Medical Device Link.

DESIGN

Fluorescence and fluorophores

Optical sensing plays a widespread role in diagnostics, frequently providing the vital link in the chain that allows the concentration of a biochemical analyte to be quantified. Light interacts with matter in many ways, but one class of interaction, namely fluorescence, stands out above all others when it comes to biochemical sensing.

Fluorescence occurs when matter first absorbs light at one wavelength and then re-emits light at different wavelengths, shifted relative to the input. The shift, almost invariably to longer wavelengths, occurs because of internal transitions that consume some of the energy of the incident photon and reduce the energy of the emitted photon.

“Optical whiteners” are probably the most common example of fluorescence. These are dyes that have been specially engineered to fluoresce in ultraviolet light and to enhance the appearance of clothes or paper. Optical whiteners are just one example of a fluorophore: a molecule that has been found or developed to exhibit useful fluorescence. In the latter case, it is now possible to go beyond conventional synthesis and to engineer “quantum dot” molecules with specifically tailored energy levels and associated excitation and emission bands.

Figure 1: A system perspective of a generic fluorimeter. (click image to enlarge)

In addition to their spectral response, fluorophores also exhibit a characteristic decay period or lifetime. Typically, lifetimes are in the nanosecond range, but some molecules decay more slowly. Allowing for the transition from fluorescence to what, more strictly, is known as phosphorescence, lifetimes can extend to milliseconds and beyond.

Much of the importance of fluorescence as a measurement method stems from the fact that the wavelength shift facilitates a sensitive “dark background” technique. This is achieved by means of spectral filters that transmit the emission band, but reject the excitation band. The additional possibility of discrimination in the time domain then serves to compound the status of fluorescence as one of the most powerful weapons in the armoury of the bioanalytical instrumentation developer.

Figure 2: A high-performance LED-based fluorimeter. (click image to enlarge)

Many fluorophores are available on the market for application as biochemical “tags” to enable the detection of a specific target or event. Some are designed to be used in isolation, but often they can be paired with other molecules that, according to their proximity, will activate or deactivate the fluorescence. This offers the assay designer an even wider range of possibilities and is essential to many applications in molecular diagnostics. It is no surprise that high performance fluorophores are valuable intellectual property and that license fees can exert a major influence on the development process.

Measurement of fluorescence

The diagram in Figure 1 shows a system-level perspective of a generic fluorimeter. The left-hand side is concerned with the excitation of the sample to stimulate fluorescence and the right-hand side is concerned with its collection, filtering, detection and post-processing.

There are, of course, many system-level and detailed design decisions to make when developing a fluorimeter for any specific application. For example, possible excitation sources include xenon flashlamps, diode and solid state lasers and, of increasing significance, light emitting diodes (LEDs). Similarly, detectors can be single devices, coarse arrays or high resolution image sensors and can respond directly or provide the front-end gain of a photomultiplier tube or a related device for ultrasensitive detection.

Figure 3: An eight-channel LED multiplexer for multiband excitation. (click image to enlarge)

Figure 2 is an example of a high performance LED-based fluorimeter showing the custom electronics and a schematic of the complete fluorimeter. The electronics use synchronous detection to achieve performance approaching the fundamental noise floor and so, despite being a simple and relatively low cost solution, it is capable of detecting less than a femtomole of the target fluorophore within the sample volume.

Paradoxically, LED-based solutions have found particular utility, not only in low-end, cost-sensitive systems, but also in state-of-the-art and ultrasensitive, molecular diagnostic systems. In these systems, biochemical amplification ensures that each analyte molecule is able to “switch-on” many fluorophores so that the sensitivity demanded of the fluorimeter is correspondingly reduced. In this context, LEDs are not only a suitable choice, but because of their robustness and reliability, also an attractive one.

Another advantage of LEDs is that they are now available in a wide range of wavelengths and with increasing optical powers. Figure 3 shows an LED multiplexer, designed to permit parallel detection of multiple analytes. Multiplexing is achieved by tagging each analyte with a different fluorophore and matching an LED to each excitation band.

In most cases, it is also desirable to demultiplex the different emission bands and Figure 4 shows a recently developed seven-channel demulti-plexer, part of a high throughput, microfluidic assay platform. The demultiplexer uses a cascade of dichroic filters to split-off each channel in turn to its own photomultipler tube.

When every photon counts

Figure 4: A seven-channel demultiplexer for multiband emission. (click image to enlarge)

Figure 5 provides a diagrammatical context for the discussion that follows. It summarises the diagnostic landscape in terms of the volume of sample available to be measured and the typical concentration of the target analyte. The diagonal lines represent the implied number of analyte molecules for combinations of volume and concentration and the regions, as overlaid, represent different applications.

Driven particularly by the increasing diagnostic significance of biomarkers, present in only trace quantities, and also by the desire to process the smallest possible sample volumes, there is a steady progression both down and to the left, towards ever smaller numbers of molecules. In many cases, biochemical amplification is not possible and the challenge is to detect small numbers of fluorophores directly.

Figure 5: A summary of the diagnostic landscape in terms of sample volume and analyte concentration. (click image to enlarge)

The seven-channel fluorimeter shown in Figure 4, developed for the microfluidic platform, is able to detect as few as 103 molecules or 1 zepto-mole of fluorophore in each of the samples that it interrogates. This requires excitation by a laser source for high power density, detection by a photomultiplier to elevate the signal above the electronic noise floor, and also the inclusion of effective measures to combat background fluorescence. The latter are necessary, because fluctuations in the background signal are generally responsible for the noise floor that determines the limiting sensitivity of these systems. Selection of materials with low autofluorescence is a vital part of the equation, but it is also necessary to pay close attention to the optical design. The seven-channel fluorimeter uses a confocal design to create a small measurement volume that is tightly centred on the sample and relatively immune to its surroundings.

Remarkably, it is now possible and practical to detect even a single molecule under the right conditions. A single fluorophore, excited with sufficient power density, may emit thousands of photons before it is exhausted by photo-bleaching. A well designed, highly efficient optical system can harvest enough of these emitted photons to achieve reliable detection when equipped with a suitable photon counting or integrating detector. This is the basis of some micro array systems that offer single copy detection without amplification.

Exploiting the time domain

Figure 6: Probability density functions for the intensity and lifetime of an unknown fluorophore obtained by the application of a Bayesian algorithm to photon arrival times and showing the narrowing of the distribution as more photons are processed. (click image to enlarge)

Fluorescence is a powerful technique when used conventionally. When the time domain is included, even more possibilities are available. One possibility is to recognise different flurophores not only by their spectra, but also by their characteristic lifetimes so that the number of parallel measurements can be increased even further. In ultrasensitive systems, these lifetimes are derived from photon counting detectors that also record arrival times. In this context powerful, probabilistic signal processing techniques have been applied to extract the maximum information from the arrival time of each successive photon. Figure 6 illustrates how this algorithm is able to progressively refine its inference of the unknown intensity and lifetime of a fluorophore in the sample.

Figure 7: A lateral flow strip labelled with a TRF label and the TRF detection principle. (click image to enlarge)

Taking advantage of the time domain is by no means confined to high-end systems. Use of fluorophores with extended lifetimes opens up a technique known as time resolved fluorescence (TRF). In this case, the lifetime is long enough to make it possible to measure the fluorescence during a brief period after the excitation is switched-off. As a result, it becomes possible to use electronic timing rather than optical filters to exclude the excitation light and thereby to extend performance without adding cost. This is relevant to the most cost-sensitive of applications. Figure 7 shows a lateral flow strip for an over the counter (OTC) diagnostic device, labelled with a suitable TRF label instead of a conventional absorber.

A bright future

There are strong drivers within the diagnostics market to detect analytes at lower concentrations, in smaller volumes and, wherever possible, at lower cost. Fluorescence, as a technique, has contributed a great deal to this progression, but still has much more to offer.

In the near future, progress is expected on two interrelated fronts. First, fuelled by the development of high-end systems for which performance takes precedence over cost, detection systems will be developed that are able to obtain more information from fewer fluorophores in less time. There is, for example, a particularly strong current interest in extending the capability of direct, single-molecule detection techniques for molecular diagnostics and rapid sequencing applications. Second, drawing on a combination of ingenuity and access to the next wave of enabling components, it will be possible to transfer many of the attributes of these high-end systems into point of care and even into OTC devices.

Mike Hazell is a Senior Consultant at Cambridge Consultants Ltd. He specialises in optical physics and optical engineering and has a particular interest in detection systems for biochemical analysis. Science Park, Milton Road, Cambridge CB4 0DW, UK, tel. +44 1223 420 024, e-mail: info@cambridgeconsultants.com, www.cambridgeconsultants.com