The concern regarding the Mitsar amplifier expressed with so much vigor by those with competing interests has met the reality test of actual recorded data. The concern expressed was over a theoretical time skewing error due to the data sampling of an older version of the Mitsar amplifier.
I suggested at the time that all the emotion was merely an example of someone yelling “the sky is falling”, like Chicken Little. There was no real problem, just lots of crying out and hand wringing.
I requested in an open international forum for anyone to send me a sample of the problem, and none could be produced. I suspected there was no real problem, as the sample issue was concerning a 500 sample/second device having a time skew… though this was in comparison to a database collected on a 100 sample per second device, with the waveforms interpolated from these samples.
It was highly suspect from my technical perspective when this issue was raised, and it was even more suspect when nobody could produce actual data showing the coherence or phase issue.
Testing now has shown that the old style Mitsar, with the non-simultaneous sampling is identical in performance to the new style amplifier that has simultaneous sampling, and thus no skewing error is possible in the newest amp. There is also an intermediate style amplifier tested, which is one of the smaller amps, but with a more current sampling design.
The data clearly show that there is no difference in coherence between these devices.
The Mitsar amplifier also has been tested with the new BranMaster Discovery amplifier, and it also was shown to have indenticle coherence findings with the Mitsar amplifier.
Clearly there is no real issue.
Data rules…. the experimental details are below.
We performed the following experiment.
We took three different Mitsar amplifiers:
1. Mitsar-EEG-201 – old model of amplifiers with relatively large time shift between channels (1.75 ms maximum) 2. Mitsar-EEG-201M – new model of amplifiers with relatively small time shift between channels (470 microsecond maximum) 3. Mitsar-EEG-202 – 32-channels amplifiers with zero time shift between channels.
Than we take Electro-Cap and put it on the head of one subject. We used linked ears referent for EEG recording.
We sequentially connect Electro-Cap and referents to three different amplifiers and perform independent recording of EEG in eyes closed condition. The duration of recording was longer than 300 seconds.
When we reconnect Electro-Cap and referents from one amplifier to another we do not touch to electrodes on the head and ears.
The total time of out experiment was approximately 20 minutes. This means the functional state of subject remain relatively stable.
Than we remontage the EEG to average referent (very important for time shift influence measurements), compute the coherence for all three EEG recordings using the same processing parameters and compare them. The duration of time interval for processing was the same for these EEG recording and was equal to 300 seconds.
We do not find any dramatic differences in coherence corresponding to different amplifiers. The small fluctuations can be explaned by amplifiers noise and non-stationarity of EEG.
The Mitsar system is calibrated at the manufacturer. There is no need to recalibrate the amplifier unless there is a serious problem from damage. The calibration button is so the user can run a test calibration signal to demonstrate that the channels are in fact correct and equal. If these were ever not correct (or equal) then the manufacturer would recalibrate the hardware device. It is blocked so that a user cannot accidentally mess the calibrations up. So in summary the calibration button in the software is pressed and then the button to observe EEG is pressed. Then the test calibration signal is generated and can be recorded. This is a good idea if the case is a medical or legal evaluation so that when the data is presented as evidence there is validation that a microvolt equals a microvolt.