Nick:
The procedure I suggest involves the following steps which should be
easy.
1. Take the original data and compute first order derivatives. This
will give you constant values between the retraces. The retraces will
have large values of the derivatives.
2. Automatically eliminate the retrace spikes by computing a
histogram of the derivative values. It should be bimodal with one value
clustering about the constant slope, and the other value about the
retrace values. The modes should be well separated.
3. Use the average value of the lowest mode as the slope and slow
phase.
4. Another way is to compute an average value of the derivatives
using a least square fit of the original data, Compute the rms
deviation. Automatically eliminate data that represent outliers (more
than one standard deviation) because they are associated with retraces.
Recompute the standard deviation without the outliers. Repeat until the
average value of the slope does not change much indicating convergence.
Only a few passes will be required, depending upon the accuracy you
require. The computation process can be made automatic. You might also
be able to do this using Excel.
I strongly suggest that you use an infrared EyeTrace system because
it is more stable and usable than EOG. I know of a light weight goggle
system for binocular horizontal measurement and recording. It has analog
outputs and data file recording and is available for only $5,000. It
connects to a PC or laptop with the serial port (RS-232) and includes
software for Windows 95.
This is one I intend to use in my own research. E-mail me if you
want more information.
Good luck.
Sol
=====
bour wrote:
> nick webb wrote:
>
>> Can anyone point me in the direction of an algorithm for analysing
>> nystagmus
>> from EOG recordings? I am interested in how the Gain of the slow
>> phase
>> changes with time. My eye movement files are in MATLAB format at the
>>
>> moment.
>>
>> Any help would be greatly appreciated....
>>
>> cheers,
>>
>> Nick Webb
>
> Dear Nick,
>
> I'm sorry to tell you that analysing slow phase of nystagmus from
> EOG recordings is of little use. This sounds as coming from the
> stone age. Especially slow phases are unreliable with EOG
> because of the poor signal to noise ratio and the spontaneous drift
> of the signal.
> The only advise I can give you is:
> Go to a better measurement procedure (infrared, video, etc)
> before you try anything else.
> This is a waste of time.
>
> --
> Lo Bour, PhD
>
> Department of Neurology/Clinical Neurophysiology, H2-222
> Academic Medical Centre, University of Amsterdam
> Meibergdreef 9, 1105 AZ Amsterdam
> The Netherlands
>
> e-mail: [log in to unmask]
> fax : +31 20 5669187
> tel : +31 20 5663515
>
>
--
======================================================================
Sol Aisenberg, Ph.D. [Technology and Business Advisor]
International Technology Group, The Technology and Information Company
Practical Inventions/Intellectual Property by Request - No Obligations
36 Bradford Road, Natick, MA 01760
Tel: 508/651-0140 Fax: 508/653-0759
http://www.tiac.net/users/itgroup/ [log in to unmask]
======================================================================
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|