Rhythmic Artifact Issue

[maximilia...@gmail.com]

Hello!

 

I’ve conducted several LFP recordings from the rat hippocampus and some recordings show a rhythmic artifact around every 400 milliseconds (the distance between artifacts appears to be very stable). The artifact resembles a large negative amplitude deflection, which can also be seen in the spike viewer within the OpenEphys GUI, with a characteristic waveform. The artifact repeatedly occurs for several minutes apparently randomly, i.e  irrespective of the animals movement (during sleep and exploration), and in all four LFP channels but in some animals also in EEG channels, though sometimes channels vary in amplitude. 

I’ve attached screenshots of the artifact and a power-spectrum over one entire raw recording (duration: 4 hours) between 0 and 300 Hz as well as a power-spectrum over NREM sleep epochs between 0 and 35 Hz (corrected for dF/F).

Animals are connected to one RHD 32-channel headstage, connected to a standard SPI interface cable. The SPI cable is connected to a custom-made commutator (30 $ conductive slip ring, MST005-22-12A), which is connected to a 1.8m standard SPI cable, which is connected to the Open Ephys Acquisition Board. The overall cable, thus, runs ~2.8 m. I’ve used Version 0.5.5 of the OpenEphys to conduct the recordings.

1.       My best guess is that the artifact might be caused by the commutator (weird signal degradation?), or relatedly, that data transmission between the Headstage and Acquisition Board is faultily, leading to the rhythmic artifact. Has anybody experienced such an artifact or has any idea what the issue might be?

2.       We’ve used the custom-made commutator, because it has a hole in the center, through which optic fibers can be run for optogenetic stimulation (I’ve attached a optical rotary joint above the commutator to release the tension of all cables). Has anybody good suggestions for a combined optical + ephys commutator, that we could try alternatively?

Thanks,

Max

[Jon Newman]

Hi Max,

Hmm, that's an interesting one.

- I kinda doubt it's the commutator, unless it somehow is picking up garbage and getting into the SPI cable's power supply. Other than that, it can only result in digital communication issues which would not look this reasonable.

- 400 ms is a very long time. Given the size and long period, it almost feels as if it could be a large electromechanical noise source in the vicinity of the rig. Something like a duct fan in the ceiling turning on periodically. Or a refrigerator compressor trying to kick on.

- If you are using a cheap DPSS laser for opto stim, they often have hilariously noisy power supplies. The laser and its power supply might be to blame as well. Another option is even the laser itself, which again if it's a cheap DPSS, does not have stable output power and engage in all sorts of funny behavior over time.

In any case, I would approach this using a process of elimination. Does it happen when the commutator is removed? Does it happen in a different room? What about if the laser is unplugged assuming there is one. Can you try a different wall wart? If that does nothing then what about an isolated bench top power supply to potentially illuminate a ground loop.

If in the case that you find the noise source and there is nothing you can do about it, this also feels like one that could be solved with a good faraday cage. 

Let us know what you find!

--
You received this message because you are subscribed to the Google Groups "Open Ephys" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-ephys+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/open-ephys/5edb8ff8-1831-4f0d-b3c6-1ebcffc47d0an%40googlegroups.com.

[maximilia...@gmail.com]

Hi Jonathan,

 

Thanks for your message!

Indeed the power supply of the laser is incredibly noisy. I’ve already spent quite some time to ground the laser and the box in which the animals is placed, which improved the signal a lot (but you can still see the power spikes in 50 Hz + harmonics in the power spectra). The reason why I think that it is not an electrical artifact from the room, though, is that I previously recorded animals in the same room configuration with a Neuralynx Digital Lynx SX Amplifier + analogue MillMax Headstage and did not see any similar artifacts. I found this thread (https://open-ephys.atlassian.net/wiki/spaces/OEW/pages/491644/Commutators), in which Rodrigo reported on weird LFP step–like chopped signals, which were solved by reducing the cabling the Acquisition Board and commutator. Rodrigos' artifacts were not rhythmic, however, and the signal looked less prototypical than the artifacts I’m experiencing.  I’m wondering whether the grounding or power supply of the headstage is somehow interrupted periodically, leading to the large negative deflection and subsequent slower return to baseline of the signal… I’m not sure whether this is possible, though. I couldn’t find any information on the transmission frequency between the headstage and the Acquisition Board, which I could compare to the rhythm in which the artifact occurs. What do you think?

However, I will try your suggestions and let you know, if I can get rid of the artifact.

 

Best,
Max

[Jon Newman]

Hi Max,

The reason I dont think its corruption is due to the commutator is because, as far as I understand it, it is a passive slip-ring device. This means the only thing it could do is mess up bits in the digital communication coming from the headstage to the acquisition box. The probability that those flipped bits result in a nice waveform that starts around the DC-level of your recording and then returns back to baseline is ~0. It would look much more like a bunch of uniformly random samples all over the place for the duration of the corruption. The commutator could still be to blame by injecting an unwanted signal into the system such that the electrodes pick it up.

Grounding -- again, you would see a massive artifact where the signal swings out of control hitting the rails of the amplifier, very unlikely.

I'm fairly confident this is an unwanted signal generated somewhere in the room. For all I know it could be coming from the acquisition box itself. Perhaps something is malfunctioning and going into a thermal overload on/off cycle. Have you tried a different acquisition board, if you have access to one?

You mentioned this artifact is not always there. How long does it last when it occurs? Is there a secondary period between the bouts of noise?

Jon

To view this discussion on the web visit https://groups.google.com/d/msgid/open-ephys/fc5d81f0-5589-486f-a196-cf50a4eabac7n%40googlegroups.com.

[maximilia...@gmail.com]

Hi Jon,

thanks for the suggestions! I'm now testing the room for some possible artifact-sources. Unfortunately I have no other Acquisition Box available at the moment, but I'm asking around in the building.

To your questions: It usually occurs for several seconds up to several minutes (not longer than 5 - 10 minutes though). If I understand you correctly, then there is a very consistent secondary period (~every 100 ms) between the bouts of noise, which you can also see in the screenshots.

I agree that if the passive slip ring messes up some bits, this would lead to random (and probably extreme) samples for the time of corruption.  However, could it be that the waveform results from the applied filters to the signal? I don't know whether the digital signal (including possible corrupted bits) from the headstage is converted to analog within the Acquisition Board again... I thought this might be the case, as I can set high and low cut filters on the Intan chip with the bandwidth interface of the Rhythm FPGA. The filters in my recording were LowCut="2.4959882418772361135" HighCut="7603.7651218333703582" by default. Is there a Digital to Analog converter in the Acq. Board?

Best,
Max

[Jon Newman]

Hi Max,

- Those filters are in the headstages prior to digitization.

- Digitization occurs at the headstage and all subsequent processing is digital

- At this point, I will just tell you that this is not due to corruption in digital communication, you need to look elsewhere

- The periodicity and low frequency indicate an external signal source that is switching on for 5-10 minute durations. Good places to look would be HVAC stuff and refrigerator compressors in the vicinity of your recording

- It's possible you can get rid of all of this with a proper faraday cage.

Jon

To view this discussion on the web visit https://groups.google.com/d/msgid/open-ephys/a3fd2ee3-3dd0-4ba6-8e9b-d2ca87b6b352n%40googlegroups.com.