex last. The brief stimulus is ideal for evoking a sharp, synchronous AP — and is the only stimulus that reliably evokes a recordable CM at the tiptrode level.[2021] Default for screening, default for any question that turns on the CM (auditory neuropathy in particular).
Tone burst
A brief windowed sinusoid — typical clinical setting: 1 or 2 kHz carrier, 1–2 ms linear rise/fall, 10–14 ms plateau, alternating polarity, around 90 dB nHL.[2000] The defining feature is what happens during the plateau: the SP persists as long as the stimulus persists. Where a click SP is a thin shoulder you have to measure with a caliper, a tone-burst SP is a flat plateau you can read off the trace directly. Tone-burst protocols therefore raise the diagnostic sensitivity for Ménière's disease from ~60% (click amplitude ratio) to ~85–92% when combined with other measures.[2017, 2019]
Chirp
A frequency-swept stimulus — broadband, but with low frequencies delivered earlierand high frequencies later, by an interval matched to the cochlea's traveling-wave delay so all cochlear regions are excited simultaneously. Elberling and colleagues introduced the CE-Chirp in 2007 for ASSR work and the wider-band ABR community has adopted it.[2007] For ECochG specifically, chirps boost the AP because the cross-frequency neural volley is more synchronous. The cost: chirps elicit a smaller, less reliable CM than clicks — about 58–63% success rate vs 92–100% for clicks at the same intensity — because the stretched-out stimulus envelope no longer produces a tight oscillation for the OHCs to follow.[2021]
TStimulus comparator
Use click when the question is about CM presence or absence (auditory neuropathy), about the AP alone (intraoperative monitoring), or for a quick first screening pass. Use tone burst when the question is about SPelevation (Ménière's, third-window conditions, fistula). Use chirp when you specifically need a larger AP (for example, supra-threshold synaptopathy work where wave I amplitude is the metric) and can accept reduced CM yield.[2010, 2021]
TRecording parameters
The synthesis and detection math in the simulator on the previous module assumed standard values for every recording parameter. Those values aren't arbitrary; they reflect physiological signal properties and a half-century of practical optimisation.
| Parameter | Typical setting | Reason |
|---|---|---|
| Stimulus rate | 11.1–11.3 /s | Slow enough to avoid neural adaptation distorting the AP; fast enough to collect 1000–2000 sweeps in < 3 min. |
| Recording window | 10–14 ms post-stimulus | Captures CM, SP, AP. Tone burst needs longer (≥14 ms) to capture both onset and offset responses. |
| Band-pass filter | 3–100 Hz HPF, 1500–3000 Hz LPF | HPF removes drift and EEG; LPF removes muscle artefact and 50/60 Hz mains. CM extraction sometimes uses a wider 100–3000 Hz band to keep the high-frequency stimulus following.[2021] |
| Sweeps averaged | 1000–2000 (clinical); 200–500 (intraop) | Signal-to-noise improves as √N. Tiptrode needs more sweeps than TT because raw signal is ~10× smaller. |
| Polarity | Alternating (cancels CM); rarefaction or condensation alone (for CM) | Alternating averaging eliminates the CM and leaves SP + AP clean. Single-polarity recording is needed when the question is the CM. |
| Intensity | 80–95 dB nHL | High enough to give a robust SP + AP; low enough to avoid stimulus artefact and patient discomfort. |
| Impedance | ≤ 5 kΩ per electrode, < 2 kΩ between | Above this, line noise and thermal noise overwhelm the 1 µV signal.[2020] |
FA workable starting protocol
For a tiptrode click recording on an alert adult patient, this set of parameters reproduces what most clinical labs are doing today. It is not a substitute for local validation, but it is a defensible starting point you can run on day one:
- Patient supine, relaxed, eyes closed. Avoid swallowing during epochs.
- Skin preparation at vertex (active) and contralateral mastoid (reference), forehead ground; impedance < 5 kΩ.
- Tiptrode in the test ear canal, snug fit, gold foil contacting the canal skin.
- Click: 100 µs, alternating polarity, 11.1/s, 90 dB nHL.
- Band-pass 3–1500 Hz, 10 ms window, average ~1500 sweeps.
- Repeat with rarefaction and condensation polarities separately if you need the CM (e.g. suspected ANSD).
- For Ménière's protocol, add 1 and 2 kHz tone bursts at 90 dB nHL, 2 ms rise/fall, 10 ms plateau, 11.1/s, 1000 sweeps.
Five things go wrong frequently and look similar — a flat trace, large stimulus artefact, no recognisable AP. Before you conclude "absent response", check: (1) impedances drifting up > 5 kΩ during the run (re-prep), (2) tiptrode foil rotated away from the canal skin (re-seat), (3) patient muscle activity contaminating epochs (relax jaw, neck), (4) stimulus polarity stuck (alternate is set), (5) insert phone tube blocked or unseated (the missing CM in particular suggests no stimulus is actually reaching the cochlea, not absent biology). The polarity-block check from Module 7 — covering the tube to confirm the "CM" disappears — is the single best way to discriminate stimulus artefact from a real biological CM.[2022]