Code for spike and burst detection

[Reinhard]

Hi, everyone,

I downloaded a folder of matlab codes in the Google site, but seems it didn't provide a code to detect the Burst. Did anyone have some codes can detect the brust, brust detection, network burst detection, cross-correlation, etc? I found a code named SPYCODE, which was developed by Dr. Chiappalone, has such features. But I didn't get a reply from her. I'm looking for somebody has SPYCODE or some others with similar features. Thank you so much!

[Jon Newman]

Burst detection is trickier than you might think because they come in all shapes and sizes. There are many algorithms to do it. The two most prominent are

1. A simple threshold set at something like 5 to 10X the baseline firing rate. This is what is used in our recent paper:

http://www.nature.com/ncomms/2015/150309/ncomms7339/full/ncomms7339.html

2. Burstiness-index. 

http://www.jneurosci.org/content/25/3/680.full.pdf+html

Extracted from that paper:

Bursts come

in different forms, so simply tallying up the

number of bursts is not sufficient to describe

the burstiness of a culture. It is essential to account

for the size of bursts, measured in terms

of number of participating neurons, aggregate

number of spikes, or duration. Fortunately, we

found that it is not necessary to identify individual

bursts to quantify the level of burstiness of a

recording. Instead, we used the following method:

divide a 5 min recording into 300 1-sec-long

time bins and count the number of spikes (total

across all electrodes) in each bin. Compute the

fraction of the total number of spikes accounted

for by the 15% of bins with the largest counts. If

the firing rate is tonic, this number, f15, will be

close to 0.15. Conversely, if a recording is so

bursty that most of the spikes are contained in

bursts, f15 will be close to 1, because even at the

highest burst rates observed during these experiments,

bursts did not occupy 45 1-seclong

bins (15%) in a 5 min recording. We then

defined a burstiness index (BI), normalized between

0 (no bursts) and 1 (burst dominated) as

BI ( f15 0.15)/0.85. (Statistical fluctuations

make the BI deviate slightly from zero even in

complete absence of bursts.)

Aside from this there are more complex ways to characterize the nature of sychronized spiking activity in cultures, e.g. looking at the distribution of sizes of sychronized events instead of looking at a single moment of the distribution. This is what Plenz has been doing for years.

http://www.maths.tcd.ie/~mnl/store/BeggsPlenz2003a.pdf

Jon

On Sun, Jul 5, 2015 at 9:53 PM, Reinhard <laos...@gmail.com> wrote:

Hi, everyone,

I downloaded a folder of matlab codes in the Google site, but seems it didn't provide a code to detect the Burst. Did anyone have some codes can detect the brust, brust detection, network burst detection, cross-correlation, etc? I found a code named SPYCODE, which was developed by Dr. Chiappalone, has such features. But I didn't get a reply from her. I'm looking for somebody has SPYCODE or some others with similar features. Thank you so much!

--
You received this message because you are subscribed to the Google Groups "NeuroRighter Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neurorighter-us...@googlegroups.com.
To post to this group, send email to neurorigh...@googlegroups.com.
Visit this group at http://groups.google.com/group/neurorighter-users.
For more options, visit https://groups.google.com/d/optout.

--

Jonathan Newman

Postdoctoral Fellow, MIT

http://www.mit.edu/~jpnewman/

[Jon Newman]

I just realized that our paper did not actually have the methodology for burst detection described in full...That makes me a little sad. Anyway, this comes from Ming's thesis where the method is described very explicitly:

Groups of temporally-correlated action potentials that were spatially-distributed among electrodes were called ’bursts’. The time points when bursts occurred were detected post-hoc using a simple thresholding approach. First, the MEA-wide firing rate histogram was calculated for 10-ms bins. Any bin that exceeded 100 times the overall pre-treatment firing rate was initially counted as a candidate burst. Because a single burst typically spans several bins, candidate bursts from consecutive bins were combined into a single burst. Because bursts sometimes show reverberating activity toward their ends, candidate bursts detected within 1 second of a previous burst were discarded. The burst rate was calculated based on number of bursts occurring within a given time. The interburst firing rate was calculated from portions of the MEA-wide firing rate histogram that occurred at least 2 seconds before and 8 seconds after a burst occurred to eliminate the possibility that any buildup or residual activity contributed to this value. For both burst rate and interburst firing rate, any activity was assessed by normalizing the 24-hour treatment period to the 3-hour pre-treatment period. Statistical significance was determined using a Kruskal-Wallis test followed by Wilcoxon rank-sum tests with Bonferroni correction for multiple comparisons. 

[Reinhard]

Hi, Jon,

Thanks a lot! I will read this references soon.

To unsubscribe from this group and stop receiving emails from it, send an email to neurorighter-users+unsub...@googlegroups.com.

To post to this group, send email to neurorigh...@googlegroups.com.
Visit this group at http://groups.google.com/group/neurorighter-users.
For more options, visit https://groups.google.com/d/optout.

--

Jonathan Newman

Postdoctoral Fellow, MIT

http://www.mit.edu/~jpnewman/