Automated Charting of the Visual Space of Housefly Compound Eyes

Summary

The protocol here describes the measurement of the spatial organization of the visual axes of housefly eyes, mapped by an automatic device, using the pseudopupil phenomenon and the pupil mechanism of the photoreceptor cells.

Abstract

This paper describes the automatic measurement of the spatial organization of the visual axes of insect compound eyes, which consist of several thousands of visual units called ommatidia. Each ommatidium samples the optical information from a small solid angle, with an approximate Gaussian-distributed sensitivity (half-width on the order of 1˚) centered around a visual axis. Together, the ommatidia gather the visual information from a nearly panoramic field of view. The spatial distribution of the visual axes thus determines the eye's spatial resolution. Knowledge of the optical organization of a compound eye and its visual acuity is crucial for quantitative studies of neural processing of the visual information. Here we present an automated procedure for mapping a compound eye's visual axes, using an intrinsic, in vivo optical phenomenon, the pseudopupil, and the pupil mechanism of the photoreceptor cells. We outline the optomechanical setup for scanning insect eyes and use experimental results obtained from a housefly, Musca domestica, to illustrate the steps in the measurement procedure.

Introduction

The compactness of insect visual systems and the agility of their owners, demonstrating highly developed visual information processing, have intrigued people from both scientific and non-scientific backgrounds. Insect compound eyes have been recognized as powerful optical devices enabling acute and versatile visual capacities^1,2. Flies, for instance, are well-known for their fast responses to moving objects, and bees are famous for possessing color vision and polarization vision².

The compound eyes of arthropods consist of numerous anatomically similar units, the ommatidia, each of which is capped by a facet lens. In Diptera (flies), the assembly of facet lenses, known collectively as the cornea, often approximates a hemisphere. Each ommatidium samples incident light from a small solid angle with half-width on the order of 1˚. The ommatidia of the two eyes together sample approximately the full solid angle, but the visual axes of the ommatidia are not evenly distributed. Certain eye areas have a high density of visual axes, which creates a region of high spatial acuity, colloquially called a fovea. The remaining part of the eye then has a coarser spatial resolution^{3,4,5,6,7,8,9}.

A quantitative analysis of the optical organization of the compound eyes is crucial for detailed studies of the neural processing of visual information. Studies of the neural networks of an insect's brain¹⁰ often require knowledge of the spatial distribution of the ommatidial axes. Furthermore, compound eyes have inspired several technical innovations. Many initiatives to produce bio-inspired artificial eyes have been built on existing quantitative studies of real compound eyes^11,12,13. For instance, a semiconductor-based sensor with high-spatial resolution was designed based on the model of insect compound eyes^{11,14,15,16,17}. However, the devices developed so far have not implemented the actual characteristics of existing insect eyes. Accurate representations of insect compound eyes and their spatial organization will require detailed and reliable data from natural eyes, which is not extensively available.

The main reason for the paucity of data is the extreme tediousness of the available procedures for charting the eyes' spatial characteristics. This has motivated attempts to establish a more automated eye mapping procedure. In a first attempt at automated analyses of insect compound eyes, Douglass and Wehling¹⁸ developed a scanning procedure for mapping facet sizes in the cornea and demonstrated its feasibility for a few fly species. Here we extend their approach by developing methods for not only scanning the facets of the cornea but also assessing the visual axes of the ommatidia to which the facets belong. We present the case of housefly eyes to exemplify the procedures involved.

The experimental setup for scanning insect eyes is: partly optical, i.e., a microscope with camera and illumination optics; partly mechanical, i.e., a goniometer system for rotating the investigated insect; and partially computational, i.e., use of software drivers for the instruments and programs for executing measurements and analyses. The developed methods encompass a range of computational procedures, from capturing images, choosing camera channels, and setting image processing thresholds to recognizing individual facet locations via bright spots of light reflected from their convex surfaces. Fourier transform methods were crucial in the image analysis, both for detecting individual facets and for analyzing the facet patterns.

The paper is structured as follows. We first introduce the experimental setup and the pseudopupil phenomenon-the optical marker used to identify the visual axes of the photoreceptors in living eyes^19,20,21. Subsequently, the algorithms used in the scanning procedure and image analysis are outlined.

Protocol

The protocol is in accordance with the University's insect care guidelines.

1. Preparation of a housefly,  Musca domestica

1. Collect the fly from the laboratory-reared population. Place the fly in the brass holder (Figure 1).
   1. Cut 6 mm from the upper part of the restraining tube (see Table of Materials). The new upper part of the tube has an external diameter of 4 mm and an internal diameter of 2.5 mm (Figure 1A). Place the live fly inside the tube, seal the tube with cotton to prevent damaging the fly, and push the fly such that the head protrudes from the tube and its body is restrained (Figure 1B). Immobilize the head with beeswax such that the eyes remain uncovered (Figure 1C-E).
   2. Cut the tube again such that the tube length is 10 mm (Figure 1C). Place the plastic tube with the fly in the brass holder, such that one eye of the fly is pointing upward when the holder is resting on a tabletop (Figure 1D,E).
2. Adjust the orientation of the tube such that with the goniometer elevation at 0° (i.e., the azimuth stage is in a horizontal position), the vertical illumination beam of the microscope is perpendicular to the eye surface in a central region, between ventral and dorsal, and between anterior and posterior edges of the eye, so that the whole eye can be scanned within the range of azimuth and elevation allowed by the setup.

2. Alignment of the goniometer's rotating azimuth axis with the microscope optical axis

1. Mount an alignment pin on the azimuth rotation stage so that the x-y position of the tip can be adjusted to coincide with the azimuth axis on the motorized stage. While viewing with the microscope, equipped with a 5x objective, focus on the tip using the z-axis joystick (Figure 2).
2. Align the x-y adjustment of the azimuth axis with the microscope's optical axis and ensure that the elevation and azimuth rotary axes are pre-aligned with the centered pin, using the x- and y-axis joysticks.
3. Manipulate the azimuth and elevation joysticks to check whether the pin is centered with respect to both degrees of freedom. When well-centered, the pin tip remains in, approximately, the same position during azimuth and elevation rotations.

3. Alignment of the fly eye with the motorized stages

1. With the elevation stage at 0°, mount the fly and its holder on the azimuth stage. Observe the fly's eye with the microscope.
2. With the illumination LED on, adjust the horizontal position of the fly so that the center of the pseudopupil is aligned with the microscope. Adjust the vertical position of the fly by using the rotating screw of the holder (Figure 1D), so that the deep pseudopupil (DPP; Figure 3)^19,20,21 is brought into focus at the level of the elevation axis.
3. Align the DPP with respect to the azimuth and elevation axes by centering it in the field of view (see Figure 2). Use the magnets glued to the bottom of the fly holder to affix it firmly to an iron plate mounted on the azimuth stage, while permitting manual sliding adjustments.
   1. Switch the view to the digital camera mounted at the microscope. Run the software initialization of the GRACE system, which includes initializing the motor controllers and the Arduino LED controller (Figure 4). Therefore, open MATLAB R2020a or higher version. Run the MATLAB script Initialize_All_Systems (Supplementary File 1).
4. Confirm whether the fly's pseudopupil (Figure 3B,C) is at the center of the projected image on the computer screen.

4. Autofocusing and autocentering

1. Bring the focus to the level of the corneal pseudopupil (CPP; Figure 3B)^19,20,21 manually by using the z-axis joystick.
2. Run the autofocusing algorithm (Supplementary File 1, script AF) to attain a sharp image at the cornea level. Check by returning the focus to the DPP level by adjusting the motorized z-axis stage. Store the distance between the DPP and CPP (in motor steps).
3. Fine-tune the pseudopupil centering by running the autocentering algorithm (Supplementary File 1, script AC). Bring the focus back to the CPP level.
4. Re-run the autofocusing algorithm. Zero the motorized stages at their current positions (X,Y,Z,E,A) = (0,0,0,0,0), where E is elevation and A is azimuth.
5. Run the scanning algorithm (Supplementary File 1, script Scan_Begin), which samples eye images along trajectories in 5° steps, while performing the autocentering and autofocusing algorithms.
6. At the conclusion of the sampling, turn off the LED Controller, and the motor controllers.
7. Process the images by applying the image processing algorithms (Supplementary File 1, script ImProcFacets).

Results

Animals and optical stimulation
Experiments are performed on houseflies (Musca domestica) obtained from a culture maintained by the Department of Evolutionary Genetics at the University of Groningen. Before the measurements, a fly is immobilized by gluing it with a low-melting-point wax in a well-fitting tube. The fly is subsequently mounted on the stage of a motorized goniometer. The center of the two rotary stages coincides with the focal point of a microscopic setup^24<...

Discussion

The spatial distribution of the visual axes of housefly eyes can be charted using the pseudopupil phenomenon of compound eyes and the reflection changes caused by the light-dependent pupil mechanism. Therefore, an investigated fly is mounted in a goniometric system, which allows inspection of the local facet pattern with a microscope setup equipped with a digital camera, all under computer control. Image analysis yields eye maps. An essential difficulty encountered is that without careful positioning of the eye at the be...

Materials

Name	Company	Catalog Number	Comments
Digital Camera	PointGrey	BFLY-U3-23S6C-C	Acquision of amplified images and digital communication with PC
High power star LED	Velleman	LH3WW	Light source for observation and imaging the compound eye
Holder for the investigated fly	University of Groningen		Different designs were manufactured by the university workshop
Linear motor	ELERO	ELERO Junior 1, version C	Actuates the upper microscope up and down. (Load 300N, Stroke speed 15mm/s, nominal current 1.2A)
Low temperature melting wax	various		The low-temperature melting point wax serves to immobilize the fly and fix it to the holder
Microscope	Zeiss		Any alternative microscope brand will do; the preferred objective is a 5x
Motor and LED Controller	University of Groningen	Z-o1	Designed and built by the University of Groningen and based on Arduino and Adafruit technologies.
Motorized Stage	Standa (Vilnius, Lithuania)	8MT175-50XYZ-8MR191-28	A 6 axis motorized stage modified to have 5 degrees of freedom.
Optical components	LINUS		Several diagrams and lenses forming an epi-illumination system (see Stavenga, Journal of Experimental Biology 205, 1077-1085, 2002)
PC running MATLAB	University of Groningen		The PC is able to process the images of the PointGrey camera, control the LED intensity, and send control commants to the motor cotrollers of the system
Power Supply (36V, 3.34A)	Standa (Vilnius, Lithuania)	PUP120-17	Dedicated power supply for the STANDA motor controllers
Soldering iron	various		Used for melting the wax
Stepper and DC Motor Controller	Standa (Vilnius, Lithuania)	8SMC4-USB-B9-B9	Dedicated controllers for the STANDA motorized stage capable of communicating with MATLAB
Finntip-61	Finnpipette Ky, Helsinki	FINNTIP-61, 200-1000μL	PIPETTE TIPS FOR FINNPIPETTES, 400/BOX. It is used to restrain the fly
Carving Pen Shaping/Thread Burning Tool	Max Wax		The tip of the carving pen is designed to transfer wax to the head of fly
MATLAB	Mathworks, Natick, MA, USA	main program plus Image Acquisition, Image Analysis, and Instrument Control toolboxes.	Programming language used to implement the algorithms