Open sathishande opened 7 years ago
liubenyuan
commented
7 years ago Hi, this is a python (initial) implementation of https://github.com/jovo/fast-oopsi
py-oopsi is mainly used for the connectomics challenges (see http://www.kaggle.com/c/connectomics). I am not a pro on this field, I may help you if you provide a small segment of this sample data or, you may ask these problem on jovo's project.
sathishande
commented
7 years ago Hi, I have attached sample fluorescent data of six neurons as excel file. This data has frame rate of 1 Hz ,i.e., frame duration = 1 sec. The issues with fast-nonnegative deconvolution with this data are as follows. 1) Even-though there is no variation in fluorescent intensity, it is identifying spikes .i.e., those are unwanted spikes. 2) when presence of noise, it is detecting too many spikes.
Should we have frame rate above 10 Hz or 20 Hz for this algorithm?
On Wed, Jul 5, 2017 at 3:59 PM, liubenyuan notifications@github.com wrote:
Hi, this is a python (initial) implementation of https://github.com/jovo/fast-oopsi
py-oopsi is mainly used for the connectomics challenges (see http://www.kaggle.com/c/connectomics). I am not a pro on this field, I may help you if you provide a small segment of this sample data or, you may ask these problem on jovo's project.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/liubenyuan/py-oopsi/issues/2#issuecomment-313065399, or mute the thread https://github.com/notifications/unsubscribe-auth/AVeMmQ8RpqGzWddzaknv00CYAWPdsMhlks5sK2WfgaJpZM4ON9V6 .
--
Thanks & Regards
SATHISH ANDE PhD Scholar Communication & Signal Processing Dept. Of Electrical Engineeering IIT Hyderabad ODF Estate, Yeddumarilaram-502205 Telangana-India Ph.no: +917893929684
sathishande
commented
7 years ago Hi, I have attached sample fluorescent data of six neurons as excel file. This data has frame rate of 1 Hz ,i.e., frame duration = 1 sec. The issues with fast-nonnegative deconvolution with this data are as follows. 1) Even-though there is no variation in fluorescent intensity, it is identifying spikes .i.e., those are unwanted spikes. 2) when presence of noise, it is detecting too many spikes.
Should we have frame rate above 10 Hz or 20 Hz for this algorithm?
On Wed, Jul 5, 2017 at 4:21 PM, Sathish Ande sathish.vic@gmail.com wrote:
Hi, I have attached sample fluorescent data of six neurons as excel file. This data has frame rate of 1 Hz ,i.e., frame duration = 1 sec. The issues with fast-nonnegative deconvolution with this data are as follows. 1) Even-though there is no variation in fluorescent intensity, it is identifying spikes .i.e., those are unwanted spikes. 2) when presence of noise, it is detecting too many spikes.
Should we have frame rate above 10 Hz or 20 Hz for this algorithm?
On Wed, Jul 5, 2017 at 3:59 PM, liubenyuan notifications@github.com wrote:
Hi, this is a python (initial) implementation of https://github.com/jovo/fast-oopsi
py-oopsi is mainly used for the connectomics challenges (see http://www.kaggle.com/c/connectomics). I am not a pro on this field, I may help you if you provide a small segment of this sample data or, you may ask these problem on jovo's project.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/liubenyuan/py-oopsi/issues/2#issuecomment-313065399, or mute the thread https://github.com/notifications/unsubscribe-auth/AVeMmQ8RpqGzWddzaknv00CYAWPdsMhlks5sK2WfgaJpZM4ON9V6 .
--
Thanks & Regards
SATHISH ANDE PhD Scholar Communication & Signal Processing Dept. Of Electrical Engineeering IIT Hyderabad ODF Estate, Yeddumarilaram-502205 Telangana-India Ph.no: +917893929684
--
Thanks & Regards
SATHISH ANDE PhD Scholar Communication & Signal Processing Dept. Of Electrical Engineeering IIT Hyderabad ODF Estate, Yeddumarilaram-502205 Telangana-India Ph.no: +917893929684
liubenyuan
commented
7 years ago I think you should try to adjust the parameters of py-oopsi or try fast-oopsi (MATLAB) of jovo. And I can not find the attached excel files at this post.
jovo
commented
7 years ago i was told recently that there might be a bug in the python implementation. possibly lambda is set to 1/lambda?
otherwise, i know that danielle witten has a new paper on the arxiv with link to code that seems much better than my original matlab code (their code is in R). let me know if that helps!
On Wed, Jul 5, 2017 at 8:42 PM, liubenyuan notifications@github.com wrote:
I think you should try to adjust the parameters of py-oopsi or try fast-oopsi (MATLAB) of jovo. And I can not find the attached excel files at this post.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/liubenyuan/py-oopsi/issues/2#issuecomment-313264315, or mute the thread https://github.com/notifications/unsubscribe-auth/AACjci5clbI2J9-genrg4W_gjx3xi2yqks5sLC2JgaJpZM4ON9V6 .
-- the glass is all full: half water, half air. neurodata.io ps - i tend to respond to email weekly so i can spend more time with my wife & baby, thank you
sathishande
commented
7 years ago Hi, I have attached sample fluorescent data of six neurons as excel file. This data has frame rate of 1 Hz ,i.e., frame duration = 1 sec. The issues with fast-nonnegative deconvolution with this data are as follows. 1) Even-though there is no variation in fluorescent intensity, it is identifying spikes .i.e., those are unwanted spikes. 2) when presence of noise, it is detecting too many spikes.
Should we have frame rate above 10 Hz or 20 Hz for this algorithm?
On Thu, Jul 6, 2017 at 6:13 AM, liubenyuan notifications@github.com wrote:
I think you should try to adjust the parameters of py-oopsi or try fast-oopsi (MATLAB) of jovo. And I can not find the attached excel files at this post.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/liubenyuan/py-oopsi/issues/2#issuecomment-313264315, or mute the thread https://github.com/notifications/unsubscribe-auth/AVeMmaoIQTrde0kLart1dBffQNT50tGQks5sLC2JgaJpZM4ON9V6 .
--
Thanks & Regards
SATHISH ANDE PhD Scholar Communication & Signal Processing Dept. Of Electrical Engineeering IIT Hyderabad ODF Estate, Yeddumarilaram-502205 Telangana-India Ph.no: +917893929684
I am running this code by taking frame rate of 1 Hz. But this algorithm identify spikes even though there is no variation in fluorescence intensity or there is presence of noise. How do i handle these issues?