A nice thing about Columbia's graduate courses is that almost all of them require semester-long projects. For PhD students, these projects offer an opportunity to kickstart larger research projects in the event that they go well.
For my Optical Systems course, we've been asked to design and simulate a relatively complex optical system. We primarily use SDOCT here as our imaging modality for imaging the cochlea, as it provides high-resolution depth-resolved images of the basilar membrane en vivo. It also allows for spectral domain phase microscopy (SDPM) inherently, giving vibration data for very small-scale vibrations.
Recently, however, we have been running into a limit of OCT - as it works based on the reflectivity of materials, translucent materials are very hard to image. This is true for many imaging modalities, but can often be overcome by staining translucent materials of interest - If one uses a fluorescent dye which targets the translucent material of interest, its fluorescence can make it visible to the user in images. This process is called fluorescent labeling.
Fluorescent labeling does not effect the output of an SDOCT system, as it does not change the reflectivity of the material in the bandwidth of interest. To see this, consider how an SDOCT system can find the distance between itself and a point reflector: the light source is split into a reference arm and sample arm, and the interference pattern generated by the reflected sample light and the reference light is used to determine the distance. This reflected light is entirely within the bandwidth of the light source, and the interference pattern is determined using photodetectors within the OCT system specified to the light source's bandwidth.
Now consider a fluorescing object rather than a point reflector. Light enters the object, is absorbed, and the object fluoresces. This fluoresced light is a narrowband signal, and has bandwidth which does not likely overlap with the bandwidth of the light source. Thus, the fluoresced light is not even picked up by the on-board photodetectors, and the interferogram will be unaffected by its existence. Thus, SDOCT cannot be used for fluorescent labeling, and cannot be used to image translucent materials without modification.
The tectorial membrane, which winds about the cochlear spiral above the organ of Corti, is a gel-like membrane which is mostly fluid. In the inner ear, the basilar membrane will vibrate due to an incoming sound. Where this vibration is strong, this pushes the stereocilia (the small 'hairs' on the hair cells) into the tectorial membrane, effectively 'tilting' the stereocilia due to a shearing force. This opens ion channels, creating the driving current for the sensory hair cells, and thus creating a neural signal which is transmitted to the brain along the auditory nerve. The tectorial membrane, at least in this simple mechanical way, is crucial to hearing.
![](https://static.wixstatic.com/media/a27d24_99b057004f72415bb6b84b0154a2b7d2~mv2.jpg/v1/fill/w_300,h_233,al_c,q_80,enc_auto/a27d24_99b057004f72415bb6b84b0154a2b7d2~mv2.jpg)
However, it may be reductive to think of the tectorial membrane entirely as "that object which the stereocilia push into." It must surely at least play this role, but its role in the en vivo mechanics of the cochlea are not yet well-understood enough to say that this is its only role.
We know ex vivo that the tectorial membrane can actually carry traveling waves much like the basilar membrane, which could suggest a coupled wave system at the heart of cochlear mechanics. But we also know well that we cannot rely on ex vivo experiments for the characterization of cochlear mechanics. Hearing is active; we know that very important elements of hearing such as the cochlear amplifier require external power to operate properly. Thus, to better understand the role the tectorial membrane plays in cochlear mechanics, we need an imaging modality that can detect the tectorial membrane en vivo.
It is easiest for the SDOCT system to remain our imaging modality of choice for vibrometry. Thus, our goal is to combine SDOCT with a second imaging modality which can provide an image of the tectorial membrane. With SDOCT, we image through the round window membrane. This means that we view a basal portion of the cochlea, seeing on a B-Scan (from top to bottom) the round window membrane, the basilar membrane and the organ of Corti.
We would like to see beneath the organ of Corti, where the tectorial membrane lies. Thus, if we wish to use a second imaging modality, it must be able to obtain images through the round window and basilar membranes, as well as through the organ of Corti. As the membranes are fairly low in reflectivity, and the desired depth here is on the order of 10s of microns, this is not the hardest constraint to work with. This modality need not be particularly fast (we will not be using it for vibrometry), but must be able to support fluorescent labeling.
It is important to note that the tectorial membrane can, in fact, be stained so as to fluoresce. It has been shown that the tectorial membrane contains a higher concentration of calcium ions than the endolymph fluid which surrounds it, meaning that any fluorescent dye which targets Ca2+ (and many exist) will stain the tectorial membrane so as to make it fluoresce
Two-photon microscopes and confocal microscopes each fit the bill for our fluorescent imaging constraints. They each have their pros and cons, however, which deserve explanation.
In two-photon microscopy, two narrowband beams are shot at a sample so as to intersect at a "point" (as close to a point as one can obtain). The wavelength of these light-sources is chosen such that a single photon will not have enough energy to excite fluorescence, but the combination of of one from each beam will. Thus, point-by-point fluorescence images can be created if both beams are scanned together.
![](https://static.wixstatic.com/media/a27d24_931fd9fca8f8456c8b4259b5ef3ec296~mv2.png/v1/fill/w_510,h_195,al_c,q_85,enc_auto/a27d24_931fd9fca8f8456c8b4259b5ef3ec296~mv2.png)
This modality requires two light sources and a relatively complex scanning mechanism to operate. Furthermore, this system naturally requires many photon to be sent from each light source, or else the rare event of two-photon fluorescence will never occur. This means that the system requires two high power light sources, making photobleaching a real possibility. One must be very careful with the power of the sources so as to generate an image without damaging the tissue of interest.
![](https://static.wixstatic.com/media/a27d24_5044d96d0f22487c846d0ae9476c0d5d~mv2.png/v1/fill/w_980,h_529,al_c,q_90,usm_0.66_1.00_0.01,enc_auto/a27d24_5044d96d0f22487c846d0ae9476c0d5d~mv2.png)
Confocal microscopy works based simply on geometric optics. Rather than developing a field of view and a focal plane, the confocal microscope is designed so that only a point is in focus at any given time. This is achieved through a "point" light source and "point" detector, as demonstrated in the figure below. This allows for optical sectioning up to the abberations in the lenses use to craft the device, and also requires point-by-point scanning to create an image. It is less precice, in most cases, than two-photon microscopy, but requires less from the scanning mechanism and light source.
![](https://static.wixstatic.com/media/a27d24_504625dc7b5741e3ab129e37feb0e4ca~mv2.jpg/v1/fill/w_409,h_350,al_c,q_80,enc_auto/a27d24_504625dc7b5741e3ab129e37feb0e4ca~mv2.jpg)
It is the confocal modality that appeals more to our team for a few reasons. Perhaps most importantly, it is significantly cheaper than the two-photon alternative, and provides "enough" for our experiments. It also does not run the risk of photobleaching as easily as two-photon.
Most interesting, however, is the scanning mechanism for confocal. By scanning the mirror in one rotational direction, a "line" image can be created along a plane. By scanning the mirror in two directions, a planar image (by most called "a picture") can be generated of a plane with strong depth sectioning. These mirror scans are built into most SDOCT systems, as they allow for the creation of B-scans and volume scans rather than simple axial line scans (or "A-Scans"). Thus, much of the scanning desired for a confocal system is already built-in to our lab's SDOCT system, allowing us to use the same mirror to image with both modalities.
Amiss in this formulation is the ability for the confocal microscope to scan into the material axially. If we want to create images with our confocal microscope which align with OCT B-Scans or the above two-photon images of the cochlea, we must move our focal point axially in to or out of the material. This can be done in one of two ways - either by moving the scanning mirror axially, or by moving the sample axially. Without changing our optical system at all, the latter method can be achieved simply with an xyz-stage, and may be the best choice for our particular interests.
So we are embarking on the design of a combined confocal microscopy and SDOCT system! We have a long way to go, but it seems promising. Thanks for reading, and have a lovely day.
Commentaires