EP4634641A1

EP4634641A1 - Microscopy of biological samples

Info

Publication number: EP4634641A1
Application number: EP23828234.7A
Authority: EP
Inventors: Richard Michael BERRY; Robert Rafgatovich ISHMUKHAMETOV
Original assignee: Oxford University Innovation Ltd
Current assignee: Oxford University Innovation Ltd
Priority date: 2022-12-16
Filing date: 2023-12-13
Publication date: 2025-10-22
Also published as: WO2024127003A1; GB202219021D0

Abstract

A method is provided for detection of birefringent particles in a blood sample from a subject. The blood sample is processed to lyse red blood cells and the processed blood sample is imaged using an optical microscope configured for back scattering dark field microscopy. The imaging comprises illuminating the processed blood sample with light having a first polarisation state, capturing an image of light scattered by the processed blood sample having a second polarisation state different from the first polarisation state, and analysing the image to identify birefringent particles. Either a magnification of the optical microscope is at most 10x magnification; or a thickness of the processed blood sample is greater than a depth of field of the optical microscope and the imaging further comprises scanning the focal plane of the optical microscope through the thickness of the processed blood sample during the capturing of the image.

Description

MICROSCOPY OF BIOLOGICAL SAMPLES

The invention relates to methods for performing microscopy of biological samples, in particular for detection of birefringent particles therein that may be indicative of a disease state.

Certain diseases are characterised by the formation of birefringent particles, such as crystals, in the body. These particles can be detected by optical microscopy in blood samples and thereby used to diagnose the disease. A particular example of this is the diagnosis of malaria.

Malaria is a serious infectious disease caused by parasitic single-celled organisms carried by mosquitos. Those infected suffer symptoms such as fever, tiredness, and vomiting. Severe cases can cause seizures, coma, or even death. The disease is widespread, particularly in countries around the equator and is estimated to cause over half a million deaths a year. Treatment with antimalarial medications can significantly improve survival rates, but this requires prompt and reliable diagnosis to identify those infected at any early stage of the disease.

Malarial parasites develop a unique intracellular biomarker: a crystalline, birefringent pigment called hemozoin (HZ). HZ forms inside every cell of the malarial parasite after it invades the red blood cells of the host. HZ is unique to malaria only and so presence of HZ in the blood of a subject is a reliable marker to diagnose malaria.

The ‘gold standard’ technique for detection of HZ in blood uses Giemsa staining of a thin and a thick smear of a subject’s blood. The stained smear is examined under an optical microscope, where individual parasite cells are visible due to the staining. While effective and reliable, this diagnosis method is labour intensive and slow because it requires careful sample preparation and skilled operators to prepare and examine the sample. This makes it unsuitable for rapid diagnosis in many of the regions of the world most affected by malaria, where medical infrastructure is often lacking.

Rapid Diagnostic Tests (RDTs) are available for malaria as an alternative to microscopic examination of blood smears. RDTs are based on immuno-chromatographic detection of certain target malarial protein biomarkers. RDTs are simple to use, cheap ($0.2-0.5/test), relatively fast (15 min/test), and need just a single droplet of blood. RDTs are therefore widely used for self-diagnostics. However, RDTs are not quantitative (so cannot quantify a severity of infection), have relatively low sensitivity, and have limited shelf life. RDTs are also known to stop working in certain areas where malarial parasites evolve to lose the protein biomarkers on which the RDTs rely.

There is therefore still a need for fast, quantitative, and sensitive diagnosis of malaria. In particular, there is a need for such diagnosis that does not require expensive equipment or trained operators, so that self-diagnosis can be performed by anyone at risk of getting malaria, including residents of malaria-endemic countries and international travellers, as well as by doctors or other medical practitioners.

According to a first aspect, there is provided a method for detection of birefringent particles in a blood sample from a subject comprising: processing the blood sample to lyse red blood cells in the blood sample; and imaging the processed blood sample using an optical microscope configured for back scattering dark field microscopy, wherein: a magnification of the optical microscope is at most lOx magnification; the imaging comprises illuminating the processed blood sample with light having a first polarisation state, capturing an image of light scattered by the processed blood sample having a second polarisation state different from the first polarisation state, and analysing the image to identify birefringent particles in the processed blood sample.

This method allows for detection of birefringent particles in the blood of a subject using a simple low-cost device. HZ is typically the only birefringent object in blood, and therefore detection of birefringent particles allows for quantitative and sensitive diagnosis. In addition, HZ is a conservative biomarker that cannot be lost by the malarial parasite, unlike the markers used by malarial RDTs. Surprisingly, the inventors have found that by lysing red blood cells and using back-scattering dark field microscopy with polarised light, the birefringent particles can still be reliably detected even when imaging a large volume of the blood sample in a single operation. In particular, the sample can be imaged at low magnifications that would be insufficient to resolve the particles under normal optical microscopy. By increasing the volume of sample that is examined for the presence of birefringent particles in a single field of view relative to other common techniques, the time taken to examine a sample can be reduced. Therefore compared to other comparable diagnostic techniques such as RDTs, the method offers quantitative, sensitive, and fast detection using inexpensive equipment that does not require specialist training. Although mainly described below in the context of malaria diagnosis, the method can also be used for other applications requiring detection of birefringent particles in blood or other physiological fluids. For example, the method can be used for detection of gout or other crystal diseases.

In some embodiments, the processing of the blood sample further comprises loading the processed blood sample into a sample cartridge and mounting the sample cartridge loaded with the processed blood sample onto the optical microscope. Use of a sample cartridge simplifies the preparation process for untrained users and can ensure greater consistency in the presentation of the sample for analysis.

In some embodiments, the sample cartridge is pre-loaded with a detergent, and the processing of the blood sample is carried out by mixing of the detergent with the blood sample on loading of the blood sample into the sample cartridge, wherein optionally the detergent is a dry detergent. This further simplifies the sample preparation process by removing the need for an additional step of adding an agent to lyse the red blood cells. This can also further improve consistency of sample preparation by untrained users by ensuring a standardised amount of detergent is added to the blood sample.

In some embodiments, a thickness of the processed blood sample through which light travels during the imaging is greater than a depth of field of the optical microscope; and the imaging of the processed blood sample further comprises scanning the focal plane of the optical microscope through the thickness of the processed blood sample during the capturing of the image. In some embodiments, the thickness of the processed blood sample through which light travels during the imaging is at least 50pm, optionally at least 100pm, optionally at least 200pm, optionally at least 500pm, optionally at least 1mm. This further increases the volume of blood examined for the presence of birefringent particles, thereby increasing the sensitivity of the method.

In some embodiments, the magnification of the optical microscope is at most 5x, optionally at most 2x, optionally at most lx, optionally at most 0.5x. Further reduction in the magnification also increases the volume of sample that can be examined in the same time, thereby improving sensitivity.

In some embodiments, the first and second polarisation states are linear polarisation states. This is a straightforward way to provide polarisation with commonly-available components, thereby reducing cost of the equipment required.

In some embodiments, the first and second polarisation states are orthogonal. This ensures that predominantly light that is scattered from birefringent particles, and therefore has its polarisation changed, is detected in the image, thereby improving contrast in the image.

In some embodiments, the illuminating of the processed blood sample with light having a first polarisation state comprises illuminating the processed blood sample with light transmitted through a first polariser configured to transmit light having the first polarisation state. In some embodiments, the capturing of the image of light scattered by the processed blood sample comprises capturing an image of light scattered from the processed blood sample and transmitted through a second polariser, the second polariser configured to transmit light having the second polarisation state. Using polarisers to polarise the illuminating light means that commonly-available and inexpensive light sources can be used. Polarisers are inexpensive and readily available, and so their use reduces the cost of equipment needed to carry out the method.

In some embodiments, the scanning of the focal plane of the optical microscope comprises moving the processed blood sample relative to an objective lens of the optical microscope along an optical axis of the optical microscope. This is the most straightforward way to scan the focal plane and can easily be achieved with standard microscopy equipment.

In some embodiments, the scanning of the focal plane of the optical microscope comprises scanning the focal plane using an actuator. Using an actuator can provide greater consistency in the rate and range of scanning, thereby improving reliability of the method when carried out by untrained operators.

In some embodiments, the capturing of the image of light scattered from the processed blood sample is performed using an image sensor having at most 8xl0⁶ pixels, optionally at most 2xl0⁶ pixels, optionally at most 5xl0⁵ pixels, optionally at most 2xl0⁵. Using image sensors having a moderate to low resolution reduces the cost of the equipment needed to carry out the method, thereby making it more accessible to malaria-endemic regions.

In some embodiments, the image is an image stack comprising a plurality of image frames captured at successive time points during the scanning of the focal plane. This allows different regions through the thickness of the sample to be independently analysed for greater accuracy and reliability.

In some embodiments, a rate of the scanning of the focal plane and a time interval between the successive time points are such that the focal plane moves at most 50pm, optionally at most 20pm, optionally at most 10pm between the successive time points. This ensures that all parts of the sample are taken into account during the imaging so that the method makes the most effective use of available information from the sample.

In some embodiments, the analysing of the image comprises forming a maximum projection image in which a value of each pixel is determined as the highest measured intensity for that pixel across the image stack and identifying birefringent particles in the maximum projection image. This ensures that each possible crystal location is analysed using the greatest light that it emitted during the scanning, thereby maximising the chance of correctly identifying birefringent particles.

In some embodiments, the identifying of birefringent particles in the blood sample comprises identifying contiguous groups of pixels in the image having an intensity above a predetermined intensity threshold. This reduces false positives by ensuring only birefringent particles of a minimum intensity are detected, and that particles are not double counted if they extend over multiple pixels.

In some embodiments, the contiguous groups of pixels are groups of pixels having a number of pixels below a predetermined size threshold. This ensures that very large areas of signal do not result in false positives, because such regions may be unlikely to correspond to the birefringent particles to be detected based on expected sizes of such particles.

In some embodiments, the analysing of the image comprises performing background subtraction on the image. This improves contrast and thereby the likelihood of correctly identifying birefringent particles.

In some embodiments, performing background subtraction comprises applying a highpass spatial filter to the image. This reduces contributions from very large areas of signal, which are likely to be due to contaminants such as dust in the optical setup.

In some embodiments, applying the highpass spatial filter comprises subtracting from the value of each pixel of the image a percentile value of the values of pixels surrounding the pixel of the image, the percentile value being optionally between a 60^th and a 90^th percentile value, optionally between a 70^th and an 80^th percentile value, optionally a 75^th percentile value. This is a convenient and computationally inexpensive implementation of a highpass filter, thereby reducing the computational requirement and cost of the analysis equipment.

In some embodiments, the method further comprises, prior to loading the processed blood sample into the sample cartridge, imaging the sample cartridge using the optical microscope to obtain a calibration image, and wherein performing background subtraction comprises using the calibration image. This helps to reduce the effect on the detection of birefringent particles of any imperfections or contaminants that may be present on or in the sample cartridge.

In some embodiments, the processing of the blood sample comprises diluting the blood sample with distilled water. This can lyse cells effectively through osmotic pressure without expensive reagents.

In some embodiments, the processing of the blood sample comprises mixing a detergent with the blood sample, the detergent optionally being saponin. This can more selectively lyse the desired cell types in the sample with reduced dilution of the sample.

In some embodiments, the processing of the blood sample is performed to also lyse at least one further blood component in addition to the red blood cells, optionally wherein the at least one further blood component comprises white blood cells and/or platelets. While red blood cells are the dominant cell type in blood, lysing other cell types as well can further reduce the background signal from particles other than birefringent particles and improve detection of the desired particle type.

In some embodiments, the birefringent particles comprise hemozoin crystals. This is appropriate when the method is used to help diagnose malaria.

In some embodiments, the method further comprises determining that the subject has malaria depending on the identified hemozoin crystals. This allows the method to be used for diagnosis by untrained users in circumstances where rapid, inexpensive diagnosis is required.

According to a second aspect, there is provided a method for detection of birefringent particles in a blood sample from a subject comprising: processing the blood sample to lyse red blood cells in the blood sample; imaging the processed blood sample using an optical microscope configured for back scattering dark field microscopy, wherein: a thickness of the processed blood sample through which light travels during the imaging is greater than a depth of field of the optical microscope; the imaging comprises illuminating the processed blood sample with light having a first polarisation state, capturing an image of light scattered by the processed blood sample having a second polarisation state different from the first polarisation state while scanning the focal plane of the optical microscope through the thickness of the processed blood sample, and analysing the image to identify birefringent particles in the blood sample.

This method allows for detection of birefringent particles in the blood of a subject using a simple low-cost device. HZ is typically the only birefringent object in blood, and therefore detection of birefringent particles allows for quantitative and sensitive diagnosis. In addition, HZ is a conservative biomarker that cannot be lost by the malarial parasite, unlike the markers used by malarial RDTs. Surprisingly, the inventors have found that by lysing red blood cells and using back-scattering dark field microscopy with polarised light, the birefringent particles can still be reliably detected even when imaging a large volume of the blood sample in a single operation. In particular, the focal plane can be scanned through a sample having a thickness greater than the depth of field of the optical microscope. By increasing the volume of sample that is examined for the presence of birefringent particles in a single field of view relative to other common techniques, the time taken to examine a sample can be reduced. Therefore compared to other comparable diagnostic techniques such as RDTs, the method offers quantitative, sensitive, and fast detection using inexpensive equipment that does not require specialist training. Although mainly described below in the context of malaria diagnosis, the method can also be used for other applications requiring detection of birefringent particles in blood or other physiological fluids. For example, the method can be used for detection of gout or other crystal diseases.

Any of the features described above in relation to embodiments of the method of the first aspect may also be implemented in the method of the second aspect.

Embodiments of the present invention will now be described by way of non- limitative example with reference to the accompanying drawings, in which:

Fig. 1 is a flowchart of a method for detection of birefringent particles in a blood sample from a subject;

Fig. 2 is a flowchart showing further detail of the step of processing the blood sample in Fig. 1;

Figs. 3 A and 3B show a sample cartridge for use in the method, Fig. 3 A being a perspective view and Fig. 3B being a cross-sectional view; Fig. 4 is a flowchart showing further detail of the step of imaging the processed blood sample in Fig. 1;

Fig. 5 shows an optical setup that can be used to carry out the method of Fig. 1;

Fig. 6 shows an alternative optical setup for carrying out the method of Fig. 1;

Fig. 7 is a flowchart showing further detail of the step of analysing the image in Fig. 4;

Figs. 8A and 8B shows an image before and after background subtraction;

Fig. 9 shows an example of a program for analysing images to detect birefringent particles in the blood sample;

Figs. 10A to 10E are images of a blood sample containing hemozoin taken using different optical techniques by way of comparison;

Figs. 11 A to 11C are images of different samples illustrating the effect of lysing red blood cells on the captured image; and

Fig. 12 shows results of applying the method to control and test blood samples.

Fig. 1 shows a method for detection of birefringent particles in a blood sample from a subject. The method is carried out using an optical microscope. As discussed above, birefringent particles in blood can indicate the presence of disease and can provide a reliable marker for assessing the presence and severity of that disease. In particular, the birefringent particles may comprise hemozoin crystals. Hemozoin crystals are indicative of the presence of malarial parasites, and therefore of a malaria infection. Compared to other techniques for malaria diagnosis that require labour intensive staining and examination of blood samples or the use of other malarial biomarkers, the current method is more reliable and can more easily be carried by untrained users with inexpensive equipment.

Although detection of hemozoin and determination of malaria are the primary example discussed below, it is not essential in general that the birefringent particles are hemozoin or that the method comprise determining if the subject has malaria. As discussed above, the presence of birefringent crystals may indicate other disease states such as gout, and so the method may also be used in the detection and diagnosis of indicators of other such diseases.

The method comprises processing SI 00 the blood sample. Step SI 00 is shown in further detail in Fig. 2. The processing SI 00 is performed to lyse SI 10 red blood cells in the blood sample. Red blood cells cause significant scattering of incident light in the sample and can obscure the detection of birefringent crystals. Lysing the cells disperses their contents and removes the distinct boundary of their cell wall, thereby reducing scattering by the cells. The effect of lysis of red blood cells on the observed image is demonstrated in Figs. 11 A to 11C. Fig. 11 A shows a blood sample cell free of malarial parasites observed using the optical setup of the present method (described in more detail below) but without lysis of red blood cells. Without lysis, the thick layer of blood (>10 pm) scatters light strongly and makes detection of birefringent particles impossible. Fig. 1 IB shows an image of a 100pm thick layer of blood free of malarial parasites after lysis of red blood cells imaged under lx magnification. As can be seen, there is no signal at all, as is desirable in the absence of any of the particles that the method intends to detect. For comparison, Fig. 11C shows an image of pure water, which appears substantially similar to the healthy blood sample after lysis. This demonstrates the effectiveness of lysis in removing unwanted signal from blood components other than the birefringent particles.

It is desirable to perform the lysis such that damage to the food vacuoles of malarial parasites is reduced, minimised, or prevented. Malarial parasites store hemozoin crystals in vacuoles within their cells. This causes a concentration of hemozoin crystals in one region that makes the crystals more easily detectable, particularly at low magnifications. Reducing damage to the food vacuoles maintains this concentration of crystals and makes it easier to detect a smaller number or concentration of hemozoin crystals in the blood sample.

The processing SI 00 of the blood sample may comprise diluting the blood sample with distilled water. Distilled water can lyse cellular structures by osmotic pressure. The processing SI 00 may comprise freezing and thawing the blood sample. Formation of ice crystals in the blood sample will lyse cells in the blood sample so that their contents are dispersed on thawing. However, using distilled water dilutes the blood sample. Using distilled water and freezing are also both likely to risk lysis of the parasitic food vacuoles. Therefore, the processing SI 00 of the blood sample may instead comprise mixing a detergent with the blood sample. The detergent can be chosen to more selectively lyse cells and other structures within the blood sample depending on their specific properties. A smaller volume of detergent than of distilled water may also be needed to lyse the red blood cells, thereby reducing dilution of the blood sample. Optionally, the detergent is saponin. When using saponin, a concentration of at least 0.5%, optionally at least 1%, optionally at least 2% saponin by volume may be used to ensure lysis of the red blood cells.

The processing S100 of the blood sample is performed to also lyse S120 at least one further blood component in addition to the red blood cells. While red blood cells are the most numerous cell in a blood sample, and therefore the dominant source of unwanted scattering, other cells may also cause unwanted scattering. For example, the at least one further blood component may comprise white blood cells and/or platelets. However it is not essential to lyse further blood components beyond the red blood cells, because adequate performance can be obtained when only red blood cells are lysed.

The processing S100 of the blood sample further comprises loading S130 the processed blood sample into a sample cartridge and mounting the sample cartridge loaded with the processed blood sample onto the optical microscope. An example design of sample cartridge 100 is shown in Figs. 3A and 3B. The sample cartridge 100 shown in Figs. 3 A and 3B is an illustrative example of a sample cartridge that is suitable for use with the present method, but other designs are possible and could be applied for use with the present method.

As shown in the perspective view of Fig. 3 A, the sample cartridge comprises an opening 102 through which the blood sample can be introduced to a cavity 106 via an input channel 108. In this design, the cavity 106 has a volume of approximately 30pl, but in general the volume of the cavity 106 can be varied as appropriate for the specific setup being used. The sample cartridge 100 also comprises a vent 104 connected to the cavity 106 via a vent channel 110 through which air can escape the cavity 106 as it is displaced from the cavity 106 by the blood sample during loading of the sample cartridge 100. The vent channel 110 has a narrower width than the input channel 108.

The sample cartridge is also shown in cross-sectional view in Fig. 3B, the cross-section being taken along the input channel 108. The sample cartridge 100 is composed of three layers, namely a base layer 116, a feature layer 114, and a cover layer 112. The cavity 106, input channel 108, and vent channel 110 are formed by cutting out corresponding regions from the feature layer 114. The feature layer 114 is sandwiched between the base layer 116 and cover layer 112 such that the base layer 116 and cover layer 112 form the ceiling and floor respectively of the cavity 106, input channel 108, and vent channel 110. The opening 102 and vent 104 are formed by sizing and placing the cover layer 112 such that the end of the input channel 108 and vent channel 110 are exposed from underneath the cover layer 112. The feature layer 114 has a thickness t, which can be used to control the volume of the cavity 106, and thus the volume of the blood sample.

While convenient for simplicity of manufacture, this design of sample cartridge is not essential and any other suitable design may be used as appropriate.

Using a sample cartridge may be convenient particularly for untrained users, as it can standardise the volume of blood used in the method based on the size of the cavity 106 and ensure the blood sample is mounted in the optical microscope such that light passes through the blood sample in the intended manner. However, although convenient in many situations, it is not essential to use a sample cartridge and any other suitable sample loading arrangements may be used.

The sample cartridge is preferably free of contaminants such as dust and other particulate matter prior to loading of the blood sample. This reduces scattering due to the contaminants that may produce false positive results. The effect of the sample cartridge is also reduced through imaging S125 the sample cartridge prior to loading of the blood sample into the sample cartridge as shown in Fig. 2. This will be discussed further below.

The sample cartridge may be pre-loaded with a detergent, and the processing SI 00 of the blood sample (specifically the lysis SI 10 of red blood cells and the lysis S120 of one or more further blood components) may be carried out by mixing of the detergent with the blood sample on loading of the blood sample into the sample cartridge. The detergent may be pre-loaded into the cavity 106 of the sample cartridge 100, for example during manufacture of the sample cartridge 100. This is again useful if the method is carried out by untrained users, because the quantity of detergent used for a given volume of blood sample can be standardised to ensure consistent results. The pre-loading also removes a processing step of measuring and adding detergent, thereby reducing the complexity of the operations needed to carry out the method. Optionally the detergent is a dry detergent. This can further reduce dilution of the blood sample.

Returning to Fig. 1, the method comprises imaging S200 the processed blood sample using an optical microscope configured for back scattering dark field microscopy. Back scattering dark field microscopy shows objects of interest as bright spots on a dark background. This reduces background light and makes automated detection of bright spots corresponding to objects of interest easier.

As shown in Fig. 4, the imaging S200 comprises illuminating S210 the processed blood sample with light having a first polarisation state.

Fig. 5 shows an example setup of the optical microscope for carrying out the present method. The processed blood sample 4 is illuminated with light from the light source 2. In Fig. 5, the illuminating S210 of the processed blood sample 4 with light having a first polarisation state comprises illuminating the processed blood sample 4 with light transmitted through a first polariser 6 configured to transmit light having the first polarisation state. This ensures that only light having the first polarisation state is transmitted from the light source 2 to the process blood sample 4. Alternatively, instead of including the first polariser 6, the light source 2 may be configured to emit light having the first polarisation state.

Light from the light source 2 passes through lens 7 and is reflected by rod mirror 12 towards the objective lens 14, passing through the back focal plane 18 of the objective lens 14. After passing through the objective lens 14, the light interacts with the processed blood sample 4. In the setup of Fig. 5, the optical microscope is arranged so that the light from the light source 2 arrives at the processed blood sample 4 travelling substantially parallel to the optical axis of the optical microscope. Some light is reflected from the processed blood sample 4. The reflected light returns through the objective lens 14 and is blocked by the rod mirror 12.

Some of the light arriving at the processed blood sample 4 is back scattered, i.e. scattered by greater than 90°. The present method makes use of the realisation that using a back-scattering configuration is particularly advantageous when imaging small crystals (such as hemozoin). Particles that are small compared to the wavelength of light, including malarial HZ crystals, scatter light nearly uniformly in all directions including back scattering. However, larger objects such as cells and dirt scatter 10s to 1000s of times more light forwards than backwards. Imaging the sample using a back scattering configuration therefore greatly increases the contrast for the particles of interest (e.g. malarial HZ) compared to forwards scattering.

Light that is back scattered by the processed blood sample 4 passes through the objective lens 14 and will bypass the rod mirror 12 and reach the tube lens 16, where it is focussed onto an image sensor 10. The image sensor 10 captures S230 an image of light scattered by the processed blood sample 4 having a second polarisation state different from the first polarisation state. Typically, the first and second polarisation states are linear polarisation states and the first and second polarisation states are orthogonal.

The capturing S230 of the image of light scattered by the processed blood sample 4 comprises capturing an image of light scattered from the processed blood sample 4 and transmitted through a second polariser 8. The second polariser 8 is configured to transmit light having the second polarisation state. This arrangement of using first and second polarisers is known as a crossed-polarisers setup. This ensures that the image captures by the image sensor 10 is only of light that has had its polarisation state altered by interaction with the processed blood sample 4.

Birefringent crystals cause a change in the polarisation state of light passing through them. Therefore, by using this arrangement, the method ensures that the image is of light that has interacted with the birefringent crystals in the processed blood sample. Light which has been reflected from the processed blood sample 4 is blocked by the rod mirror 14. Light scattered by non-birefringent particles in the processed blood sample 4 will still be in the first polarization state, and so will be blocked by the second polariser 8 and not reach the image sensor 10. This makes it easier to detect the birefringent particles of interest.

Fig. 6 shows an alternative setup for the optical microscope for carrying out the present method. In this arrangement, the optical microscope is configured so that light from the light source 2 arrives at the processed blood sample 4 travelling at an angle with respect to the optical axis of the optical microscope. For example, the angle may be greater than 15°, optionally greater than 30°, optionally approximately 45°. The angle of incident light onto the processed blood sample 4 may alternatively be chosen to be close to the Brewster angle to minimize reflected light. This arrangement may be referred to as side scattering, but is still a form of ‘back scattering’ because the majority of light transmitted through the objective lens 14 is still scattered by more than 90° by the processed blood sample 4.

The arrangement of Fig. 6 eliminates the need for the rod mirror 12 because light travels directly from the light source 2 to the processed blood sample 4. In addition, light transmitted through or reflected by the processed blood sample 4 bypasses the objective lens 14 and is stopped by beam blocks 20. This means there is no need to provide the rod mirror 12 to act as a beam block in the optical path from the processed blood sample 4 to the image sensor 10. Bypassing the objective lens with the illuminating light also removes unwanted scattering or reflection from the objective that might contaminate the image.

The capturing S230 of the image of light scattered from the processed blood sample 4 may be performed using an image sensor 10 having at most 8xl0⁶ pixels, optionally at most 2xl0⁶ pixels, optionally at most 5xl0⁵ pixels, optionally at most 2xl0⁵. Using a low resolution sensor may reduce the cost of the apparatus and make it more accessible for use by individual users or in regions with poor access to organised healthcare.

As discussed above, it is advantageous to increase the volume of the processed blood sample 4 that is examined for birefringent particles in a single operation. The method shown in Fig. 1 achieves this in two ways. Firstly, a magnification of the optical microscope is at most lOx magnification. Secondly, a thickness of the processed blood sample 4 through which light travels during the imaging S200 is greater than a depth of field of the optical microscope, and the imaging S200 of the processed blood sample further comprises scanning S220 the focal plane of the optical microscope through the thickness of the processed blood sample 4 during the capturing S230 of the image, as shown in Fig. 4. Although the scanning S220 of the focal plane and the capturing S230 of the image are shown as sequential steps in Fig. 4, they are carried out substantially simultaneously.

While the method illustrated in the figures includes both of these features of a) low magnification and b) scanning of the focal plane through a thick sample, this is not essential and the method can still achieve advantageous results relative to prior art methods using only one of the two features.

Reducing the magnification of the microscope increases the volume scanned significantly because halving the magnification will quadruple the area of sample visible to the microscope. Reducing the magnification also reduces the cost associated with the apparatus used to carry out the method because high-quality, high-magnification lenses are not required. Further reductions in magnification are also possible. The magnification of the optical microscope may be at most 5x, optionally at most 2x, optionally at most lx, optionally at most 0.5x.

Surprisingly, the inventors have found that by using back-scattering dark-field microscopy with polarised light, and detecting only the light with a polarisation different from the incident light, it is still possible to reliably detect the light scattered from the birefringent particles even at magnifications where it is not possible to actually resolve the birefringent particles themselves. This enables the use of low magnifications to image large volumes.

Scanning the focal plane of the optical microscope through a sample thicker than the depth of field of the optical microscope also increases the volume of the blood sample that can be imaged in a single operation. This allows the method to make use of signal from a larger sample while also removing the need for additional processing steps to ensure the sample is thin enough to lie within the depth of field, as would be necessary with the gold standard technique of Giemsa staining.

Typically the depth of field of the microscope would be in the region of 10pm, so the thickness of the processed blood sample through which light travels during the imaging of the processed blood sample is typically at least 10pm. Of course, using a thicker sample enables the imaging of a larger volume, and so preferably the thickness is greater than 10pm, for example at least 20pm. The thickness of the processed blood sample through which light travels during the imaging of the processed blood sample may be at least 50pm, optionally at least 100pm, optionally at least 200pm, optionally at least 500pm, optionally at least 1mm. The thickness of the processed blood sample through which light travels during the imaging of the processed blood sample can be controlled by the design of the sample cartridge 100 and specifically the dimensions of the cavity 106.

The scanning S220 of the focal plane of the optical microscope may comprise moving the processed blood sample 4 relative to an objective lens 14 of the optical microscope along an optical axis of the optical microscope. This may be preferred (for example over moving one or more of the lenses) because it allows all of the optical elements of the apparatus to remain fixed relative to one another, thereby reducing the chance of accidental misalignment.

The scanning S220 of the focal plane of the optical microscope may comprise scanning the focal plane using an actuator. Use of an actuator can allow the scanning to be automated. This will provide improved consistency of processing, particularly in the case where the method is carried out by untrained users. The actuator may be an electrical actuator such as a motor or servo. Alternatively, purely mechanical actuators may be used, such as a clockwork mechanism. This would reduce the electrical power required by the apparatus, which may be valuable in areas lacking reliable electricity supply.

In some embodiments, the image may be an image stack comprising a plurality of image frames captured at successive time points during the scanning S220 of the focal plane. This allows the thickness of the processed blood sample 4 to be fully analysed by capturing image frames at different points through the thickness of the processed blood sample 4.

A rate of the scanning S220 of the focal plane and a time interval between the successive time points may be such that the focal plane moves at most 50pm, optionally at most 20pm, optionally at most 10pm between the successive time points. Ideally, every point through the thickness of the processed blood sample 4 will be taken into account, and so the rate of the scanning S220 of the focal plane and a time interval between the successive time points may be such that the focal plane moves by a distance approximately equal to the depth of field of the optical microscope between the successive time points.

As an alternative to the image comprising an image stack, the image may be a single image frame captured during the scanning S220 of the focal plane. For example, the image may comprise a single image frame captured by a continuous exposure through the entire duration of the scanning S220. In this case, the image comprises a single image frame that integrates the light received at the image sensor 10 during the scanning S220. This may be used as an alternative to determining a maximum projection image, as described further below.

Returning to Fig. 1 and Fig. 4, the imaging S130 of the processed blood sample 4 comprises analysing S250 the image to identify birefringent particles in the processed blood sample 4. Fig. 7 shows further detail of the analysing S250 of the image.

The analysing S250 of the image comprises performing S252 background subtraction on the image. This is important to remove signal in the image from sources other than the birefringent particles.

Performing S252 background subtraction may comprise applying a highpass spatial filter to the image. The birefringent crystals will generally be small in comparison to other features in the image or background light, so a highpass spatial filter preserves the signal due to the birefringent crystals.

Applying the highpass spatial filter may comprise subtracting from the value of each pixel of the image a percentile value of the values of pixels surrounding the pixel of the image. This will mean that signal in pixels very similar in magnitude to those around them is reduced. The percentile value is optionally between a 60^th and a 90^th percentile value, optionally between a 70^th and an 80^th percentile value, optionally a 75^th percentile value.

The surrounding pixels may be, for example, the eight pixels immediately adjacent to the pixel of the image. Alternatively, the surrounding pixels may be those in a region of interest (ROI) around the pixel of the image, for example centred on the pixel of the image.

In the embodiment shown in Fig. 7, the image comprises an image stack as described above and the analysing S250 of the image comprises forming S254 a maximum projection image. The maximum projection image is a single image frame in which a value of each pixel is determined as the highest measured intensity for that pixel across the image stack. The analysing S250 comprises identifying birefringent particles in the maximum projection image. Typically each birefringent particle will scatter a maximum intensity of light to the image sensor 10 when the focal plane of the microscope is centred around the birefringent particle in the thickness of the processed blood sample. The maximum projection image therefore effectively compresses the birefringent particles detected throughout the entire volume of the processed blood sample in the frames of the image stack into a single image frame for ease of analysis.

The identifying of birefringent particles in the blood sample comprises identifying S256 contiguous groups of pixels in the image having an intensity above a predetermined intensity threshold. In Fig. 7, the identifying S256 of contiguous groups of pixels in the image comprises identifying contiguous groups of pixels in the maximum projection image. These are groups of pixels likely to correspond to an actual object in the image and having a minimum level of scattering of the light.

Figs. 8 A and 8B shows an example of background subtraction using this approach. Fig. 8 A shows an image of cultured malarial parasite P. falciparum at 12-15 hrs old. The parasites are diluted in blood lysed by freezing to a level of approximately 200 parasites/pl. The image is taken with a 30ms exposure time and uses a logarithmic grey-scale. The scattered light from the birefringent hemozoin crystals is visible as bright spots, but the background contains significant signal as well.

Fig. 8B shows the image from Fig. 8A after application of a highpass spatial filter by subtracting from each pixel the 75^th percentile values of surrounding pixels in a 5x5 pixel ROI centred on the pixel. A mask is also applied to exclude pixels having a postfiltering intensity below a predetermined threshold. As can be seen, the contrast of the bright spots corresponding to the hemozoin crystals is greatly improved.

The contiguous groups of pixels may be groups of pixels having a number of pixels below a predetermined size threshold. Large objects in the image are likely to be contaminants such as out-of-focus dirt. Smaller objects are more likely to correspond to birefringent particles based on the typical expected sizes of those particles. For example, the predetermined size threshold may be 20 pixels, optionally 10 pixels, optionally 5 pixels. The predetermined size threshold may be chosen based on the magnification of the optical microscope and/or an expected size of the birefringent crystals.

Where the image comprises an image stack and a maximum projection image is formed, the background subtraction may be carried out individually on each image frame before forming the maximum projection image. Alternatively, the background subtraction may be carried out on the maximum projection image.

The identifying of birefringent particles may output a list of identified particles including their size and/or position in the image. The identifying may also output other pixel value statistics, such as distributions of pixel values. The identifying may be carried out using any suitable algorithm, for example a “Best First” algorithm.

Fig. 9 shows an example of an interface of a computer program displaying outputs of the method. At the right of the figure (top of the drawing sheet) is shown the maximum projection image, with coloured spots indicating identified birefringent particles. Various statistics concerning the birefringent particles and pixels of the image are shown at the left of the figure (bottom of the drawing sheet).

Where the image comprises an image stack, the position in the image of the birefringent particles may also comprise a position in the thickness of the processed blood sample. The position in the thickness can be determined from which image frame in the image stack provided the value in the maximum projection image for the pixel(s) identified as corresponding to the birefringent particle.

The positions in the thickness of the birefringent particles can be used to determine other statistics, such as a distribution of birefringent particles through the thickness of the processed blood sample, as shown in the histogram at the bottom-left of Fig. 9. Such a distribution may be used to further improve the analysis S250 of the image, for example by excluding birefringent particles if they are identified at a position in the thickness having an anomalously high number of identified birefringent particles. The number of identified birefringent particles at a given position in the thickness of the sample may be identified as anomalously high if it meets a predetermined condition. For example, the predetermined condition may be that the number is above a predetermined threshold and/or above a predetermined multiple of an average (e.g. median or mean) of the number of identified birefringent particles across all positions in the thickness of the sample. In the example of Fig. 9, the number of birefringent particles identified in the final image frame is much higher than in the other image frames, and so the birefringent particles identified at the corresponding position in the thickness of the sample are likely to actually be dust particles or imperfections on the surface of the sample cartridge. Therefore, the birefringent particles identified at this position in the thickness may be excluded.

As shown in Fig. 2, the method further comprises, prior to loading S130 the processed blood sample into the sample cartridge, imaging S125 the sample cartridge using the optical microscope to obtain a calibration image. The performing S252 of the background subtraction comprises using the calibration image.

The imaging S125 of the sample cartridge is shown occurring after the lysing SI 10 of the red blood cells and the lysing S120 of further blood components in Fig. 2. However, this is not essential and the imaging S125 of the sample cartridge can occur at any point prior to the loading SI 30 of the sample cartridge. It is not essential to image S125 the sample cartridge prior to loading SI 30 the sample cartridge and use the calibration image in the performing S252 of the background subtraction. However, doing so improves the ability of the background subtraction to compensate for light in the image due to, for example, imperfections or contaminants present in or on the sample cartridge.

In the method of Fig. 1, the birefringent particles comprise hemozoin crystals, and the method comprises determining S300 that the subject has malaria depending on the identified hemozoin crystals. In existing “gold standard” techniques based on examining thin blood smears, determining if the subject has malaria may involve calculating a density of malarial parasites in the blood and determining that the subject has malaria if the density is sufficiently high. In the present method, the features of individual parasites may not be resolvable depending on the magnification used. However, each contiguous group of pixels will generally correspond to a single malarial parasite, because the group of pixels corresponds to the HZ crystals within the parasite.

It is possible that at very low magnifications, neighbouring parasites may be imaged as a single group of pixels. However, the density of parasites that would be required for this to be likely is very high, and the subject would typically already be displaying other symptoms of malaria at such an advanced stage of disease. In such cases it would also be possible to reduce the sensitivity (e.g. by increasing the predetermined intensity threshold) to detect only the brightest particles, or to use thinner or more dilute samples.

In the present method, determining S300 that the subject has malaria depending on the identified hemozoin crystals may comprise determining that the subject has malaria if a number of contiguous groups of pixels (also referred to as spots) identified in the image is above a predetermined diagnosis threshold. The contiguous groups of pixels counted for this determination may be those with an intensity above the predetermined intensity threshold and optionally below the predetermined size threshold, as discussed above.

The diagnosis threshold may be set in any appropriate way. For example, it may be set by calculating an expected number of spots for the particular configuration of the optical setup based on a parasite density that is considered sufficiently high for a positive diagnosis using the gold standard method. Alternatively, the diagnosis threshold may be set by taking empirical measurements of samples having different levels of malarial infection.

Figs. 10A to 10E show a comparison of hemozoin imaging under various magnification and optics quality to demonstrate the effectiveness of the optical setup of the present method. The three rows in the figure show images of young malarial parasites (top row, left-hand side of drawing sheet), mid-aged parasites (middle row), and mature parasites (bottom row, right-hand side of drawing sheet).

Fig. 10A illustrates the gold standard technique for malaria diagnostics of Giemsa staining of a thin blood smear. The sample shows malarial parasite Plasmodium falciparum inside a red blood cell under bright field microscopy with a 40x objective lens. The green arrows indicate the hemozoin pigment.

Fig. 10B shows an image taken of the same thin blood smear sample as in Fig. 10A using back-scattering dark-field (BSDF) microscopy with crossed polarisers as described in relation to Fig. 5. The hemozoin is visible as bright spots in the image.

Fig. IOC shows images taken of the same thin blood smear sample as in Figs. 10A and 10B but prior to Giemsa staining. The left-hand column is taken with the same bright- field optical setup as Fig. 10A, and the right-hand column with the BSDF setup of Fig. 10B. As can be seen, the hemozoin is less visible without staining in the bright-field images, especially in the young parasites. Furthermore, the parasites surrounding and containing the hemozoin are nearly invisible without staining. However, the visibility of the hemozoin in the BSDF image is largely unaffected by the lack of staining.

Fig. 10D shows images of the same stained thin blood smear sample as in Figs. 10A and 10B with bright-field (left) and BSDF (right) setups described in Fig. 6, but taken at lx magnification. The image contrast is stretched (to the same extent) in each image to make low-contrast spots appear brighter. In the bright-field image it is impossible to discern the presence of hemozoin at this low magnification. However, the hemozoin is still detectable as a bright spot in the BSDF image even at low magnification. Fig. 10E shows an image of the same sample as Fig. 10D with the same BSDF setup as the right-hand column of Fig. 10D, but taken with low-cost optics. The optics used in Fig. 10D had a cost of approximately £1,400, while the optics of Fig. 10E had a cost of approximately £60. Similar results are obtained as in Fig. 10D, illustrating that the present method can still be carried out even with low-cost equipment, thereby greatly improving its availability.

Fig. 12 shows results of the method carried out on control and infected blood samples. The two control samples are from a previously frozen sample of blood (RBC control) and a fresh sample from a healthy subject lysed by detergent (healthy control).

The 12-15 hrs infected sample had a parasitic load (parasitemia) of 0.004% (verified by parallel Giemsa-stained counting before 2000x dilution). This load corresponds to a predicted-200 parasites per microliter. The 12-15 hrs sample corresponds to the same parasite lifecycle stage (young parasites) as the top row of images in Fig. 10. The schizonts and gametocytes infected samples correspond to the mid-aged and mature parasite of Fig. 10 respectively. Schizonts and gametocytes are generally much easier to detect by Giemsa staining than the young parasites because they are much larger and brighter than 12-15hr old parasites, as illustrated in Fig. 10A. This difference in brightness is also visible in the results of Fig. 12.

Spots (contiguous groups of pixels corresponding to potential hemozoin crystals) were detected in the young parasite sample using the method above, with a sample thickness of 1mm.

The highpass spatial filter was applied by subtracting from each pixel the 75^th percentile value of pixels in a 5x5 pixel ROI centred on the pixel. For this choice of filter parameters, the 75^th percentile value of the 25 pixel ROI would fall between the values of the 6^th and 7^th brightest pixels in the ROI. This means that the central pixel will be set to zero value (black in the image) unless it is brighter than the 6^th or 7^th brightest pixel in the ROI, depending on the exact configuration of the processing.

If pixels were either all either white or black (i.e. maximum or minimum intensity measurable by the image sensor), this implementation of the highpass filtering would exclude all contiguous groups of pixels larger than 6 or 7 pixels, effectively setting the predetermined size threshold. However, the pixels within contiguous groups will in reality have varying brightness profiles, and so 6 or 7 pixels is only an approximation of the typical contiguous group size that is passed in practice in this example.

The pixel intensity threshold for inclusion in a spot was set to 0.1 times the number of milliseconds of exposure multiplied by 1/255 of the saturating pixel intensity, i.e. the intensity that would saturate the pixel of the image sensor. For example, for a 30ms exposure time, the intensity threshold was set to 3/255 of the saturation intensity, and for a 50ms exposure time the intensity threshold was set to 5/255 of the saturation intensity. This particular choice of threshold is not essential and other threshold values may be chosen depending on the setup. For example, a higher threshold may be used for an image sensor with a higher gain.

As can be seen, the number of spots detected in the infected sample was much higher than in either control sample, indicating probable presence of malarial parasites. The schizonts and gametocytes infected samples were only imaged in thin smears so spot counts are not shown because they were less meaningful and not comparable to the spot counts for the young parasite sample.

The present method provides a new way of detecting malaria and other diseases indicated by the presence of birefringent particles in the blood or other physiological fluids. The method can be implemented using a low-cost miniature optical sensor, which can be operated under the control of a user' s computer or smartphone. This can significantly improve the availability and reliability of testing. The method allows direct optical detection of individual malarial parasites in a single droplet of freshly lysed liquid blood. The method also allows data collection by users with little or no medical training, automated counting of parasites, and digital quantitative diagnosis and reporting. The method requires only minimal equipment other than the optical sensor. All that is required is a standard set for finger pricking, and the sample cartridge (if used). The method can also be carried out without needing liquid consumables. Thereby, the present method can provide faster, more reliable, quantitative detection by untrained users in a manner that is not available with existing diagnostics.

The combination of back scattering dark field (BSDF) microscopy with crossed- polarisers to observe a blood sample in which the red blood cells have been lysed allows birefringent particles such as hemozoin to be observed as bright spots. The birefringent particles remain bright and detectable even when a large volume is observed in a single operation by using a low magnification and/or scanning through a thick blood sample.

Claims

1. A method for detection of birefringent particles in a blood sample from a subject comprising: processing the blood sample to lyse red blood cells in the blood sample; and imaging the processed blood sample using an optical microscope configured for back scattering dark field microscopy, wherein: a magnification of the optical microscope is at most lOx magnification; the imaging comprises illuminating the processed blood sample with light having a first polarisation state, capturing an image of light scattered by the processed blood sample having a second polarisation state different from the first polarisation state, and analysing the image to identify birefringent particles in the processed blood sample.

2. A method according to claim 1, wherein the processing of the blood sample further comprises loading the processed blood sample into a sample cartridge and mounting the sample cartridge loaded with the processed blood sample onto the optical microscope.

3. A method according to claim 2, wherein the sample cartridge is pre-loaded with a detergent, and the processing of the blood sample is carried out by mixing of the detergent with the blood sample on loading of the blood sample into the sample cartridge, wherein optionally the detergent is a dry detergent.

4. A method according to any preceding claim, wherein: a thickness of the processed blood sample through which light travels during the imaging is greater than a depth of field of the optical microscope; and the imaging of the processed blood sample further comprises scanning the focal plane of the optical microscope through the thickness of the processed blood sample during the capturing of the image.

5. A method according to claim 4, wherein the thickness of the processed blood sample through which light travels during the imaging is at least 50pm, optionally at least 100pm, optionally at least 200pm, optionally at least 500pm, optionally at least 1mm.

6. A method according to any one of the preceding claims, wherein the magnification of the optical microscope is at most 5x, optionally at most 2x, optionally at most lx, optionally at most 0.5x.

7. A method according to any one of the preceding claims, wherein the first and second polarisation states are linear polarisation states, and/or the first and second polarisation states are orthogonal.

8. A method according to any one of the preceding claims, wherein the illuminating of the processed blood sample with light having a first polarisation state comprises illuminating the processed blood sample with light transmitted through a first polariser configured to transmit light having the first polarisation state.

9. A method according to any one of the preceding claims, wherein the capturing of the image of light scattered by the processed blood sample comprises capturing an image of light scattered from the processed blood sample and transmitted through a second polariser, the second polariser configured to transmit light having the second polarisation state.

10. A method according to claim 4 or any preceding claim dependent thereon, wherein the scanning of the focal plane of the optical microscope comprises moving the processed blood sample relative to an objective lens of the optical microscope along an optical axis of the optical microscope.

11. A method according to claim 4 or any preceding claim dependent thereon, wherein the scanning of the focal plane of the optical microscope comprises scanning the focal plane using an actuator.

12. A method according to any one of the preceding claims, wherein the capturing of the image of light scattered from the processed blood sample is performed using an image sensor having at most 8xl0⁶ pixels, optionally at most 2xl0⁶ pixels, optionally at most 5xlO⁵ pixels, optionally at most 2xl0⁵.

13. A method according to claim 4 or any preceding claim dependent thereon, wherein the image is an image stack comprising a plurality of image frames captured at successive time points during the scanning of the focal plane, optionally wherein a rate of the scanning of the focal plane and a time interval between the successive time points are such that the focal plane moves at most 50pm, optionally at most 20pm, optionally at most 10pm between the successive time points.

14. A method according to claim 13, wherein the analysing of the image comprises forming a maximum projection image in which a value of each pixel is determined as the highest measured intensity for that pixel across the image stack and identifying birefringent particles in the maximum projection image.

15. A method according to any one of the preceding claims, wherein the identifying of birefringent particles in the blood sample comprises identifying contiguous groups of pixels in the image having an intensity above a predetermined intensity threshold, optionally wherein the contiguous groups of pixels are groups of pixels having a number of pixels below a predetermined size threshold.

16. A method according to any one of the preceding claims, wherein the analysing of the image comprises performing background subtraction on the image.

17. A method according to claim 16, wherein performing background subtraction comprises applying a highpass spatial filter to the image, optionally wherein applying the highpass spatial filter comprises subtracting from the value of each pixel of the image a percentile value of the values of pixels surrounding the pixel of the image, the percentile value being optionally between a 60^th and a 90^th percentile value, optionally between a 70^th and an 80^th percentile value, optionally a 75^th percentile value.

18. A method according to claim 16 or 17 when dependent on claim 2, wherein the method further comprises, prior to loading the processed blood sample into the sample cartridge, imaging the sample cartridge using the optical microscope to obtain a calibration image, and wherein performing background subtraction comprises using the calibration image.

19. A method according to any one of the preceding claims, wherein the processing of the blood sample comprises diluting the blood sample with distilled water.

20. A method according to any one of claims 1 to 18, wherein the processing of the blood sample comprises mixing a detergent with the blood sample, the detergent optionally being saponin.

21. A method according to any one of the preceding claims, wherein the processing of the blood sample is performed to also lyse at least one further blood component in addition to the red blood cells, optionally wherein the at least one further blood component comprises white blood cells and/or platelets.

22. A method according to any one of the preceding claims, wherein the birefringent particles comprise hemozoin crystals, optionally wherein the method further comprises determining that the subject has malaria depending on the identified hemozoin crystals.

23. A method for detection of birefringent particles in a blood sample from a subject comprising: processing the blood sample to lyse red blood cells in the blood sample; imaging the processed blood sample using an optical microscope configured for back scattering dark field microscopy, wherein: a thickness of the processed blood sample through which light travels during the imaging is greater than a depth of field of the optical microscope; the imaging comprises illuminating the processed blood sample with light having a first polarisation state, capturing an image of light scattered by the processed blood sample having a second polarisation state different from the first polarisation state while scanning the focal plane of the optical microscope through the thickness of the processed blood sample, and analysing the image to identify birefringent particles in the blood sample.

24. An apparatus comprising an optical microscope and configured to carry out the method of any of the preceding claims.

25. A computer program comprising or a computer-readable medium having stored thereon instructions which, when carried out by the apparatus of claim 24, cause the apparatus to carry out the method of any of claims 1 to 23.