All About Imec Probes

Part Numbers -> Types

SpikeGLX, CatGT, C_Waves use the probe part number to load attribute data from an internal database, code here.

The main categories are these:

• NP 1.0: First-gen rodent probes 10 mm length
• NP 2.0: Next-gen rodent probes 10 mm length (linear sites with closer z-pitch)
• NHP: Non-human primate, thicker shanks, {10, 25, 45} mm lengths.
• UHD: Ultra high density sites for ground-truth studies.
• OPTO: Integrated light emitters on shank (under development).

For curious programmers, SpikeGLX software internally describes an abstract probe base class, and the type codes shown in the following tables correspond to probe subclasses (but even those still have some parameters). The type is shown in the GUI and in metadata primarily for SpikeGLX diagnostic reasons. If you are implementing your own software to classify probes, just be aware that each different part number has been issued by imec because some probe attribute is different, so you should go by part number.

These are the curently supported probe varieties.

Rodent, NP 1.0 imro format

NHP, NP 1.0 imro format

Rodent, NP 2.0 type 21 imro format

Rodent, NP 2.0 type 24 imro format

UHD, NP 1.0 imro format

UHD, type 1110 imro format

OPTO, NP 1.0 imro format

Geometry

Probes have either one or four shanks.

The sensors on the shanks are called {electrodes, sites, pads} and occupy only the front face of the probe.

Some probes have a staggered site layout, some are linear.

The indexing of {shanks, electrodes, channels} is zero-based.

On 4-shank probes, electrode indices start over again at zero on each shank, see figure:

Indexing of Shanks and Electrodes

Channels and Channel Settings

The width of the shank limits how many physical wires can run along its length. Consequently, we can currently read out a maximum of 384 electrodes from a shank. Most probes have many more electrodes than that, so the probes contain switches to select which electrodes are connected to the read-out channels.

There are some exceptions to this:

• Probes {NP1200, NP1210} only have 128 electrodes (and channels).
• Some UHD probes contain only 384 electrodes, so there are no switches.

But generally, there are 384 read-out channels, and for each channel there are some user adjustable parameters:

• Which site is connected.
• Which reference is applied for differential voltage measurements.
• The gain of each band.
• If AP bandpass filtering is applied.

All of these settings are managed by the imec read-out table editor (imro editor), discussed later.

Banks and site selection

For most probes, the imec API function call that connects a channel to a site needs to know the (shank, bank) of the target site. The shank index is in the range [0..3] as shown in the figure above (single-shank probes only have index-0).

Banks are groups of 384 sites. The set of 384 sites nearest the tip (lowest) compose bank 0. The next set is bank 1, and so on, up to the maximum bank index for that probe. Note that some probes have a non-integral number of banks. For example, the NP1000 has 960 sites. 960/384 = 2.5 banks. In this case, some channels can be set to (shank,bank) = {(0,0), (0,1), (0,2)}, while others can only be set to {(0,0), (0,1)}.

The probes fall into four different categories with respect to their internal wiring diagrams and switch matrices. These categories correspond to type codes {0, 21, 24, 1110}. For each category there are selection rules describing which sites can be connected to a given channel. Generally, each site can be connected to only one read-out channel.

Selecting sites and obeying the rules is simple when using the imro editor. You will click on a picture of the probe to select blocks of sites at a time. As you select blocks remaining areas are blackened to show they are forbidden according to the selection rules for ths probe. The available space fragments as you place more blocks into it. You are only permitted to place new blocks in allowed regions.

If you want to roll your own custom imro table files that can be loaded into SpikeGLX, use the tools in the IMRO_Tables/MATLAB_Scripts subfolder of your download package.

Open Ephys uses our formats for imro tables , so you can use our tools to create imro tables and run them with Open Ephys GUI.

Type 0 wiring

This has a simple internal single-shank structure. If you think of the 384 sites in a bank as a 1-d array: {i}, i = 0,383, then channel i can connect to sites: i + 384*b, b = 0,max-bank.

Type 21 wiring

Experimental wiring was developed for the 2.0 single-shank probe making it possible to connect more than one electrode to the same channel. The idea was to improve read-out speed and unit yield by adding sites together. Untangling who was who was done in post-processing. To make it easier to disentangle, the channel-site mappings within each 32-site block are scrambled. This works but at the cost of reduced SNR. Connecting more than one site to a channel is not recommended for normal use and the graphical tools in SpikeGLX do not support that. Nevertheless it is still possible to do that by writing your own imro file.

Lee, Kyu Hyun;Ni, Yu-Li et al. (2021) Electrode pooling can boost the yield of extracellular recordings with switchable silicon probes Nature Communications

The block scrambling causes more fragmentation than with other probes, hence, the graphical editor limits how small a block size you can select for these probes.

Type 24 wiring

A 48-channel block structure is used to allow selected sites to be spread over all four shanks. This causes a different fragmentation pattern than seen with other probes.

Type 1110 wiring

UHD probes have another novel imro data structure. Technically they allow some site selection patterns, like long stripes, that are accessible if you write your own files using the tools in IMRO_Tables/MATLAB_Scripts.

Reference choices

Bands and gains

Bandpass option

Imro Table, Editing, Files

Survey Mode

Other Maps

Output Data Format

• Channel order same
• Sync Channel
• Metadata

Sharpening

• Wade
• Tim's protocol
• Emily protocol -> Nick protocol.

Reference, Ground Wires, Soldering

Runs, Settle Time

• Start run early, let electronics warm and settle.

• Don't jiggle the ground wires, or let the animal transition between grounded and isolated. This causes amplifiers to saturate and resettle.

Testing, Failure Modes

Light Sensitivity

Cleaning

fin