Scanning Electron Microscope Eye Structure Analysis

Joseph R. Lust
Optics 307, University of Rochester, Institute of Optics,
Rochester, NY 14627



Through eyes and sight animals are capable of perceiving their surroundings and in turn are able to interact with their environment and each other. In this project eyes of three different animals were imaged and their constituent structures analyzed in the University of Rochester’s LEO 982 SEM. The subjects were a mouse (Peromyscus maniculatus), a Starling (Sturnus vulgaris), and a fish (Betta splendens). Following exhaustive preparation, special attention was given to the components of the eye's imaging, photon detection, and neural interconnection systems. The histology was found to be comparable to that of the literature and the images taken amazing in their copious biological detail.




Images provided by the eye allow organisms to make myriad judgments of their surrounding which in turn stimulate decision making. The perception of hue, brightness, contrast, texture, and patterns allow identification and differentiation of predator, prey, and mate alike. Though sent, sound, taste and touch also facilitate perception, such senses are limited in their range as well as their resolution. The acuity of sight in most animals is however relatively high and thus necessitates more neural bandwidth than any other sensing organ, leading to large optic nerve interconnects located relatively close to the brain, and hence often in the head. Thus three major components that allow the eye to carry out this task are the lens system, photon detection system, and neural interconnect.




Eyes were extracted from three different specimens: (1) Betta Fish (Betta splendens), (1) Starling (Sturnus vulgaris), and (2) White Mouse (Peromyscus maniculatus). All animals were sacrificed in accordance with ARVO practices and humane and respectful treatment with exception to the Betta Fish which died from natural causes. All samples were fixed in 2.5% gluteraldhyde water mixture to prevent bacteria growth and prevent decay of retinal cellular constituents which would otherwise occur in 1 – 2 hours. All viable eyes were excised and the remains disposed of by appropriate means.


A. Drying

 Samples were dried to remove all volatiles from the material. The SEM is a high vacuum environment (10-6 Torr) and thus requires that all samples be void of water or other low vapor pressure fluids. While drying biological samples, the surface tension differential created when fluid evaporates from the atmospheric side of a membrane before it does on the internal side will cause the surface structures to collapse and morphology characteristics to be damaged. The following two methods were used to preserve the structures of the eye tissues.


1. Critical Point Drying (CPD)

Samples were first dehydrated through a six step ethanol bath (30%,50%,70%,80%,95%,100%) where all water was exchanged from the sample to the ethanol bath. The samples were then placed in the CPD baskets and inserted into the CPD. The CPD was used to exchange the secondary fluid (ethanol) with the transition fluid (CO2) through iterative filling and purging of the CPD bomb with liquid CO2. The secondary fluid is miscible with both the initial fluid (Water) and the transition fluid (CO2). Water and CO2 however are in no way miscible, hence the need for CPD. Using high heat and pressure, the liquid CO2 saturated samples were navigated around the Critical Point of CO2, 1072 psi and 31.1° F, keeping the samples in the fluid stage until they passed the critical point, where they instantly transitioned from liquid to gas phase without the damaging side effects of air drying. Additionally it should be noted that the reason this cannot be done with the preexisting water in the samples is that the critical point of H2O is 3212 and 374° F which is very likely to be deleterious to most sample tissues.


2. Hexamethyldisilazane Drying (HMDS)

Due to the heavy operator, equipment, and CO2 requirements of CPD, HMDS is a popular alternative to CPD drying of soft tissues. HMDS is a volatile liquid with a much lower surface tension than water (18.2 dynes/cm vs 72.8 dynes/cm). Through the above dehydration process followed by addition of HMDS (ethanol miscible), the tissue can air dry without deleterious side effects. Recent research has demonstrated that for many biological samples, HMDS and CPD render indistinguishable results. In this project fish eyes were dried with HMDS and mouse eyes with CPD. Both yielded high quality images without notable surface morphology disruption.


B. Coating

A scanning electron microscope images samples using an electron beam to interact with the sample on the surface at an excitation volume. In this volume, the incident electrons are scattered (SEI) and drawn back out to a detector to be quantified and rastered into the resultant image viewed by the user. Biological materials are a poor source of such Scattered Electrons (SE) and thus an additional material coating is desired on the sample to serve as a source of these imaging electrons. Additionally, the electrons that strike the surface of the sample must be quickly drawn off to prevent a negative charge from accumulating on the sample and repelling the negatively charged incident beam. The conductive gold coating serves this purpose as well. Inside the eye there are also myriad small particles which can be moved by the beam. Such particles are cemented in place with the gold coating. 

The conformal gold coating was applied to the samples using a Desk II sputter coating machine to deposit a 100 Angstrom layer. Thinner layers were initially tested, but rapid charging, despite all measures taken required the greater thickness, even though the coating was visible in high magnification images. The coating was used with an electron beam accelerating voltage of 3.00 KeV as greater voltages further exacerbated the anathematic charging problem.




To allow accurate imaging of the sample eye tissues, numerous steps were taken to prevent the charging of the biological tissues. All images were collected using a low 3.00 KeV accelerating voltage and a small column aperture (20.0 um). Additionally long working distances (WD) were used to minimize beam cross section on the sample and in turn reduce electron deposition and charging. In-column and in-chamber SE detectors were combined using the signal mixer to achieve better contrast. Gamma controls were also used to further utilize the dynamic range and produce high contrast images.




The purpose of the imaging system of the eye is to capture photons emanating from objects of interest and refract them to form an image on the photon sensors which is then integrated and relayed to the brain for high level interpretation. The principle refraction is done by the cornea and lens, while image capture is achieved by rods and cones, and neural integration and relaying by the ganglion cells and Cranial Nerve II.


A. Lens System 

Photons from an illumination source reflect off of an object and emanate outward into space. To form an image of the original object, the photons must be redirected, or refracted, to bring them to a second focus. In the biological eye, as in many optical systems, this is achieved by use of materials of differing refractive indices. In the eye these materials are the external medium (air or water) and the corneal material. In land mammals the difference between air (n=1.00) and the cornea (n=1.33) is large and most refraction takes place here. Following this the light travels through the eye lens (n=1.38) which has much less effect, but is changeable to modulate the focal length and achieve a focused image at the retina. 

Most animals require the ability to focus at both near and far distances. The eye in its relaxed state is focused at infinite, allowing distant objects to be easily viewed, but to focus on closer objects additional focal power is required. Additional refracting power is created through altering the shape of the eye lens by stretching and elongating the lens to alter its radius of curvature thus resulting in a change in power. The lens and cornea are thus of great importance to image formation and in turn their structure is of great interest to those researching vision. Here only the eye lens was analyzed as the cornea did no survive the preparation process.


1. Eye Lens Structure

The tissue of the eye lens is unique as it must be strong and flexible, but also completely transparent. To do so requires that the cytoplasm, nucleus, and all organelles be transparent to the visible spectrum otherwise they would interfere with images. Additionally there can be no blood vessels, capillaries, or nerves in the lens. As seen in Figure 1.1 below, the lens consists of densely packed fibers which are very tightly interconnected. The structure of the fibers is of a squashed hexagon with two wide parallel sides and four other smaller sides. The lens fibers interconnect into planar sheets at these smaller sides to form planar sheets as seen in Figure 1.1-3 which shows their excellent tessilating capability.  


Figure 1.1 - Lens Fiber cut away view

Note interconnecting ball and socket structures on short edges. Also note relative absence of ball and sockets on planar side of superficial fibers allowing planar movement.


Figure 1.2 - Lens Fiber Hook and Eye Structures - Nuclear Lens Fibers

Note alternating rows of hooks and complementary eyes. Spines hook into next row of eyes to interlock layers like Velcro.


Figure 1.3 - Lens Fiber Ball and Socket closeup

Note ball and socket joint interlocking at superficial cortical fiber edges. Planar surface of fibers interlock with next layer of offset fibers (removed).


The eye lens is very similar to an onion with many laminated concentric layers. In the case of the eye lens, the outer layer is a 4 micron thick [3] outer epithelium which encases the lens (Figure 1.4). Deep to this layer is the cortex of the lens which transitions quickly to the superficial cortical lens fibers. These lens fibers interconnect at their edges by “ball and socket” joints (Figure 1.3) and are the most articulate of layers as they must move the most during focal accommodation. The lens fibers stretch from the top to the bottom of the eye where they converge together as seen in Figure 1.5. Pores in the fibers allow the nutrient laden fluid known as vitreous humor to bath and support the fibers in the absence of a arteriole and venous network.


Figure 1.4 - Lens Epithelium Layer and Cortical Layer

Note lens epithelium at image center and cortex structure directly below.


Figure 1.5 - Len Fiber Convergence at Upper Pole

Note sagital view of anterior fibers on left and transverse cut of nasal fibers on right. Fibers are converging at the top of lens known as the pole.


Deep to the superficial are the inner are the inner nuclear fibers which constitute the nucleus. The nucleus has both adult and fetal portions from the periods during life when they were created [1]. All lens fibers are conserved over the organism’s lifetime. The inner fibers are packed more densely due to their reduced movement during accommodation and are thus more tightly interconnected with rows of spines and accompanying eyes (holes) which they interlock with (Figure 1.2). Figure 1.6 showcases the spines and their hollow nature constructed helically of microfilaments. A crack in the lens during cleaving in Figure 1.7 shows the dramatic landscape of the inner lens nucleus and the gap junction plates which join the fibers on their planar sides.


Figure 1.6 - Nuclear Lens Fiber Spines

Note fractured and broken spines along medial line of fiber. Spine in second hole on left demonstrates helical fibrous construction while fractured spines show hollow interiors.


Figure 1.7 - Lens Fiberscape Cavern

Note Cantilevered pair of nuclear lens fibers extending across a crack created in lens nucleus during sample cleaving for preparation.

* Image digitally altered to remove charging artifacts


In the case of the mouse lens seen above, the lens is spherical unlike the biconvex human lens as a mouse lacks the faculty of accommodation and thus relies heavily on whisker and olfactory sensory input for extrasensory perception. 

If lens fibers become damaged, the animal will lose the ability to focus and their eye will become a fixed focal length camera. When an eye develops cataracts, the lens fibers have absorbed fluid and have swollen, altering the refractive index of the fibers and making them, along with the lens opaque and inhibiting vision.  


B. Photon Detection

Equally important as the imaging optics of the eye are the dedicated structures which sense the imaged photons. The types of sensor cells are rods and cones. Rods are used to sense in low light conditions while cones are used to sense in bright light conditions. Additionally, due to the large photon flux they experience, cones are capable of discriminating between different wavelengths of visible light through three typical types of cone named for their absorptive wavelength peaks: Long (L), Medium (M), and Short (S). In most animals however, only rods exist as seen in the micrograph of the mouse retain in Figure 1.8-9. Both rods and cones achieve their greatest density in the eye at the fovea, or center of visual focus.  


Figure 1.8 - Retinal Layers at Fovea

Note Ganglion cells (green), rod inner and outer segments (blue), and Retinal Pigmented Epithelium (orange). Ganglion cells are especially densely packed as well as rods due to foveal position. (Ganglion debries from sample cleaving)


Figure 1.9 - Foveal Rods and Reticulocyte

Note Ganglion cell bed (green), rod inner and outer segments (blue), and escaped choroidal reticulocytes (red). Partial reticulocyte damaged during sample preparation.


In Figure 1.8 the three principle structures of the retina are visible. Here the upper portion consists of cuboidal ganglion cells, the middle section rods, and the lower section retinal pigmented epithelium (RPE). The ganglion cells connect the output of the rods to the brain as part of the neural interconnect while the RPE supported the rods and serves as the regeneration location for their pigment, rhodopsin. Counter to intuition, light enters the eye from the top of the figure and travels through the neural interconnect and ganglion cells, through the upper half of the rods (inner segment), and into the photon sensing pigments of the bottom of the rod (outer segment). To allow this to occur the neural interconnect and inner segments are also transparent. The only portions of the eye which are not transparent are the RPE, vascular choroid, and sclera layers which serve to absorb extraneous light from within the eye. 

When a photon of light enters the rod, it travels through hundreds of stacked layers of rhodopsin molecules of which there are millions per rod. Eventually the photon strikes a rhodopsin molecule where the absorbed energy converts it from its 11-cis to all-trans form, causing an electrical pulse of Ca2+ ions to travel out the attached ganglion cell [4]. The spent rhodopsin is then transported into the RPE where it is converted back to the cis-11 form and returned to the rod.


C. Neural Interconnect

The photons imaged by the eye are converted to electrical pulses and sent to the brain. Large amounts of neural interconnects within the eye preprocess this information before passing it out through the second cranial never (CN II) known as the Optic Nerve. The bed of ganglion cells seen in Figure 1.8 is just the beginning of the myriad interconnecting neural cells. This bed is extremely densely packed due to the myriad interconnects here including bipolar, amacrine,  and ganglion cells.

An example of the neural interconnect found in a starling is seen in false color in Figure 1.10. This image was taken of the front of the retina and numerous nerve bundles in yellow can be seen. The red represents the extracellular matrix which serves to anchor the cells and nerves in place while the blue shows bulbous ganglion cells integrating nerve signals.


Figure 1.10 - Neural Interconnect and Cellular Matrixf

Note nerves and nerve bundles (yellow), extracellular supporting matrix (red), and ganglion cells (blue).


The interconnecting cells form numerous structures which allow the sensing of motion, periodicity, and light accommodation. Some ganglions poll multiple rods and only activate if more than one rod activates, thus acting as a noise filter. For sensing motion ganglions similar to the noise filters are bundled to construct low resolution images of the visual field which exacerbate motion. Still other concentric bundles compare between the inner and outer concentric bundles and are sensitive to light on one segment while there is dark on the other segment. These ganglions allow certain frequencies of patterns to be sensed. Many of these types of “on” or “off” bundles with differing frequency sensitivities are used to convert the visual field to Fourier space for processing at the V1 cortex of the brain’s visual cortex.  

In the brain’s visual cortex multiple maps of the visual field are created and stored. Comparison between maps over time allows movement and scene changes to be sensed as well. In this area the visual field is magnified in a process known as cortical magnification so that the relative size of the information from the fovea is 30x larger than the periphery due to the high density of receptors and in turn the large amount of information being created at the fovea compared to the periphery. In the case of the mouse under study, the neural interconnect was diminutive due to the animal’s high dependence on other sensory faculties, but in humans it consists of a the massive optic nerve with over 1.5 million individual nerves.  


V. Conclusion 

From the above discussion it is evident that the SEM allows for incredible imaging of minute and intricate biological structures given proper preparation and execution. In this project 15 hours were spent painstakingly in preparation, though in retrospect more should have been spent and methods of freeze fracturing used as well as greater charging abatement. The eye theory discussed above and literature reviewed agreed well with the histological features observed in situ for the experimental specimens when they could be observed. Thus through the use of scanning electron imaging many details which cannot otherwise be observed in situ due to transparency, minute dimensions, and diffraction can be both qualitatively and quantitatively imaged and the myriad intricacy and beauty of organisms observed.


VI. Acknowledgements


VII. References  

  1. Hans Biomendal. Molecular and Cellular Biology of the Eye Lens. John Wiley & Sons; New York, 1981.

  2. Rae, James L. and Mathias, Richard T. “The Physiology of the LensThe Ocular Lens. Ed. Harry Maisel. Marcel Dekker: New York, 1985. p 93-118.

  3. Rafferty, Nancy S. “Lens MorphologyThe Ocular Lens. Ed. Harry Maisel. Marcel Dekker: New York, 1985. p 1-53.

  4. Rodieck, R. W. The vertebrate retina: principles of structure and function. W. H. Freeman and Company: San Francisco, 1973.