[Cyborg] medical mirror

Mike Schachter mike at mindmech.com
Thu Oct 21 09:47:26 PDT 2010


Todd, seems like you're going to have to use OpenCV to identify the
region of interest in the fish face though (j/k). They speculate that
the heart rate signal is contained mostly in the green channel:

"The second component typically, but not always contained the
strongest pulse signal. It is unclear why this is the case but one
might expect it to be related to the green channel and the fact
that hemoglobin absorptivity is highest in green/yellow light."

Do you have your zebrafish fixed while you're trying to take data?

In general, is anyone interested in this stuff from a machine-learning
perspective? We give (almost) weekly talks at the machine learning
group and I'd be happy to go over the paper and a high level overview
of ICA. We also have a guy who is familiar with OpenCV for the face
recognition. Maybe we could all get together on a Machine Learning
Wednesday (7:30pm)?

  mike



On Wed, Oct 20, 2010 at 11:51 PM, Todd Anderson <tanders at stanford.edu>wrote:

> Yea, I was hoping to just feed it the fish data.  I have experience with
> ICA, so I think I'll just use the paper as inspiration!
>
>
>
> On Wed, Oct 20, 2010 at 7:54 PM, Eric Boyd <mrericboyd at yahoo.com> wrote:
>
>> The article I linked
>>
>>
>> http://www.opticsinfobase.org/abstract.cfm?URI=oe-18-10-10762
>>
>> actually has enough technical details that you could probably write code.
>>  Of course that's a big project :-).
>>
>> Eric
>>
>>
>>
>>
>>
>> On 10/20/10 10:46 PM, Todd Anderson wrote:
>>
>>> Did anyone ever find a software package or code for this?  I'd like to
>>> see if it can track zebrafish heart rates...
>>>
>>>
>>>
>>> On Mon, Oct 11, 2010 at 1:03 PM, Eric Boyd <mrericboyd at yahoo.com
>>> <mailto:mrericboyd at yahoo.com>> wrote:
>>>
>>>    My guess would be that they can't really get HRV.  It looks to me like
>>>    their analysis system requires multiple beats just to get the
>>> heart-rate
>>>    accurately (they do the equivalent of an FFT to get it). But given
>>> that
>>>    they can get HR at all, I'm sure getting HRV is just a matter of time.
>>>    Might be hard with only 15Hz webcam though - if my calculations are
>>>    right that limits intra-beat timings to +/-4BPM, yes?  Or could you
>>>    calculate HRV over a big data set?!?
>>>
>>>    I am looking forward to having my new pulse-choker stuff (using polar
>>>    heart rate receiver and a heart-shaped circuit I have designed), I'm
>>>    totally gonna hack a logger into one of them and when I do so I'll
>>> share
>>>    the data here :-).
>>>
>>>    Eric
>>>
>>>    On 10/9/10 6:03 PM, Tomm wrote:
>>>     > A tray app that monitors your heart rate is a great idea.  I
>>>    wonder if
>>>     > the time resolution is good enough to measure heart rate variation
>>>     > (HRV)?  Variations in heartrate is also pretty interesting:
>>>     >
>>>     > "[...] asked 52 adults to report how often they experienced
>>> positive
>>>     > emotions like happiness, awe, and gratitude and how socially
>>>    connected
>>>     > they felt in their social interactions every day for a period of
>>> nine
>>>     > weeks. The researchers also measured the HRV of each individual
>>>    at the
>>>     > beginning and end of the study by measuring heart rate during a
>>>     > two-minute session of normal breathing. HRV at the beginning of the
>>>     > study predicted how quickly people developed positive feelings and
>>>     > experiences of social connectedness throughout the nine-week
>>>    period. In
>>>     > addition, experiences of social connectedness predicted increases
>>>    in HRV
>>>     > at the end of the study, demonstrating a reciprocal relationship
>>>    between
>>>     > heart rate and having satisfying social experiences."
>>>     >
>>>
>>> http://www.scientificamerican.com/article.cfm?id=the-neuroscience-of-heart
>>>     >
>>>     >       Tom
>>>     >
>>>     > On Sat, Oct 9, 2010 at 8:21 AM, Eric Boyd <mrericboyd at yahoo.com
>>>    <mailto:mrericboyd at yahoo.com>
>>>     > <mailto:mrericboyd at yahoo.com <mailto:mrericboyd at yahoo.com>>>
>>> wrote:
>>>     >
>>>     >
>>>    http://www.youtube.com/watch?v=LyWnvAWEbWE&feature=player_embedded
>>>    <http://www.youtube.com/watch?v=LyWnvAWEbWE&feature=player_embedded>
>>>     >
>>>    <http://www.youtube.com/watch?v=LyWnvAWEbWE&feature=player_embedded
>>>    <http://www.youtube.com/watch?v=LyWnvAWEbWE&feature=player_embedded>>
>>>     >
>>>     >     Basically, using nothing other than a webcam (1.3MP, 15 Hz!),
>>>    they can
>>>     >     read your heart rate.  Then they built that into a "medical
>>>    mirror", so
>>>     >     you can see your heart beat in the morning when you brush
>>>    your teeth :-)
>>>     >
>>>     >     I found one of his papers about it online, full text:
>>>     >
>>>     >     Non-contact, automated cardiac pulse measurements using video
>>>    imaging
>>>     >     and blind source separation
>>>     > http://www.opticsinfobase.org/abstract.cfm?URI=oe-18-10-10762
>>>     >
>>>     >     Looks to me like the big breakthrough here is the motion
>>>    independence,
>>>     >     which they got through a combination of face recognition and
>>>    cropping
>>>     >     and separating the video into it's three component colors.
>>>      They claim
>>>     >     an average error from the reference pulseox reading of less
>>>    than 1 BPM
>>>     >     even with movement, that's pretty impressive.
>>>     >
>>>     >     Anyone want to write a laptop app which would just keep the
>>>    webcam on
>>>     >     all the time and log your heartbeat?
>>>     >
>>>     >     Eric
>>>     >     _______________________________________________
>>>     >     Cyborg mailing list
>>>     > Cyborg at lists.noisebridge.net
>>>    <mailto:Cyborg at lists.noisebridge.net>
>>>    <mailto:Cyborg at lists.noisebridge.net
>>>    <mailto:Cyborg at lists.noisebridge.net>>
>>>     > https://www.noisebridge.net/mailman/listinfo/cyborg
>>>     >
>>>     >
>>>    _______________________________________________
>>>    Cyborg mailing list
>>>    Cyborg at lists.noisebridge.net <mailto:Cyborg at lists.noisebridge.net>
>>>    https://www.noisebridge.net/mailman/listinfo/cyborg
>>>
>>>
>>>
>>>
>>> --
>>> http://www.octotod.net
>>>
>>
>
>
> --
> http://www.octotod.net
>
> _______________________________________________
> Cyborg mailing list
> Cyborg at lists.noisebridge.net
> https://www.noisebridge.net/mailman/listinfo/cyborg
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.noisebridge.net/pipermail/cyborg/attachments/20101021/b1332f88/attachment.htm 


More information about the Cyborg mailing list