Experimental Section/Methods

Fabrication of the DOME

The DOME is an assembly of parts 3D printed in PLA plastic. The original device was printed using an Ultimaker 2+ printer, and has been replicated using an Anycubic i3 Mega, a lower specification machine. Parts were designed using Autodesk Inventor and Fusion 360 and can be printed without requiring dissolvable supports. Z-plane focusing is achieved using a linear rail set, where manually rotating the lead screw raises or lowers the sample stage with respect to the imaging lens and camera. An x-y translational stage is attached to the sample stage to allow easy sample adjustment in this plane. This affects only the positioning of the sample and has no bearing on the relative positions of the optical components. The imaging column faces upwards towards the sample stage from the base of the DOME and contains an optical filter holder on a pivot hinge and printed internal threads to attach magnification lenses. The imaging column has two different attachments that can be used, one for a lower magnification option in which only a tube lens is required and an extended version for higher magnification, where a microscope objective can be attached.

Optical set up

The digital light projector (DLP) module is fixed on the sample stage, and thus is unaffected by any adjustment in z-plane focus. It is instead focused independently using a screw lever attached to the projector. Light from the projector is focused by a condenser lens (50mm diameter PCX condenser lens, Edmund Optics), resulting in a total projection size of 14.5 mm × 26 mm. A white LED (RS Components) can also be attached behind the condenser lens to act as a brightfield light for standard microscopy. To provide an illumination across the sample, a ground glass diffuser (Thor Labs) is placed between the LED and condenser lens. Note that although the DOME is capable of bright field illumination, this feature was not used in the experiment presented here. A camera (Camera Module V2, Raspberry Pi) sits on the base of the imaging column pointed upwards at the sample stage. Optics such as wavelength or neutral density filters can be added into the optics holder within the imaging column on an application specific basis. For the lower magnification configuration, the imaging column ends with a 9× tube lens (Eyepiece Cell Assembly, Edmund Optics) screwed into a threaded cylindrical casing. For higher magnification applications the cylindrical section is extended, ending in an RMS thread to fit a standard microscope objective (finite conjugated 10X Semi-Plan Standard Objective, Edmund Optics). While the length of this lens piece is specific to the lenses used here, due to the modular nature of the DOME it would be trivial to adjust this dimension to suit alternative optics. Positioning both the camera and projector perpendicular to the sample stage results in significant lens flare through which imaging is difficult. To circumnavigate this the projector is angled at 10°, positioning the bright spot created by the light source of the projector out of the camera field of view (FOV).

Characterisation of imaging and projection modules

The modular design of the DOME allows for interchangeable levels of magnification using a tube lens and an RMS threaded tube to mount different microscope objectives. A low magnification of 9X is suitable for larger microsystems of the order of hundreds of microns in size such as multi-cellular algae, while a higher magnification of 90X is appropriate for smaller agents such as mammalian cells or bacteria.
            To assess the imaging and projection capabilities, we began by comparing the 9X and 90X magnification settings and used Volvox as an example subject. Volvox are an algae 350–500 µm in diameter where a single spherical colony houses up to 50,000 cells. At 9X magnification, many colonies can be seen in low detail whereas at 90X magnification only a few at are visible but smaller features such as daughter colonies within the body of each Volvox are clearly seen (Figure 1e). A scale for both magnifications was calculated by imaging a measuring ruler.
Next, we considered light projection. When light from the projector is focused through the condenser lens, the total projection area on the sample stage is 14.5 mm x 26 mm, making each projected pixel theoretically 30 µm2 in size. Due to differing FOVs for each magnification, this fixed projection area leads to a trade-off with the number of projected pixels that are visible to the camera (300 x 300 pixels for 9X, and 88 x 66 pixels for 90X magnification).
To test the precision of projected light patterns, a series of line triplets of differing size and spacing were projected onto a neutral density filter (Supplementary Figure 1a,b). The resulting camera images were then analysed by averaging the intensity for each pixel row (Supplementary Figure 1c,d). High precision projected patterns would result in clear differences in light intensity for even closely spaced lines. For 9X magnification, lines of 1-pixel width (30 µm) were difficult to distinguish, with improvements at separations of 2 pixels and clear differences at separations of more than 3 pixels (90 µm). For the 90X magnification, distinct peaks are seen for lines separated by just 2 pixels (60 µm). Measuring across the peaks for each line set in pixel distance and multiplying by the scale factor allows direct measurement of the observed projector pixel size. At both magnifications this agreed with the theoretical projector pixel size of 30 µm2.

Light spectra measurements

Another key feature of the DOME is the ability for each projected pixel to have a different colour. This offers the means to provide multiple light-signals to different agents and supports multiplexed communications for more complex behavioural control. This capability is possible due to the projector containing three separate LEDs for red, green and blue light. As living microagents are often sensitive to limited wavelengths of light, we characterised the light spectra of each LED separately. The light spectra produced by the projection module was performed using a calibrated spectrophotometer (Ocean Optics). To collect the readings the optical fibre used for measurement was attached to the DOME at the sample plane facing upwards. The projector was set to a full screen display where all pixels had value (0, 0, 255), (0, 255, 0) or (255, 0, 0) respectively for red, blue and green.