A basic introduction to adaptive optics for microscopy

Introduction

This document provides an introduction to adaptive optics for microscopy from a practical perspective. It is aimed at those who already have some experience in optics and preferably have experience in designing microscopes. It should prove useful to anyone who has an interest in using adaptive optics for their applications, but does not yet have experience with this technology.

We provide an overview of the concepts of adaptive optics and practical matters concerning their implementation in typical microscopes. This will include general concepts and principles. We will make the reader aware of important questions that should be considered when planning and implementing adaptive optics systems. Full details of specific implementations will not be provided here, but rather in separate documents that will be made available in due course.

Much of the content will also be relevant to other microscope-related applications, including those systems that use similar microscope type optics. This encompasses optical tweezers, laser-based micro/nano-fabrication and optical photo-stimulation systems, such as those used in optogenetic applications.

As we aim to provide generally useful information through this guide, we do not include a comprehensive review of developments in this area, nor do we seek to cover all possible implementations.

What is adaptive optics?

In adaptive optics (AO), one uses a dynamically reconfigurable optical element to correct aberrations. Aberrations are distortions in the phase of the light waves that may be present in microscopes either due to imperfections in the optics, or due to the optical structure of specimens. In addition to the adaptive element, an AO system also requires a method of aberration measurement, which may use a wavefront sensor or an indirect method, where the aberration correction is optimised through a sequence of measurements. Both of these measurement methods have been deployed in microscopes.

The AO concept was first introduced for military applications and then for astronomical telescopes. The need for AO in these applications was to overcome the problematic effects of atmospheric turbulence, which introduced time varying aberrations. While the fundamental concept of AO should be transferable from telescopes to microscopes, there are considerable differences in the optical systems involved. Indeed, it has not generally been possible to transfer technologies directly from astronomy, so other innovations were required.

We note that other terms, such as “active optics” or “dynamic optics” are sometimes encountered. While the phrases are to some degree interchangeable with “adaptive optics”, these other terms are often applied to situations that use reconfigurable elements, but not necessarily for aberration correction.

What is a microscope?

This may seem like a question with an obvious answer. However, we pose it here to illustrate an important point with regard to the implementation of AO. The difficulty in answering the question is that there is a multitude of different types of optical microscope, which employ different optical systems and rely upon different physical phenomena. The type of microscope will influence the choice of AO solution and the optimal solution may be different in each case. We elaborate on these choices further in this article.

Aberrations in microscopes

Aberrations, defined in the broadest sense, are the deviation of light from its ideal form. While there could be changes in amplitude or polarisation in a beam, which could be considered aberrations, we are usually concerned in AO with aberrations that are a deviation in the phase of a wavefront from the ideal. A microscope may have collimated beam paths, where the rays of the beam are parallel and the wavefronts are planar. Alternatively, in the focussed beam paths, rays should converge to (or diverge from) a point and wavefronts form caps of a spherical surface. An ideal lens, when used appropriately, can be considered as a device that converts planar wavefronts to spherical wavefronts and vice versa.

Aberrations may arise from low-quality optical components, poorly designed systems, inappropriately used or poorly configured optics. Lenses are never perfect and can thus introduce aberrations, but are more likely to do this if misaligned. Mirrors and dichroic filters can never be perfectly flat, so can cause aberrations in wavefronts reflected from their surfaces.

Aberrations may also be due to specimens. A common source is when there is a mismatch in refractive index between the objective lens immersion medium (e.g. using an oil immersion lens to view a water-based specimen). The light is refracted at the interface, meaning that the wavefronts transmitted through the interface are distorted. Similar distorted wavefronts are encountered when using a coverglass of incorrect thickness (particularly with water immersion lenses). Further, more complicated aberrations arise through the optical inhomogeneity of the specimen. Specimens consist of regions of differing refractive index, which cause distortions when light propagates through them. The resulting aberrations become larger in amplitude and more complex in form, the deeper one focuses into a specimen. The exact nature of these specimen-induced aberrations varies from specimen to specimen and even between different parts of the same sample. This variation is the main reason why an adaptive method is required for aberration correction.

It is important to note that most aberrations encountered in microscopes are static on the time scale of many imaging experiments, so – unlike in astronomical telescopes – continuous updating of the correction may not be necessary. Specimen-induced aberrations arise mainly from the overall shape of the specimen. Although there may be dynamic processes within cells and tissues, if they do not affect the overall structure then the aberrations at a particular point in the specimen will not change significantly. This means that after the aberration correction has been obtained at that point, it can be used without further online updating of the measurement.

The most important question to answer before you start

The most pressing question to answer, before you embark upon using AO in your system, is the following:

“Do I really need adaptive optics for my microscope?”

One assumes, if you have got to this stage, then you believe already that you have a problem with aberrations in your microscope.

It is safe to say that AO should not normally be needed in a well-designed and configured microscope that is used to image correctly prepared thin specimens. AO is most likely to be needed when imaging through thicker specimens or in unusual application where the ideal optical components are not available. Objective lenses with correction collars can be used to (partially) correct spherical aberration arising from a refractive index mismatch and they should be used wherever possible.

Note that certain types of microscopes, or microscopes used for particular applications, may be more sensitive to aberrations than others. For example, the imaging quality of super-resolution microscopes tends to suffer more from small aberrations than conventional resolution microscopes.

There is more than one type of microscope

A diverse range of high-resolution optical microscopes have been introduced for use particularly in biomedical sciences. They employ different contrast mechanisms (e.g. reflection, phase contrast, fluorescence, multi-photon fluorescence, and other non-linear optical processes) and different optical systems (e.g. widefield, laser scanning, and parallelised scanning systems). There are many more exotic variations that build upon these ideas.

AO systems have been implemented at least once in many of these microscopes, but there is no uniformly applicable implementation yet developed that can be deployed across all of the different microscopes. While effort is being made to create more universal solutions, it is important to understand why a solution packaged and optimised for one type of microscope may not be the appropriate solution for another type. A large part of this concerns the physical configuration of the microscope and the nature of the image formation process, which may include both optical and computational aspects.

One should also consider in which paths of the microscope aberration correction is required. For example, in a conventional widefield fluorescence microscope, aberrations are only problematic in the imaging path, as the illumination path is used only to provide uniform illumination of the specimen. In this case, aberration correction is only necessary in the emission/imaging path. In other microscopes, correction is required in both illumination and imaging paths; this includes the confocal microscope or structured illumination microscopes, where imaging fidelity is required to generate the illumination patterns and to create an image at the detector. Microscopes that use non-linear phenomena, such as the two-photon excitation fluorescence microscope, derive their resolution from the illumination path; aberrations in the imaging path have no effect in the most common configuration using a large area detector. In these microscopes, aberration correction is only required in the illumination path.

Choice of adaptive optical element

The two main options available for AO microscopy are deformable mirrors (DMs) and liquid crystal spatial light modulators (SLMs). The fundamental purpose is the same for either element: to add an equal but opposite (conjugate) aberration into the system in order to compensate for the unwanted problematic aberrations.

Deformable mirrors have reflective surfaces whose shape can be changed by the application of forces via an arrangement of actuators. Actuation can be implemented using electrostatic, electromagnetic or piezoelectric forces. The majority of DMs used in microscopy have continuous surfaces, although segmented mirrors are also available. Being mirrors, they have high reflectivity, typically above 95% or higher if appropriate coatings are used. They exhibit broadband operation and are also polarisation independent, making them ideal for use with fluorescence. Typically, DMs have tens to hundreds of actuators and with the right electronic drivers can operate at rates of up to several kHz. Some DMs have large stroke (or surface deflection) of several tens of micrometres; however, the aberrations encountered realistically in microscopes are usually much smaller than this. It should be noted that having more actuators or more stroke may provide striking contrasts between models on manufacturers’ data sheets, but it does not necessary mean that a particular DM is better for a particular application. Such determinations can only be made by a more thorough comparison of expected aberrations and device capability.

Spatial light modulators come in numerous forms, but the most commonly used type in adaptive microscope is the liquid crystal on silicon (LCOS) reflective mode, parallel aligned, nematic liquid crystal SLM. These tend to have lower optical efficiency than DMs and they can only be used with polarised light, so are well suited to correction of laser illumination, but not to fluorescence. For practical reasons, they also are frequently used as diffractive (sometimes termed “holographic”) elements, which makes them wavelength dependent. They are usually limited to a few tens of Hz update rates, although some have been demonstrated at hundreds of Hz. The number of driveable pixels is typically much higher than the number of actuators of a DM (e.g. pixel numbers may be in the hundreds of thousands). The modulation range is however limited to very small multiples of a wavelength. This limitation in range can be overcome using phase wrapping – where large phases are “wrapped” into the range of a single wavelength – which can be used for correction of very large magnitude aberrations, albeit for narrowband light.

Transmissive adaptive elements have recently become available. These consist of transparent fluid filled chambers whose shape can be changed in order to adapt the thickness of the element. They have not yet been used extensively in microscopes, but seem to show performances comparable to low order DM models. Being transmissive devices, though, they do have the potential to make AO design easier, especially when adding devices to existing optical systems, as they avoid the need for folded paths required for reflection mode DM and SLM devices.

Various aspects should be considered before making a choice of adaptive element for your application:

• Is correction only required for laser light? This is narrowband and polarised and, as there is often much more laser power available than is needed, there are not significant limitations on efficiency. SLMs are suitable for this task.

• Is correction required for fluorescence light (or other weak emission)? This light is broadband and unpolarised and high efficiency of collection is required. DMs are best suited for these applications.

• Do you need to correct multiple ranges of wavelengths simultaneously, for example in different excitation or emission paths? A DM may be the optimal choice in this case if it can be used to correct all paths simultaneously.

• Is optical efficiency important or can you tolerate some losses?

• Do you need to correct large aberrations (of many wavelengths in magnitude) or are small corrections (order of a wavelength) sufficient?

• Will the aberrations be complex (e.g. heavy scattering through thick tissue) or simple (e.g. mostly spherical aberration)?

• Is high speed operation necessary?

• Do you have budget constraints? Unsurprisingly, prices depend on capability (e.g. number of actuators, correction range and speed) and it makes little sense to pay for capability you are unlikely ever to use.

Aberration measurement

Aberration measurement can be performed either using direct wavefront sensing or indirect measurement – or “sensorless AO”. While direct wavefront sensing – particularly in the form of the Shack Hartman sensor – is widely used in AO systems in other applications, it is less widely used in microscopy. The reason for this is that the Shack Hartmann sensor relies upon presence of a point-like source of light. In general, the specimen in a microscope is three-dimensional and light can emanate from throughout a volume of the specimen. Such light would overwhelm, or at least confuse, the sensor measurement. These sensors have been used effectively in situations where point like light sources are present. The most useful of these applications is where two-photon fluorescence is used as the source for wavefront sensing. As two-photon fluorescence is generated only in the focal spot, the emission behaves like a point source. In confocal fluorescence microscopy, wavefront sensors have been used with point like objects, such as beads or highly localised markers in biological specimens. They cannot however be used more generally in all specimens, because of the presence of out-of-focus light. These limitations are also present for widefield microscopes, where such direct sensing is only likely to be effective for isolated point sources.

Sensorless AO avoids the problems of direct sensing by using the images produced by the microscope. A sequence of images is obtained with different aberrations intentionally added using an adaptive element. These images collectively contain information that can be used to infer the necessary aberration correction. Careful choice of the applied aberrations and the estimation algorithm means that the correction can be inferred from a small number of images. This approach has the advantage that the only modification to an existing microscope design is the addition of the adaptive element and associated optics. The sensorless AO approach, as it is based upon the practical acquisition of regular (albeit aberrated) images, is conceptually compatible with any type of microscope. However, as the images must be acquired sequentially, the aberration measurement process does take longer than a direct sensing approach. This is not necessarily a problematic limitation, as most aberrations in microscope specimens are static over the timescale of many imaging experiments.

Conclusion

This document has provided a brief overview of the major considerations in the design and application of microscope AO systems. Further details on each of the aspects covered and on practical tips for their implementation will be discussed in further articles.