Thank you for a worthwhile presentation.
The most important and difficult point is considered to be the determination of counting the number of real peaks of waveforms coincided with the swallowing movement. Your threshold-based algorithm is so simple and comprehensible. But applying a peak-counting algorithm is expected to be more detectable.
1. In case of Figure 4, If a certain hysteresis(ignore any movement within the range) is applied using a peak-counting algorithm, the small peak would be neglected on the count 3 without Vth1. If the value of the lower peaks is unstable (e.g. in case of dysphagia) and the value is slightly over the Vth2, I wonder the counting of the peak will be ignored while the swallowing is performed. It would be helpful if you have any idea or comments on things above.
2. The [Vmean] is indicating the value of the rest position of the larynx. The term [mean] reminds us average value. So it would be delightful for our clinicians if you use [Vrest] in stead of [Vmean].
Thank you for your question and I am very sorry to late the reply.
1. The peak-counting algorithm will be nice when the measurement is performed without body motion of the subject.
We were afraid of the noise coincided with the body motion.
The MI sensor will also catch the movement of the subject
because there are the magnetic field related to the earth magnetism.
Hence, we used the two threshold values Vth1 and Vth2
as much as possible to reject the counting by noises.
The threshold-based algorithm increases the count when a pair of the signals <under the Vth2> and <over the Vth1> are found.
So, we can neglect the other patten signal caused by the noise.
When we use the peak-counting algorithm or
we want better result using the threshold-based algorithm,
the hardware should be improved to reject the noises of the body motion.
2. Thank you for your advice.
We can define the Vrest as the rest of the larynx.
However, we cannot obtain the rest *position* from the signal.
The output signal of the MI sensor shows the changes of
the magnetic field between 0.3 Hz to 5Hz; there are a bandpass filter in the MI sensor.
When there are no changes of the magnetic field,
the MI sensor output the Vmean(=2.5V).