Retina press (for multi-electrode array)

This post describes a cap used to gently push a piece of retina onto the recording area of a multi-electrode array (MEA).

The cap and press were designed by Carola Yovanovich and Daniel Koziowski, originally for use with 3Brain CorePlate MEA chips, although they have also been successfully used on Multi Channel Systems's MEA chips also.

Pressing a retina

If a piece of retina is simply placed on a recording chip, the electrical signals from retinal ganglion cell action potentials are picked up only weakly, if at all. To improve the strength of the signal, the retina can be pressed down.

In the Tom Baden lab, for one of our setups, we use a 3D printed cap to press down the retina. We press down from above, pressing against the photoreceptors. Our MEA is a 3Brain BioCAM X, and we project light onto the tissue from above. To press the tissue against the recording area, whatever we place on top of the tissue must be trasparent. We use the following polyester membranes from Sterlitech:

Before the membrane is glued to the cap, this is what the cap looks like:

This cap is printed with PLA. We have also printed caps with resin, which gives a nicer result but is otherwise functionally equivalent. One benefit of the resin caps is that it's easier to illuminate the central membrane area when under a microscope.

An acrylic cap, after being pressed on a piece of Xenopus retina, is captured under a microscope:

Retinal damage

Pressing against the tissue must be gentle, but, as can be seen in the third image, the process does damage the photoreceptors. In the above example, the cap was removed after 20 hours of recording, and the yellow "dust" you see are the outer segments of the rod photrecoptors, which are large in Xenopus.

Cleaning

After a recording, the membranes usually wash clean; however, sometimes tissue adheres to them or they get indentations. For this reason, the membranes are treated as disposable and are often replaced. I've yet to figure out the best glue to use; currently I'm using an epoxy glue. For washing we cycle with Alconox and Milli-Q water.

Gas bubbles

One outstanding concern is the appearance of gas bubbles near the edge of the membrane. The bubbles were not present when the cap was initially pressed down and must have formed during or after the recording. Luckily, there is a margin between the edge of the membrane and the edge of the electrode array, which is smaller and sits within the membrane window. Even with this margin, the bubbles can affect responses of ganglion cells sitting within the array but whose receptive fields are wide, or it might be the cases that spikes are picked up from axons of ganglion cells sitting in the well but outside the array. When designing the size of the membrane window, making it larger than the size of the array can insure these bubbles are kept at the periphery. This means you want the window to be large relative to the array area. For 3Brain chips, the constraint is the size and shape of the depressed area that 3Brain calls the "exposed chip area" as this area is slightly depressed vertically. For all of the single well chips from 3Brain (as of 2026), the exposed area is listed at approximately 40 mm² (~6.32 mm side length). Array side lengths vary (5.12 mm, 2.67 mm, 3.80 mm), and choosing a larger side length will reduce the padding possible between the window edge and the sensing array. A related parameter you must balance the thickness of the edges of the cap that glue to the membrane, as thinner edges give you a larger window but reduce the surface area for adhering to the membrane.

Meniscus

The sides of the submerged walls of the cap form a meniscus. This slight bending of the liquid surface has a small but not negligible effect on light transmission. Future designs should try and reduce the amount of material in the walls by switching to thin columns. The balance will be reducing the wall material but keeping them strong enough for pressing.

Press

The actual pressing of the cap onto the chip and retina needs fine control. For this, we use a custom mechanical press, which can allow the cap to be pressed onto the retinal with just enough force:

The little fork speeds things up by allowing the cap to be lowered to a safe height quickly. Once that's done, you would then move the press under the microscope. It's very obvious when the membrane makes contact with the tissue. To get high quality spike signals, I find that you want to press enough that the tissue stretches out—for the Xenopus retina, by about 20-40%.

Petri dish as a case

We house the caps in petri dishes, using a ring to keep the membrane suspended and not in contact with another surface. This reduces the chance of the membrane being damaged and makes it easier to clean it.

Adhesive tape inner

The inside circumference of the cap is lined by a 1 mm thick foam adhesive tape. The tape makes a friction fit between the cap and the well of the chip. The fit is loose enough so that the cap can be pressed down, but tight enough that the cap holds its position once in place.

Design files

This cap can be constructed with the following design file and materials:

• cap.f3d
• cap-fused.3mf (single piece cap design)
• cap-fused-alternative.f3d (an alternative single piece design)
• pressing-arm-L-attachment.f3d
• pressing-arm.f3d
• rectangular-chamber.f3d
• petri-spacer.scad

Unfortunately, with different people working on these designs, there is a mix of desgin file types. f3d for Autodesk Fusion, scad for OpenSCAD and the 3mf file where the original design seems to be lost.