Testing of micro-fluidic systems for Raman spectroscopic measurements on biological cells

Testing of micro-fluidic systems for

Raman spectroscopic measurements

on biological cells

Malin Berger

Engineering Physics and Electrical Engineering, bachelor's level 2018

Testing of micro-fluidic systems for

Raman spectroscopic measurements on

biological cells

Malin Berger

Bachelor Thesis in Engineering Physics and Electrical Engineering

Supervisor: Kerstin Ramser

Abstract

Pulmonary Artery Hypertension (PAH) is a condition that can affect people as a consequence of infections or diseases such as lung disease, high blood pressure or pneumonia. When afflicted by these diseases, low oxygen content in the lung tissue causes the pulmonary arterial soft muscle cells (PASMC) located in the walls of the pulmonary arteries to chronically swell up. As a result, the arteries are constantly narrowed. This can in many cases be fatal as the arteries become clogged and the heart is forced to pump more blood to the lungs, causing an enlargement of the right heart chamber which eventually may lead to heart failure. This irreversible swelling of the PASMC is the cause for PAH. To find a treatment for this incurable disease, the mechanisms of the vasoconstriction need to be investigated.

Spectroscopy is the study of the interactions between light and matter and is a tool that can be used to gain knowledge in the matter of the expansions of the PASMC. In particular, Raman spectroscopy that targets the inelastic interactions can be used since it registers dynamic changes of cells.

To simulate an oxygen deprived environment, a micro-fluidic system designed for use in cellular experiments has been developed. Tests of the prototypes showed strong Raman signals from the polymeric material of the system itself. These signals over-shadowed the signals from the observed sample. The objective of the experiments presented in this report was to test whether the signals from the micro-fluidic system could be eliminated by adding spacing between the polymer and the sample.

The experiment was conducted by collecting data of samples from baker’s yeast pre-pared in the micro-fluidic system at different z-distances. By this the optimal spacing between the polymer of the micro-fluidic system and the sample could be determined. This experiment concluded that the sample needed to be placed 1.54 mm further from the micro-fluidic system in order to test human lung tissue at 2 mW laser intensity.

Acknowledgements

During this project I have had great support from many people.

First and foremost, I want to thank Kerstin Ramser for being an excellent supervisor. Thank you for pushing me to do my very best, and for being patient as I asked the same questions over and over. I also owe a big thanks to Joel Wahl for all help in MATLAB. Without your guidance I would have been lost in all the code. A big thanks to Hillevi Sandell for proof-reading the report, and for your brutally honest comments. Lastly, thank you Jonathan Hjelm for all the times you cooked, cleaned, made the dishes and brought me tea during late nights of writing.

Lule˚a, May 2018. Malin Berger

Glossary

CCD - Charged-coupled Device.

PAH - Pulmonary Artery Hypertension.

PASMC - Pulmonary arterial soft muscle cells. PBS - Phosphate-buffered saline.

PC - Polycarbonate.

PMMA - Polymethyl Methacrylate. PS - Polystyrene.

Contents

3 Background and concepts . . . 5

3.1 Spectroscopy . . . 5

3.1.1 Vibrational Raman scattering . . . 5

3.1.2 Fluorescence . . . 7

3.2 The micro-fluidic system . . . 8

3.3 Raman spectrometer components . . . 9

3.3.1 The compound light microscope . . . 9

3.3.2 Spectrometer . . . 10

3.4 Interpreting the data . . . 10

3.5 Baker’s yeast . . . 11

4 Experiments and signal analysis . . . 13

4.1 Experimental set-up . . . 13

4.2 Testing on polymeric materials . . . 14

4.3 Testing on baker’s yeast . . . 14

4.4 Z-experiments on PMMA micro-fluidic system . . . 15

4.4.1 Performing the experiment . . . 15

4.4.2 Determining the distance between the sample and the polymer 15 4.5 Signal analysis . . . 16

5 Analysis of results and data . . . 20

5.1 Result from polymeric measurements . . . 20

5.2 Results from baker’s yeast measurements . . . 21

5.3 Results from z-experiments . . . 22

6 Discussion and conclusions . . . 25

References . . . 27

1 Introduction

In the human body, blood travels from the lungs through the pulmonary arteries to the heart. The walls of these arteries contain pulmonary arterial soft muscle cells (PASMC) that have the ability to change their shape and swell up if little oxygen is present in the surrounding tissue. In the event of low oxygen levels, the arteries vasoconstrict (become narrower), they get clogged, and the blood is directed into regions which are full of oxygen. When the blood has been redirected to an oxygenated region the cells can relax and return to their normal, thin shape. However, if the oxygen content stays low for longer times, like under pneumonia or anesthesia, this process can become irreversible. This is called Pulmonary Artery Hypertension (PAH). In this case, the heart must pump more blood into the lungs, which leads to an enlargement of the right hearth chamber, eventually leading to heart failure [1] [2].

According to Socialstyrelsen there is an increase of people in Sweden who are diag-nosed with PAH due to attentiveness regarding symptoms [3]. The issue is that PAH is an incurable disease. The only treatment available are methods of slowing down the process [3]. Even though there has been much research done in the field, there is still not a full understanding regarding the form shift the lung cells exhibit, making the disease difficult to cure. Raman spectroscopy has the potential to answer some of these questions.

With Raman spectroscopy the specimen (in this application surgically removed or cultured lung tissue) can be irradiated by a laser and a Raman spectrum can be retrieved. By taking spectra of cells under oxygen deprived situations and comparing them to spectra of fully functional cells, a deeper understanding regarding the form shift of the cells can be gained.

To measure the spectrum of the cells during oxygen deprived situations an air-tight micro-fluidic system is necessary. Two prototypes of the system have been developed using two different polymeric materials (plastics), polycarbonate (PC) and polymethyl methacrylate (PMMA). Upon testing the prototypes, strong Raman signals from the polymers were detected which overshadowed the signals from the specimen.

2 Thesis objectives

The objective of the measurements presented in this paper is to test whether the signals from the micro-fluidic systems can be eliminated by adding spacing between the systems and the sample, see figure (1). This will be done by collecting the Raman spectra of yeast cells in one of the micro-fluidic systems, and changing the measurement point in the z-direction away from the sample to see at which height the polymer signal vanishes.

Figure 1: Drawing of the micro-fluidic system from its side. The sample of yeast has been diluted using PBS, which is a liquid solvent commonly used in biological research. The objective of the report is to test if the polymer signals from the micro-fluidic system can be eliminated by adding spacing between the polymer of the system and the sample.

3 Background and concepts

This section will explain the theory of the methods, equipment and materials used during the experiments.

3.1 Spectroscopy

Optical spectroscopy is the study of the interactions between light and matter. It is used in astronomy to identify constituents in distant stars and galaxies, it is used in hospital procedures such as x-rays and it is used in research to identify the chemical composition and physical structure of matter. These are just a few examples to illustrate how the techniques are utilized, but applications can be found in most fields of technological science [4].

In this thesis vibrational Raman spectroscopy has been used since it targets molecular vibration of cellular components without interference of water as compared to infrared spectroscopy. Furthermore, it has a better resolution than absorption spectroscopy.

3.1.1 Vibrational Raman scattering

Raman scattering can be explained using quantum mechanics, but for simplicity it will here be explained in a classical approach. In classical physics, light is seen as electromagnetic radiation with an oscillating electric field, ~E. If the radiation interacts with matter in an inelastic manner, an electric dipole, ~p, is induced. This electric dipole is dependent on the electric field and the polarizability, α, of the studied specimen [5].

~

p = α ~E (3.1)

As the polarizability varies a change in the dipole moment can be observed. This is the cause for inelastic scattering [5].

Most scattering will be performed elastically, called Rayleigh scattering. In this sce-nario the interactions of light and matter does not result in an induced dipole. Here the incoming photon, ν0, will have the same frequency as the outgoing photon [6].

A small part of the scattering will happen in an inelastic manner. There are two types of inelastic scattering. If the molecule initially is in the ground state, S0, and

is hit by a photon of less energy than the excitation energy of the molecule, it will be pushed up to a virtual excitation level, S00. While it drops back down and emits

a photon it will stay in a higher vibrational energy stage, thus emitting a photon of lower frequency. This is called Stokes Raman scattering and will be the main focus of this report. [6].

νout = ν0− νv (3.3)

If the molecule initially is in an excited vibrational state and returns to the ground state a photon of higher frequency and shorter wavelength will be emitted. This is called anti-Stokes Raman scattering [7].

νout = ν0+ νv (3.4)

This can be further understood by the schematic drawing in figure (2). Here νv is the

vibrational frequency caused by the inelastic scattering.

Figure 2: Different forms of light scattering: a) Rayleigh scattering, b) Stokes Raman scattering, c) anti-Stokes Raman scattering. The incoming photon causes a jump from the initial state S0 to a virtual energy state S00 followed by

an emission of a photon with higher, lower or unchanged frequency. The virtual energy state lies between the ground state and the first excited state S1.

3.1.2 Fluorescence

Fluorescence is a process which occurs naturally in organic molecules due to the molecular structure and the bonds between the atoms [8]. Since this effect is difficult to explain purely by classical physics, we first need to venture into the field of quantum mechanics.

In 1913 Niels Bohr presented a theory in which he explained that a substance would have a set of predetermined energy levels, which differed greatly from classic theory [9]. This model, while extraordinary in its thought, has later been built upon as it only works for atoms as simple as hydrogen. For more complicated atoms a more intricate model was needed and quantum mechanics soon started to take its shape [9].

The more advanced models explains the energy levels with quantum numbers and sub-levels. Here, the first excited level does not only contain one level. It contains several sub-levels. This means that the absorption and emission can go from S0 to S1

and back to ground level, but still emit a photon of higher or lower frequency. This is known as fluorescence [10].

Figure 3: A diagram showing the sequence of events leading to fluorescence. A photon cause an excitation to a higher level. The transition between the sub-levels cause no photon emission. This causes the transmitted photon to have another wavelength than the incoming photon, thus another energy.

Fluorescence has a lot of applications in medical an chemical analysis of organic matter due to its high sensitivity [11], but in Raman spectroscopy the detection of fluorescence is more a disadvantage. By comparing figure (2) and (3) it is seen that

the events of fluorescence is a process of higher frequency and the spectrometer used to collect the data will register both the Raman signals and the fluorescence signals during experiments as it registers differences in wavelength. If the specimen exhibits intrinsic fluorescence, the fluorescence signals will overshadow the Raman peaks. The ways to prevent fluorescence from an intrinsic sample is to either use a laser of different wavelength which does not generate fluorescence, or to filter away the fluorescence using signal analysis [12].

3.2 The micro-fluidic system

In order to measure the Raman spectrum of cells or other biological matter in an oxygen deprived environment a micro-fluidic system has been developed.

The main function of the system is to control the oxygen content, therefore two of them have been constructed using PC and PMMA. These polymers are suitable for this application due to their low oxygen permeability [13][14].

Figure 4: Picture showing the function of the micro-fluidic system.

The system has several channels, as seen in figure (4). There is one inlet and an outlet that can be used to flush reactants through the sample [15]. There is also a channel that fits an optical fiber that measures the oxygen content, and a tunnel that can fit an electrode for voltage measurements. The top of the component has a larger hole that fits a small pipette with an integrated electrode.

3.3 Raman spectrometer components

The Raman spectrometer is a set-up consisting of an excitation source (a laser), a sampling device (a microscope) and a detector (spectrometer) [12].

In order to observe a Raman spectra the specimen needs to be excited, and the most common way of achieving this is by illuminating the molecules with a laser. The most important requirement of the laser is that it is frequency stable which makes the instrument quite expensive. Several different types of lasers and wavelengths may be used depending on the specimen and application [12]. The following subsections will be focused on the microscope and the spectrometer as they both have components worth further mentioning.

3.3.1 The compound light microscope

A traditional compound light microscope consists of an eyepiece, an objective lens, a condenser lens and a light source. In these experiments however, the light never hits the eyepiece as it is re-directed to a charged-coupled device camera (CCD-camera) that is used to view the sample on the computer screen. The light passes through the condenser lens. Its purpose is to focus the light onto the specimen that is to be observed to gain a sharp image when magnified. The light travels through the specimen onto the objective lens. This is the most important part of the microscope since it is the component accountable for the highest magnification [16]. The light hits a mirror that redirects it into a CCD-camera connected to a computer which has the ability to take photos of the specimen [17]. A schematic drawing of this can be seen in figure (5).

Figure 5: Schematic drawing of the compound light microscope used during experi-ments.

3.3.2 Spectrometer

A spectrometer is an optical instrument that can detect spectral lines and measure their wavelengths and intensities [12].

The Raman signals are collected from the microscope through an optical fiber. The fiber leads the light into the detector through a slit. The purpose of the slit is to block stray light from the surrounding to enter the spectrometer, and it plays an important part in the optical resolution of the final image as it controls the spectral width [18]. Next the light is redirected through a mirror to the collimating mirror which focuses the light upon the grating. The grating splits up the light into its different wavelengths and determine what range of wavelengths to be analyzed [19]. Next is the focusing mirror which redirects the light through an exit slit into the CCD. This is the heart of the detector and it registers all the separated wavelengths from the grating. For high performance spectroscopy the unit needs to be cooled in order to reduce dark noise [20]. This set-up can be viewed in figure (7).

Figure 6: Schematic drawing showing the functions of a spectrometer.

3.4 Interpreting the data

In a Raman spectrum the intensity is plotted versus the wave number. The wave number shows the energy difference between the incoming and the emitted photons in inverse centimeters. The intensity is an arbitrary measure on the likelihood that

Figure 7: An example of how a Raman spectrum may look like. This is a spectrum of PC. The likelihood of achieving Raman scattering (The intensity) is plotted against the energy difference of the incoming and emitted photons (Raman shift).

3.5 Baker’s yeast

Figure 8: The building blocks of the yeast cells.

The experiments will be performed on baker’s yeast. It is used because of its properties as a eukaryotic organism which share many basic biological properties with the human cells. It is also easy to access and can be bought in any super market [21].

The baker’s yeast cells are a strand of the fungi S. cerevisiae and has an elliptical or egg shaped form and a size of roughly 6 µm. A cell wall and a membrane encloses

the cell. Inside the cell the nucleus, vacuole, mitochondria and the cytoplasm can be seen [22]. A drawing of this can be seen in figure (8). The cell contain carbohydrates, proteins, lipids, DNA, and a lot of other substances much like the human cells. The baker’s yeast studied here also shares about half its genes with mammals. Its also a unicellular, making it easier to study than human tissue [23].

4 Experiments and signal analysis

In this section the experimental set-up, the experiments performed and the signal analysis of the raw data will be explained. The experiments covered in this section consists of:

• Raman spectra of polymeric materials. This will be performed to get familiar of the expected outputs of these spectra.

• Raman spectra of baker’s yeast will be collected to be accustomed to the place-ment of the largest peeks in the yeast spectrum. Also, a spectrum of baker’s yeast inside the micro-fluidic system will be measured to ensure that the com-ponent does not function as intended before moving on the further experiments. • Z-experiments on PMMA micro-fluidic system will be performed the conclude

the optimal distance between the sample and the polymer.

• Signal analysis of raw data. This step will be performed to process the raw data to optimize the analysis of it.

4.1 Experimental set-up

Figure 9: Schematic drawing of the experimental set-up used during experiments. The laser light was first redirected through a microscope objective and onto the specimen. The laser light was then transferred to the CCD-camera and computer, and also through an optical fiber to the spectrometer. The spec-trometer examined the shift in wavelengths and sent the Raman spectrum to the computer.

All experiments were performed using a 532nm green Yd:NVO4 laser, two

micro-fluidic system prototypes, a compound light microscope, a spectrometer equipped with a 1200 lines/mm grating, and fresh baker’s yeast. A schematic drawing of the set-up can be viewed in figure (9).

4.2 Testing on polymeric materials

The first tests were made on slabs of polystyrene (PS) and on the two empty micro-fluidic systems. The slab of PS was tested to see if the set-up was sufficiently cali-brated and if the path of light was correctly lined up. This was also an opportunity to get to know the equipment since PS does not exhibit any fluorescence. The Raman measurements could be analyzed on the spot without signal analysis.

The PS was tested by focusing the microscope on the surface and irradiating it with the laser with an intensity of 5 mW using an exposure time of one second. The largest peak in the PS spectra is known to be at x=1001.4 cm-1 _{[24], so the set-up}

was calibrated to align the Raman peak to this position. The spectrum can be seen in figure (15) on page 20.

After the calibration was done, the micro-fluidic systems constructed of PC and PMMA could be tested. They were tested in the same manner as the PS, but the exposure time was set to 20 seconds. The exposure time was increased as PC and PMMA is not as Raman active as PS. These spectra can be seen in figure (16) and (17) on page 20-21.

4.3 Testing on baker’s yeast

The sample of baker’s yeast first needed to be diluted with Phosphate-buffered saline (PBS) as the specimen needed to be in a liquid form when studied in the micro-fluidic system. PBS is a liquid solution often used in biological research as it is non-toxic to biological cells and has the same salt level as the human body [25]. 1.8 ml of PBS was mixed with a tiny amount of yeast.

The experimental testing of the yeast was first performed on a microscope slide to get familiar with the expected output and to fine-tune the system in preparation for the crucial measurements with the micro-fluidic systems. The background radiation was collected by measuring on a microscopic slide with only PBS. After that the prepared sample of PBS and yeast was put on the microscopic slide and measured using a laser

The measurements on the micro-fluidic system were first done with just the PBS inside. After these testings were done, the PBS and yeast solutions were tested inside the system. These measurements were performed several times with integration times from 120 to 180 seconds. The laser intensity was varied to try to find the best setting and it was found to be 5 mW. A spectrum of yeast inside the system constructed of PMMA can be seen in figure (19) on page 22.

4.4 Z-experiments on PMMA micro-fluidic system

The prototypes of the micro-fluidic system have been seen to show strong signals from the polymeric materials of which they are constructed. Therefore, an experiment testing whether they can be improved by adding distance between the sample and the polymer were performed. An explaining picture of this theory can be seen in figure (1) in the introduction of the report on page 4.

4.4.1 Performing the experiment

The experiment was performed on the micro-fluidic system constructed of PMMA, and it was prepared with the yeast solution inside of it. The microscope was focused on the yeast cells and a Raman spectrum was measured using a laser intensity of 10 mW and an exposure time of 120 seconds. Using the microscopes focusing wheel, the focus was shifted 100 µm away from the sample and a new Raman spectrum was collected. The process was repeated until the peaks corresponding to the polymer had been reduced or vanished. The yeast cells were used to find the correct focus at the beginning of the experiment, and because of the small size of the cells the Raman signals from the yeast quickly vanished. As the objective of the experiment was to gain a spectrum without any peaks from the PMMA the yeast signals were not relevant.

4.4.2 Determining the distance between the sample and the polymer

Two procedures were developed to determine at what distance from the polymer the sample should lie. The first one was to do the experiment explained in the section above until the Raman peaks from the polymer vanished. As this was seen to be a time-consuming process, measurements were taken until the polymer signals was reduced. After measurements, one peak in the spectrum was selected. The data point of each measurement corresponding to the largest value (the top of the selected peak) was stored in a vector and plotted against the z-distance from focus in MATLAB.

From these points a first-degree polynomial estimation could be retrieved. The dis-tance was then estimated by examining where the estimated polynomial intersected the z-axis of the plot. This process is shown in figure (10).

Figure 10: Picture showing the theory behind the z-measurements. The intersection point of the estimated first-degree polynomial and the z-axis is the distance between the polymeric material of the micro-fluidic system and the sample. This distance is needed in order to not get disturbing signals from the polymer when measuring the Raman spectra of the sample.

The experiment explained above rests on the theory that Raman spectroscopy is a linear process. This means that if a laser of lower intensity is used, the peaks of the spectra will be lower. Hence, by studying figure (10), it is seen that the required distance between the polymer and the sample will be smaller if the experiment were to be performed using a laser of lower intensity.

4.5 Signal analysis

The raw data obtained during Raman measurements needed to be processed to remove the cosmic ray spikes, fluorescence, noise and background radiation. Without this signal analysis the Raman peaks were barely visible. As a finishing step the data was normalized. All the steps explained below are implicated in MATLAB

Cosmic ray spikes in the Raman spectra are inevitable due to the sensitivity of the CCD used in the detector [26]. Since these are abnormal signals not relevant to the study of the spectrum, they needed to be removed. This was the first step of the signal analysis. They were removed by usage of the second derivative and polynomial fitting. For the data obtained during measurements, a third-degree polynomial was

calculated, which produced a band of noise centered around a common level. Here the deviating values (the ray spikes) could clearly be seen. The width of the band was calculated by the standard deviation of the data multiplied by the square root of two [26]. Thereafter, the program identified the deviating values and ignored them in the polynomial fitting, producing a curve with no cosmic ray spikes. The raw data can be seen in figure (11) and the data after the ray removal process can be seen in figure (12).

(a) (b)

Figure 11: Raw data of (a) sample containing yeast with a ray spike at approximately x=1750 cm−1, (b) background radiation with a ray spike at approximately x=1900 cm−1. For this data a laser intensity of 10 mW and an exposure time of 120 seconds was used.

The ray spike removal process was repeated for several raw data measurements taken at the same place in the sample and at the same exposure time and laser intensity. When the spikes had been removed, the mean of all raw data was calculated to produce one single set of data points. This was done to improve signal to noise conditions in measurements. This was done for all data except that of the polymeric materials, as these measurements were exact enough to not require a mean.

Next, noise was reduced with a second order Savitzky-Golay filter [27]. The filter fits a polynomial to a small segment of data points centered around every point in the signal. The original values are replaced with the center values of the fitted segments, producing a new curve with less noise. For the data from experiments performed on yeast, 2*12+1 data points were used. The collected data of the background radiation was smoothed using 2*25+1 data points. The reasoning behind using bigger segment size while filtering the background radiation was to intentionally make these

measure-(a) (b)

Figure 12: Data after cosmic ray spike removal of (a) sample containing yeast, (b) background radiation. Both spikes seen in figure (11) has successfully been eliminated.

ments smoother than the yeast measurements. This was done to not introduce more noise into the data as the background later was removed. The data after smoothing can be seen in figure (13).

(a) (b)

Figure 13: Data after filtering using the Savitzky-Golay filter (smoothing) of (a) sam-ple containing yeast, (b) background radiation.

from the environment) and fluorescence from the measurements [28]. The background radiation and fluorescence will simply be called ”fluorescence” in the upcoming dis-cussions to not be confused by the background radiation measurements done in the lab. A second-degree polynomial was used for the fitting. The polynomial was cal-culated and compared to the original curve. If the calcal-culated value was larger than the original one it was used, otherwise it was discarded. This process was repeated until all coordinates of the new curve were below the raw data. Here about 1000 iterations was found to be sufficient for all analyzed data. Pictures of the estimated fluorescence of the data can be seen in figure (14). The estimated fluorescence was then subtracted from the data.

(a) (b)

Figure 14: Estimated fluorescence (dotted line) of (a) sample containing yeast, (b) background radiation.

After all data had been processed, the data from measurements without yeast was scaled and subtracted from the corresponding measurements containing yeast, pro-ducing a curve only containing the Raman peaks from the yeast cells.

The last step of the signal analysis was to normalize the curve. This was done using the build in function ”norm” in MATLAB.

The computer analysis for the polymers were made using the same steps as explained above, but no background measurements was collected and subtracted.

5 Analysis of results and data

In the following section the result from the experimental measurements will be pre-sented and analyzed.

5.1 Result from polymeric measurements

Figure 15: Spectrum of PS after signal analysis. A reference line is inserted at x=1001.4 cm−1. This is the desired placement of the tallest peak of the PS spectra [24], and the system was calibrated to have this placement.

Figure 16: Spectrum of PC after signal analysis. The tallest peaks has been marked by dotted lines at (from left to right): x=633 cm−1, x=705 cm−1, x=826 cm−1, x=886 cm−1, x=1112 cm−1, x=1180 cm−1, x=1238 cm−1and x=1605 cm−1 [29].

Figure 17: Spectrum of PMMA after signal analysis. The tallest peaks has been marked by dotted lines at (from left to right): x=598 cm−1, x=811 cm−1, x=987 cm−1, x=1187 cm−1, x=1455 cm−1 and x=1732 cm−1 [30].

5.2 Results from baker’s yeast measurements

Figure 18: Mean of seven spectra of yeast cells on a microscope slide. A large elevation around x=850 cm−1 can be seen. This peak is not relevant to the yeast spectrum as it is the Raman peak the microscopic glass slide exhibits. The relevant peaks and their corresponding meaning is marked in the plot [31] [32].

Figure 19: Mean of three spectra of yeast cells in the micro-fluidic system constructed from PMMA. The dotted lines mark the peaks from figure (18) where yeast can be seen and the solid lines corresponds to the peaks from PMMA seen in figure (17). All large peaks corresponds to the PMMA.

In figure (19) a spectrum of yeast cells inside the micro-fluidic system constructed of PMMA can be viewed. This was the first result attained while measuring in the system using yeast inside of it. By comparing this spectrum with figure (18) where a pure yeast spectrum was measured, it is seen that this result is less clear. Although the spectrum shows several important yeast peaks, most of them coincide with the peaks corresponding to the PMMA. It is therefore difficult to conclude there is any biological significance to them as they might belong to the yeast or the PMMA. The conclusion drawn from this was that the Raman signals from the polymeric material were too high, overshadowing the yeast signals.

5.3 Results from z-experiments

The z-experiment was performed and some of the raw data from the measurements can be seen in figure (20). Measurements were abandoned after the focus had been changed to 6 mm below the sample as the signals from the PMMA had not vanished. The distance between the sample and the polymer were therefore determined using the method explained in section 4.4.2. The result of this experiment can be seen in figure (21). Here a peak at x=1455 cm−1 was chosen for the experiment.

Figure 20: Raw data from measurements. From top to bottom: Spectrum of yeast 1) in focus, 2) 600 µm below focus, 3) 800 µm below focus, 4) 1 mm below focus, 5) 2 mm below focus, 6) 3 mm below focus, 7) 4 mm below focus, 8) 5 mm below focus, 9) 6 mm below focus. The vertical lines mark the peaks seen in the spectra. These are the same peaks shown in figure (17), therefore they correspond to the material of the micro-fluidic system, PMMA.

Figure 21: The largest value of the x=1455 cm−1 peak at different focuses, plotted against the distance. The straight line is an estimate of the values and is seen intersecting the z-axis at about z=7.7 mm.

As seen in figure (21) the sample needs to be placed 7.7 mm further down from the polymer when experiments in the micro-fluidic system are performed at this laser intensity. However, when this setup is to be used to measure human lung tissue, the experiment will not be performed with an intensity of 10 mW. Instead approximately 2 mW laser intensity will be used. The low intensity is used to ensure that the cells of the human lung tissue will not be damaged, which it would at the laser intensities used during the experiments performed in this report. This was unfortunately something that was overlooked during experiment and a laser of to high intensity was constantly used.

As discussed in section 4.4.2, the result from figure (21) can be used to determine at what z-distance from the polymer the sample should lie while executing the experi-ment with a laser of lower intensity. Since only one fifth of the laser intensity is to be used during real experiments, the sample should therefore lie 7.7/5=1.54 mm further away from the polymer.

6 Discussion and conclusions

The objective of this thesis was to assess whether the Raman signals from the polymer of the micro-fluidic system could be eliminated by adding spacing between the poly-meric material of the system and the observed sample. First the system was tested using baker’s yeast cells to ensure that it did not work as intended. The result of this was that the system, as suspected, gave rise to Raman signals from the polymer which obstructed the Raman signal from the yeast cells. Z-experiments concluded that 7.7 mm extra spacing was needed to block out the signals from PMMA at a laser intensity of 10 mW. Calculations showed that 1.54 mm extra spacing was needed if experiments were to be performed using a laser of 2 mW intensity.

During experiments there were several sources of error such as measurement settings, the specimen and the analysis of the data. The calibration of the system was one of the first obstacles. There was a lot of parameters that could be adjusted, such as the positioning of the sample, the focusing of the laser and calibration of the computer software. A lot of small adjustments could lead to different results. This took some getting used to. As the data gained from measurements needed to be analyzed to determine whether the reading was good or bad, the measurement settings were difficult to define. To execute the experiments two computers were used and both computers had software to control aspects of the measuring process. One computer was used to control the microscope, and one controlled the spectrometer. The laser was controlled manually. Besides this there was a path of light that needed to be perfectly aligned. With all this equipment and all the settings available, errors in measurements were prone to happen. Also the specimen used during experiments were baker’s yeast diluted in saline buffer solution allowing the cells to move during measurements. This movement could in some cases lead to a loss of signal.

Although the experiments proved to be more difficult than expected, the most chal-lenging part of this thesis was the signal analysis. All programs needed for the signal analysis needed to be created, which was a time-consuming process and could not be completed before testing began. This caused a certain time delay in the project and after analysis many measurements had to be repeated.

While starting this project I had never worked with, or heard of, Raman spectroscopy. During the course of the experiments I constantly gained more knowledge and a greater understanding for the power of the small effects that can be registered. I have also gained a deeper understanding regarding programming in MATLAB, which is a skill that will be useful as I move forward towards my professional goals.

Recommendations for future research

Experimentally determine the z-distance for a 2mW laser and adjust the micro-fluidic system to have the correct distance.

References

[1] Mayo Clinic, 22nd of March 2016. Pulmonary hypertension. https://www.mayoclinic.org/diseases-conditions/

pulmonary-hypertension/symptoms-causes/syc-20350697 (retrieved: 2017-11-07)

[2] Veith, Cristine., Schermuly, Ralph T., Brandes, Ralf P. 1 March 2016.

Molecular mechanisms of hypoxia-inducible factor-induced pulmonary arterial smooth muscle cell alterations in pulmonary hypertension.

The journal of Physiology. 594 (5): 1167–1177 [3] Socialstyrelsen, 9th of December 2014.

Pulmonell arteriell hypertension och kronisk tromboembolisk pulmonell hypertension.

http://www.socialstyrelsen.se/ovanligadiagnoser/

pulmonellarteriellhypertension#anchor_2 (retrieved: 2017-11-06) [4] Encyclopaedia Brittanica. Spectroscopy.

https://www.britannica.com/science/spectroscopy#toc558901main (retrieved: 2017-11-16).

[5] Tiwari, Gaurav. February 28, 2017. Classical theory of Raman effect. https://gauravtiwari.org/2011/02/01/

raman-effect-raman-spectroscopy-raman-scattering/ (retrieved: 2017-12-03)

[6] Banwell, Colin., McCash, Elaine. 1994.

Fundamentals of molecular spectroscopy, 4th edition. Chapter 5. Berkshire: McGraw-Hill Publishing Company.

[7] Ferraro, John R., Nakamoto, Kazuo., Brown, Chris W. 2003. Introductory Raman spectroscopy. San Diego: Academic Press. [8] Chemistry and Light. Fluorescence.

http://www.chemistryandlight.eu/index.php/fluorescence/ (retrieved: 2018-01-09).

[9] Encyclopaedia Brittanica. Orbits and energy levels.

https://www.britannica.com/science/atom/Orbits-and-energy-levels (retrieved: 2017-11-14).

[10] ThermoFisher Scientific. Fluorescence Fundamentals.

molecular-probes-the-handbook/

introduction-to-fluorescence-techniques.html (retrieved: 2017-11-14). [11] Kommu, Naresh. 2014. Applications of Fluorescence Spectroscopy.

Journal of Chemical and Pharmaceutical Sciences. (5): 18-21.

https://www.jchps.com/specialissues/Special%20issue5/05%20jchps% 20si5%20k%20naresh%2018-21.pdf (retrieved: 2018-01-09).

[12] B&W Tek. Components of a Raman Spectrometer.

http://bwtek.com/raman-components-of-a-raman-spectrometer/ (retrieved: 2017-11-03).

[13] Goodfellow. Polymethylmetacrylate Material Information.

http://www.goodfellow.com/A/Polymethylmethacrylate.html (retrieved: 2018-01-09)

[14] AZO Materials, 11th of April 2003.

Polycarbonate - Supplier Data by Goodfellow.

https://www.azom.com/article.aspx?ArticleID=1934 (retrieved: 2018-01-09)

[15] Ramser, Kerstin. 2006.

Optical manipulation and spectroscopy of single functional globin-containing cells in microfluidic systems. G¨oteborg: Chalmers tekniska h¨ogskola.

[16] Amscope.com. Microscope parts and functions.

http://www.amscope.com/microscope-parts-and-functions/ (retrieved: 2017-11-06).

[17] Kenneth R. Spring, Thomas J. Fellers, Michael W. Davidson. MicroscopyU. Introduction to Charge-Coupled Devices (CCDs).

https://www.microscopyu.com/digital-imaging/

introduction-to-charge-coupled-devices-ccds (retrieved: 2018-01-09). [18] B&W Tek. Spectrometer knowledge. Part 1: the slit.

http://bwtek.com/spectrometer-part-1-the-slit/ (retrieved: 2017-12-06). [19] B&W Tek. Spectrometer knowledge. Part 2: diffraction grating.

http://bwtek.com/spectrometer-part-2-the-grating/ (retrieved: 2017-12-06).

2017-12-06).

[21] yourgenome.org. Why use yeast in research?.

https://www.yourgenome.org/facts/why-use-yeast-in-research (retrieved: 2017-11-01).

[22] Feldmann, Horst. 2012. Yeast Cell Architecture and Functions. Wiley-VCH Verlag GmbH & Co.

https://application.wiley-vch.de/books/sample/3527332529_c02.pdff (retrieved: 2017-12-06).

[23] Explore yeast. Yeast, a cellular model.

http://www.exploreyeast.com/article/yeast-cellular-model (retrieved: 2017-12-08).

[24] McCreery Research Group, 2014. Standard Spectra. http:

//www.chem.ualberta.ca/~mccreery/ramanmaterials.html#polystyrene (retrieved: 2018-01-11).

[25] Medicago AB. SMARTBUFFERS.

http://www.medicago.se/sites/default/files/pdf/productsheets/PBS_ Buffer_v._01.pdf (retrieved: 2017-11-22).

[26] Schulze, H. Georg., Turner, Robin F.B. 2014.

A two-dimensionally coincident second difference cosmic ray spike removal method for the fully automated processing of Raman spectra.

Analytical Chemistry. 68 (2): 185-191. [27] Savitzky, A., Golay, M.J.E. 1964.

Smoothing and Differentiation of Data by Simplified Least Squares Procedures. Society for Applied Spectroscopy. 36 (8): 1627–39.

[28] Alex Cao, Abhilash K. Pandya, Gulay K. Serhatkulu, Rachel E. Weber, Houbei Dai, Jagdish S. Thakur, Vaman M. Naik, Ratna Naik, Gregory W. Auner, Raja Rabah, D. Carl Freeman. 2007.

A robust method for automated background subtraction of tissue fluorescence. Journal of Raman Spectroscopy. 38: 1199–1205.

[29] Prospector. Raman Spectroscopy 101 for Plastics Identification. https://knowledge.ulprospector.com/1365/

pe-raman-spectroscopy-plastics-identification/ (retrieved: 2018-02-01). [30] Ramansystems.com. Sample Spectra.

(retrieved: 2018-02-01).

[31] Erjavec, Nika. Pinato, Giulietta. Ramser, Kerstin. 2016.

Raman spectroscopy as a tool for detecting mitochondrial fitness. Journal of Raman spectroscopy. 47: 933-939.

[32] Huang, Yu-San. Karashima, Takeshi. Yamamoto, Masayuki. Ogura, Takashi. Hamaguchi, Hiro-o. 2004.

Raman spectroscopic signature of life in a living yeast cell. Journal of Raman spectroscopy. 35: 525-526.

Appendix A: MATLAB codes for signal analysis

Function for ray spike removal

1 f u n c t i o n [ s 2 p r i m ] = RamanRaySpikeRemoval ( s , n , m ) 2 3 %Removal o f c o s m i c r a y s p i k e s . 4 5 % s : raw d a t a v a l u e s . 6 % n : d e g r e e o f p o l y n o m i a l f o r p o l y f i t . 7 % m: l e n g t h o f segment f o r p o l y f i t . 8 9 s 1=s ( : , 1 ) ; %x−v a l u e s o f raw d a t a . 10 s 2=s ( : , 2 ) ; %y−v a l u e s o f raw d a t a . 11 s 2=s 2 ( : ) ’ ; 12 13 f i g u r e 14 p l o t( s1 , s2 , ’ k ’) ; %p l o t t i n g raw d a t a . 15 x l a b e l(’Raman s h i f t (cmˆ{ −1}) ’) 16 y l a b e l(’ I n t e n s i t y ( c o u n t s ) ’) 17 18 s 2=p a d a r r a y ( s2 , [ 0 m] , ’ r e p l i c a t e ’) ; %Padding t o make s u r e no c o s m i c r a y s i k e s a t e d g e s . 19 20 FY=g r a d i e n t(g r a d i e n t( s 2 ) ) ; %s e c o n d d e r v a t i v e o f y−v a l u e s o f raw d a t a .

21 y=−s q r t( 6 ) . ∗s t d(FY) ; %e s t i m a t i n g width o f FY . 22 a=f i n d(FY<y ) ; %f i n d any d e v i a t i n g v a l u e s . 23 s 2 p r i m=s 2 ; 24 25 f i g u r e 26 p l o t( s1 , FY(m+1:end−m) , ’ k ’) %p l o t t i n g s e c o n d d e r i v a t i v e . 27 x l a b e l(’Raman s h i f t (cmˆ{ −1}) ’) 28 29 f o r p=1:l e n g t h( a ) ; 30 31 x=(a ( p )−m: a ( p )+m) ; %c r e a t e s a r r a y o f v a l u e s t o be p r o c e s s e d . 32 q=l o g i c a l (1− ismember ( x , a ) ) ; %s e e s i f t h e c r e a t e d a r r a y c o n t a i n s any d e v i a t i n g v a l u e s .

33 34 35 b=s 2 ( a ( p )−m: a ( p )+m) ; %c e n t e r i n g segment around d e v i a t i n g v a l u e ( r a y s p i k e ) . 36 [ c , d , mu]=p o l y f i t( x ( q ) , b ( q ) , n ) ; %f i t t i n g c u r v e around d e v i a t i n g v a l u e s . 37 g=p o l y v a l( c , x , [ ] , mu) ; %c a l c u l a t i n g c o o r d i n a t e s o f e s t i m a t e d c u r v e . 38 s 2 p r i m ( a ( p ) )=g (m+1) ; %r e p l a c i n g m i d d l e ( where r a y s p i k e e x i s t s ) v a l u e o f e s t i a t e d c u r v e . 39 40 end

41 s 2 p r i m=s 2 p r i m (m+1:end−m) ; %remove padding . 42 43 f i g u r e 44 p l o t( s1 , s2prim , ’ k ’) ; %P l o t t i n g f i n a l image w i t h o u t r a y s p i k e s . 45 x l a b e l(’Raman s h i f t (cmˆ{ −1}) ’) 46 y l a b e l(’ I n t e n s i t y ( c o u n t s ) ’) 47 48 end

Savitzky-Golay filter

3 % Removal o f n o i s e from raw d a t a . 4

5 %s : Raw d a t a from measurements .

6 %n : Order o f p o l y n o m i a l f o r p o l y f i t . 7 %m: Length o f segment . 8 9 s 1=s ( : , 1 ) ; %x−v a l u e s o f raw d a t a . 10 s 2=s ( : , 2 ) ; %y−v a l u e s o f raw d a t a . 11 12 f i g u r e 13 p l o t( s1 , s2 , ’ k ’) ; %p l o t t i n g raw d a t a 14 x l a b e l(’Raman s h i f t (cmˆ{ −1}) ’)

17 s 1=s 1 ( : ) ’ ; %c o n v e r t i n g t o row−v e c t o r s 18 s 2=s 2 ( : ) ’ ; 19 20 s 1=p a d a r r a y ( s1 , [ 0 m] , ’ r e p l i c a t e ’) ; %padding a r r a y t o e n s u r e a smooth s t a r t and f i n i s h i n p l o t . 21 s 2=p a d a r r a y ( s2 , [ 0 m] , ’ r e p l i c a t e ’) ; 22 23 f o r p=m+1:l e n g t h( s 1 )−m; 24 25 [ c , d , mu]=p o l y f i t( s 1 ( p−m: p+m) , s 2 ( p−m: p+m) , n ) ; %f i t t i n g c u r v e t o s m a l l segment 26 s 2 p r i m ( p−m: p+m)=p o l y v a l( c , s 1 ( p−m: p+m) , [ ] , mu) ; % c a l c u l a t i n g c o o r d i n a t e s f o r e s t i m a t e d c u r v e . 27 s 2 ( p )=s 2 p r i m ( p ) ; %r e p l a c i n g v a l u e s i n a r r a y . 28 29 end 30 31 f i g u r e 32 p l o t( s1 , s2 , ’ k ’) ; %p l o t t i n g f i n a l n o i s e f r e e image . 33 x l a b e l(’Raman s h i f t (cmˆ{ −1}) ’) 34 y l a b e l(’ I n t e n s i t y ( c o u n t s ) ’) 35 36 end

Function for Background removal

1 f u n c t i o n [ s1 , s 2 ] = RamanBgRemoval ( s , n , i ) 2

3 % F u n c t i o n t h a t e s t i m a t e s t h e background r a d i a t i o n o f Raman

measurements .

4

5 %s : Raw d a t a from measurements .

6 %n : Order o f p o l y n o m i a l f o r p o l y f i t . 7 %i : Number o f i t e r a t i o n s . 8 9 s 1=s ( : , 1 ) ; %x−v a l u e s o f measurements . 10 s 2=s ( : , 2 ) ; %y−v a l u e s o f measurements . 11 12 % p l o t ( s1 , s2 , ’ k ’ ) ; %p l o t t i n g raw d a t a . 13 % x l a b e l ( ’ Raman s h i f t (cmˆ{ −1}) ’ )

14 % y l a b e l ( ’ I n t e n s i t y ( c o u n t s ) ’ ) 15 % h o l d on 16 17 s 1=s 1 ( : ) ’ ; %c o n v e r t t o row v e c t o r s . 18 s 2=s 2 ( : ) ’ ; 19 20 f o r p=1: i ; 21 22 [ c , d , mu]=p o l y f i t( s1 , s2 , n ) ; %f i t t i n g c u r v e o f d e g r e e n t o t h e raw d a t a . 23 s p r i m=p o l y v a l( c , s1 , [ ] , mu) ; %c a l c u l a t i n g c o o r d i n a t e s o f e s t i m a t e d c u r v e . 24

25 q=f i n d( sprim>s 2 ) ; %f i n d v a l u e s where raw d a t a i s s m a l l e r

than c o o r d i n a t e s o f e s t i m a t e d c u r v e . 26 s p r i m ( q )=s 2 ( q ) ; %R e p l a c e s m a l l e r v a l u e s . 27 s 2=s p r i m ; %p r e p a r e f o r n e x t r e p e a t i n l o o p . 28 29 end 30 31 % p l o t ( s1 , s2 , ’ k−−’) ; %p l o t t i n g e s t i m a t e d background .

32 % l e g e n d ( ’ Raw d a t a from e x p e r i m e n t a l measurements ’ , ’ E s t i m a t e d

c u r v e f o r Background removal ’ )

33 34 end