<Spikeling v2.0 an artificial neuron for neuroscience enthusiasts!>

Order PCBs:

Compare PCB Prices:

Buy Parts

Adjust quantity:
%
 
... 9 more lines
View All

Spikeling V 2.0

Spikeling V2.0 is an educational tool for neuroscience students and enthusiasts!

It is an artificial neuron that can receive different inputs, integrate them and send and outputs its computation, just like a spiking neuron would!

Technically, it consists on a microcontroller (an ESP32) running the computationally efficient Izhikevich model of a spiking neuron.

This project is licensed under the GNU General Public License v3.0
The hardware is licensed under the CERN OHL v1.2



If this all sounds familiar to you, it is because this project is a derivative (and update) of the Spikeling project started at the BadenLab. With feedback from users and students, we are developing this version, which has added capabilities, and also revisits how lessons can be taught, and what kind of experiments can be done with it!

While this is not a stable version just yet, enthusiasts and potential collaborators can source the electronic bits for this project from KitSpace


The main upgrade on the PCB concerns the use of an ESP32 module and 2 ADC 8bit expander chips that together allow spikeling to compute more input output with specific input/output ports and dedicated potentiometer. The previous version had limited ports and potentiometers and different functions had to be encoded on the arduino code hence impacting the versatility of the model.

Stereo (3-ways) audio jacks 3.5mm replace the previous BNC cable. Both ports and cables are therefore cheaper, also ports are less bulky and cables more flexible. Moreover, now one single port can carry two distinct information. In this case digital and analog input/output.

Since the user cannot access the microcontroler board anymore (acrylic sheet on top of the board), a reset button has been added to the back of the board.

Previously, "modes" were encoded manually on the microcontroler, and the user could check the set mode number by the amount of flashes the microcontroller built-in LED was providing at the begining of the routine. Now, the microcontroller is pre-set with 12 distinct "neuron modes" that the user can switch through, 12 proxy LED indicate continuously which mode is currently being set.

The buzzer is now smaller and is powered by the ESP32 3.3V (instead of the 5v from the arduino), making its clicking noise quieter, which is more pleasant for the user ear, especially after a couple of minutes using the device.

A couple of functions have been added to the board and can now be controlled through potentiometers (synaptic polarity(inhibitory/excitatory), photoreceptor gain and polarity, stimulus strength and polarity(only for current input not LED stimulation)).

Another major upgrade concerns the use of the two DAC GPIO from the ESP32. These two are conencted to the synapse out analog, which can be read either by another Spikeling synaptic in analog input port, or either by an external oscilloscope. The other DAC port is connected to the stimulus analog out, whcich drives the Current-in port (from the patch clamp module), therefore improving the direct stimulation from the board.


The entire layout of the Spikeling board has been reconceived and a laser-cut acrylic sheet repesenting a neuron now sit on top of the board. All ports and potentiometers are now strategically placed.

The ESP32 is placed below the soma, the synaptic inputs from other Spikeling units are placed on the left, the synaptic output is placed on the right. Along the axon now sits a RGB LED, the red LED brightness follows as before the Vm status of the neuron, and the LED sparks in white when the neuron spike. The spike buzzer also sit along the axon, a small hole on the acrylic sheet sitting just above so the click can be heard coming from that direction.

The Vm potentiometer and the Current-in port are now grouped together and sit on the acrylic sheet at the bottom of an electrode pipette. They can now be considered as a voltage clamp experiment variables.

A photoreceptor is now drawn on the acrylic sheet, below the opsin sits a photodiode, and a potentiometer on the photoreceptor body now controls the photoreceptor sensitivity and its polarity (the user decides then if the PR has an excitatory or an inhibitory effect on the neuron)

The noise potentiometer now sits in a box by itself as it represents parasitic noise from the experiment environment and is independent from the rest of the Spikeling functions. This potentiometer is different from the others as it is not center-detented as it generates noise from a zero to a maximum value.

Next is the Neuron mode box which contains the 12 available modes to the users and a push button that allows the user to switch between them.

Finally, the last box contains all experimental tools allowing stimuli generation. As detailled on the acrylic sheet, the user can control the stimulus frequency and the stimulus strenght, along with its polarity. The stimulus output can either be directly connected to the current in port to simulate a current applied to a voltage held neuron through the pipette. Or be connected to a cable with a 5mm LED soldered at its tip which can be placed directly on top of the photoreceptor opsin (photodiode below). The photoreceptor possesses functions (decay/recovery) simulating how a real photoreceptor would integrate a light stimuli and how it will adapt to prolonged stimulation.


Spikeling now possesses its own GUI, written in python and using the latest pyQt6 library. The script then uses pyInstall so it can be translated into a desktop app, ready to be easily distributed to users.

The GUI is divided (for the moment) into distinct tabs. The first one is the main GUI window where the user connects the board to the computer, can choose to either select the neuron mode on the board directly or through the GUI (note that custom modes will have to be selected from here, and in this instance the GUI takes priority over the board commands).

The main window display the Spikeling activity (Neuron Vm, total current input from all sources & stimulus state). If connected to other spikeling, the GUI can display fore both synaptic input, their incoming neuron Vm and their spiking event, which will be translated into input current for the main neuron. Traces can be selected through checkbox and superimposed over the same graph with a common timeline.

Below the main window, Spikeling traces can be recorded either live or for only a fixed amount of stimulation loop.

On the right hand side two control columns can be found. For all commands, when the checkbox is clicked, the GUI takes over the potentiometer and controls directly the Spikeling variable. This allow advanced user to finely design an experimental protocol for example.

Below the stimulus frequency and strength sliders can be found a custom stimulus display. Here the user can choose to use instead of the classical square wave, either a stimulus from a pre-design library (comprising sine wave, chirp, white noise, etc.), or either a custom made stimulus generated from the GUI "stimulus generator" tab.

One the first Spikeling version, if the user wanted to change the set variable for the photoreceptor and synaptic set variable (gain, decay, recovery, polarity, etc.), one had to manually reupload the arduino code with these changes, and no proper experimental protocol could be establish. Here by clicking the checkbox, the user had the opportunity to change these value directly and therefore is able to play further with each Spikeling parameter, opening the way for teacher to come up with their own personal exercises.


In this tab, the user can play directly with the Izhikevich model. A short text from Izhikevich publications explain the role of each variable (later upgrade will include hyperlink to these publication along with pop up window detailing further what the code is actually doing).

The main window computes the code with a current input of +15 (as in Izhikevich publication). User can change the 4 variable of the code and display the resulting "neuron mode". This is also where user can come up with its own neuronal modes to experiment on if the 12 pre-set one are not enough. This is also where teachers can generate their own custom made neuron, save them, and impose them as experimental model for i.e. homework exercises.


Stimulus generator and data analysis tabs needs to be finished (cf. development log) but they will include ... a stimulus generator, allowing the user to come up with custom made stimulation loop, and ... a data analysis routine for early computation course where student could upload their recorded traces and extract information from it, generating raster plot, computing data and generate basic statistics. For advanced courses an integrated python exercise manual (no IGOR !!!) will guide students through creating a data analysis pipeline using .csv files saved from the main window.


Development log

The negative photogain does not decay towards zero, an absolute value has to be implemented

  • Serial reading is slow and pyQtgraph does not update fast enough the serial values.

FIXED: The entire code was reimplement using a lower BaudRate (9600 instead of 115200) and the pyQtgraph routine was improved. Currently the GUI functions are being reimplemented with the PySerial readWrite functions. Updates to come.

  • The audio jack footprint is not correct. the back pin and holes are mm ahead of their correct position.

FIXED: the new PCB in the 2.2c version now posses the correct footprint. Previous version will need to cut the plastic pieces in front of the audio jack female and force the component into position

  • The LEDs on the board might be to Bright

The resistance of the 4 CharliePlex and the SpikeLed resistors must be adjusted to the acrylic material chosen for the front cover. The current 100 Ohm is sufficient for opaque PMMA, but clear transparent ones should have 1 kOhm soldered below so that the LEDs do not affect the user vision.

  • The CharliePlex footprint orientation (+/-) is not labeled

  • Stimulus generation tab to be reimplemented

  • Data Analysis tab to be reimplemented

  • About tab to be created

The custom 1, 2, 3 text has to be changed to the last three neuron mode implemented on the arduino code

FIXED


Cost estimation spreadsheet for productionhttps://docs.google.com/spreadsheets/d/1JO1rpfb5DoqT5x59RL3RnkQFxRdzm5al6X8uEcsVFo8/edit#gid=0