Home
This walk-through will help you build a camera-based FIP system as was first presented by Christina Kim in this paper and simplified by us. With this system you can record from up-to 7 fibers simultaneously. The system is highly flexible and can be conveniently adapted to record from different color fluorophores or for simultaneous optogenetics.
Note: If you don't feel like building your own system, a pre-build version of this system is available from NeuroPhotometrics.
See the "parts_list" Excel sheet on this same Github.
You are essentially building a rudimentary epifluorescence microscope. A computer (Arduino, or National Instruments pulse generator) generates alternating pulses that trigger LEDs. That same pulse generator generates the pulses that trigger a camera to take pictures of a fiber bundle. Individual fibers are then cropped from these pictures and the pixel intensity is recorded.
Start by assembling the filter sets and the filter cubes. In this example we are using a standard set to look at a GFP-derived sensor.:
- Filter cubes: DFM1.
- As a cost-saving measure, it is possible to 3D print these cubes, the CAT files are on the ThorLabs website.
- 405nm excitation bandpass filter FBH405-10
- If you use wider bandpass filters (say 35nm) you can get higher light output, but it's easy to get cross-channel leakage on multi-color setups.
- 470nm excitation bandpass filter FBH470-10
- Longpass filter (dichroic) DMLP435R
- This will reflect 405nm light and so blend 470nm and 405nm light.
- Longpass filter (dichroic) FF495-Di03
- This will reflect 405nm and 470nm excitation light into the brain while letting the 510nm (GFP Emission) light pass through to the detector.
- A 525nm emission bandpass filter MF525-39
The precise filter set will depend on your particular needs and the fluorophore you are interested in. Note that Thorlabs or other supliers will often sell you a complete filter set at a discount, so you don't have to to mix-and-match yourself. For instance, see the bottom of the page here.
(Click on the image for a larger version.)
After you undo the setscrew using a hex key, you should be able to take out the inserts of the filter cubes. If you cant, your filters might be screwed too deep into the cubes (which is why we recommend that you install the mirrors first). When inserting the dichroic mirrors into the insert, make sure not to break them. You prevent this from happening by properly loosening the 'clips' that are inside the cubes using a small Phillips screwdriver.
Installing the filters into the cubes is fairly straightforward. Note that they have a preferred light direction which is indicated by an arrow.
The collimators and collimator-adapters serve to hold the fibers that guide the LED light and focus the light into the system. In a future version I want to 3D print of build something that will let me mount the LEDs directly at the filter cubes. Here we use:
- Collimator F671SMA-405
- I don't think it matters much which wavelength you get for this particular application.
- Collimator holder AD11F
- A 20x objective
- New objectives are very expensive. But a lightly beat-up objective from Ebay will do totally fine! Make sure to get an adapter that will let you screw the objective into the filter cube. For instance, for this Nikon PLAN APO (Ebay, $349), I used this adapter SM1A12.
- The exact lens that you will be using depends on the size of the fiber bundle that you want to image (use a calculator like this one). Here I use AC254-100. Which is fine for a two-fiber setup (one fiber is slightly off-screen, but that is ok).
The cage is what let's you move the adapter that contains the fiber bundle, this is how you bring the fiber bundle into focus. It consists of:
After putting the cage together, you can do fine focus adjustments of the fiber bundle by undoing the set screws on the CP33 cage plate and moving it closer or further away from the objective.
If you are using the provided Arduino code, the outputs are by default:
- Camera: 10
- LED 470: 51
- LED 405: 50
If you are using an Arduino Uno instead of a Mega, change the outputs for the LEDs in the Arduino code before uploading.
NOTE: currently the Arduino version of the acquisition code works in conjunction with a USB DAQ (NI USB-6001). You split the camera trigger into channel 4 of the DAQ. The reason we do this, is that the internal clock of the NI system is better than that of an Arduino. There are more details about this around line ~512 om the "FIP_acquisition.m" code.
If you are using a NI card that has "counter output channels" such as the NI PCIe-6343, the default channels are:
- Camera: PFI 0
- LED 470: PFI 2
- LED 405 PFI 1
To be sure: if you run FIP_acquisition for the first time, the program will print the ports that you should use to the MATLAB command line. To find out how these ports map to the terminals on your breakout board use an overview of the pin layout like the one on page 15 of this manual (or just Google "PCIe-6320 Pinout").
Regardless of what NI system you are using, the FIP_acquisition code is setup to automatically record all inputs on analog input (AI) channels 0-7. If you are not using all 8 channels, you might see "ghosting" on the unconnected channels. This is not a problem, but you can prevent it by simply connecting the channels that you are not using to ground or by changing the code (see the daq_setup method) to ignore those channels.
Essentially you shine light into the fibers, or use the LEDs in combination with something fluorescent. Next you move the camera over the breadboard until you have the fibers in focus. The best is to do this in a dark room. For precise focusing of the fiber bundle, slide the adapter closer or further away from the objective.
A disadvantage of these systems over systems that use a lock-in amplifyer or other modulation is that they are quite sensitive to ambient light. Make sure to make your system completely dark using 3D printed parts, or just black tape and aluminium foil.
