Epilepsy: seizure detection

Hi, I am a french engineer working (for free) on the top of my regular work for a project called ?epimouv?.

The idea is to bring on the market devices allowing Epilepsy seizure detection and alarm. This would really help parents who are taking care of epileptic children (that is my case).
Today?s device is based on a movement detector (MEMS chip attached at the whistle). Signal is processed and is able to detect seizure (today?s application is only when the patient is laying down and asleep).

The main constraint is that the patient needs to wear ?wearable? device.

I am then very interested by your devices probing skin conductivity (biofeedback) and mind waves as well. And before starting experimentation, I would like to know if someone else has some experience or information about it.

My questions are the following:

• Did any of you work on such a project yet ?
• Is it feasible with biofeedback sensor to discriminate seizure ?
• Is there already an application ready to use that is detecting seizure ?
• Are the mindplace products enough "open" to apply a different signal processing allowing seizure detection ?

Thanks for any information.

Greetings, Falconu--

Your project sounds fascinating & I hope you will keep us informed of your progress. We've not done any work on such a device ourselves, as we do not make medical devices. I have read of seizure detection software used by hospitals, using EEG as input signal, but I'm not sure this has made its way to the consumer market. Also, Barry Sterman and others have used neurofeedback training to allow epileptics to avert a seizure - in Sterman's case, but having the subject alternative between two different EEG frequency bands; this resulted in fewer and less severe seizures. I'm not aware of GSR changes during or immediately prior to a seizure, but have not reviewed that literature. The ThoughtStream USB does emit a fairly simple serial data stream, 14 bits per sample and 20 Hz sample rate, and we can send you the specification if you'd like

-Robert

Thank you Robert,

That is very interesting to know that using neurofeedback training could result in less severe seizures => I will spread this information around.
But this is not exactly my today's topic.
I would appreciate if you can send the spec of the thoughtstream USB interface. On my side , I will keep you informed of my progress.

Gabriel

This thread has information about the data that comes from the Thoughtstream.

Ok great. Thanks a lot !

Quote:

---

Originally Posted by Falconu

Hi, I am a french engineer working (for free) on the top of my regular work for a project called “epimouv”.

The idea is to bring on the market devices allowing Epilepsy seizure detection and alarm. This would really help parents who are taking care of epileptic children (that is my case).
Today’s device is based on a movement detector (MEMS chip attached at the whistle). Signal is processed and is able to detect seizure (today’s application is only when the patient is laying down and asleep).

The main constraint is that the patient needs to wear “wearable” device.

I am then very interested by your devices probing skin conductivity (biofeedback) and mind waves as well. And before starting experimentation, I would like to know if someone else has some experience or information about it.

My questions are the following:
Did any of you work on such a project yet ?
Is it feasible with biofeedback sensor to discriminate seizure ?
Is there already an application ready to use that is detecting seizure ?
Are the mindplace products enough "open" to apply a different signal processing allowing seizure detection ?

Thanks for any information.

---

Hmmm, I am a bit puzzled by how you think any device will detect seizures that is NOT worn in some way. Even with the skin conductivity, the subject has to "wear" some device that holds the electrodes. Even biofeedback sensors that are typically available must be "worn" to get the sensor (usually electrode[s]) in contact with the skin. As far as I know, even in the hospital with EEG monitoring, seizures cannot be predicted; you can see EEG changes AFTER the seizure is started.

I'm sure if you have searched the literature that you have found no reliable home-use devices are commercially available. The few that exist have extremely high false alarm rates.

One that is still under development is the EpiLert by BioLert http://www.biolertsys.com/
They already did clinical trials at NYU medical center.
http://www.biolertsys.com/dynamic.as...ress%20Release

This works by monitoring the subject's movements all the time, then within about 10-30 seconds, determines if a seizure has started based on the movement patterns. It is probably very similar to your "epimouv", but with better detection and false alarm rates. I think their trials were with a lot more subjects than you used.

I suspect that the skin conductivity, while interesting, would not be very reliable due to so many types of events that will trigger skin conductivity increases. A lot more research is required to determine if there is some aspect of skin conductivity change that is unique to onset of a seizure that could be used to distinguish from non-seizure conductivity changes.

As for biofeedback, that does not seem to be a detection method, but rather a method to condition a subject to greater awareness of impending seizures or perhaps even partial control of seizures.