Week of 02/19/2018 - Progress Report 11

Tamara Jovanovic

Topics covered: Acquiring glucose data

Materials used: Optical Setup of the Non-Invasive system, glucose, glucose solutions, lab equipment, Menlo Systems C-fiber Laser, Yokogawa Spectrometer

This week’s progress involved getting glucose data and doing analysis. Because of the sensitivity of the system, we discussed various ways to improve it. From getting more optical alignment, to stabilizing the rotational stage, the only thing we didn’t try when it came to stabilization is getting glucose data with the cuvette never leaving the stand. This was done by using the pipette to put the solution in the cuvettes, and to get it out. The process is depicted in the picture below.

This way of taking data ensured that it would be stable. Because of this new way of taking data, we decided to do a little calibration and make a reference for ourselves for future testing.The spectrometer can only do 7 trials at a time in one spreadsheet, so this maximum of trials was used to do calibration. In order to get the most accurate results, the following things were tested:

• 7 trials of distilled water in the Quartz cuvette

• 7 trials of distilled water in the Glass cuvette

• 7 trials of a 100 mg/dL glucose solution in the Quartz cuvette

• 7 trials of a 100 mg/dL glucose solution in the Glass cuvette

First, a new glucose solution was prepared. Last week, we had our midterm presentation for the faculty of Seaver College of Science and Engineering and we got a some really good ideas for future steps in this project. One of those ideas was to test one glucose concentration, which is closer to what is realistically to be found in a person’s blood with diabetes. Because of that, we decided that 100 mg of glucose is the perfect concentration for our research. So, 100 mg of glucose was dissolved in 100 dL of distilled water, and the testing started.

Beaker with the new solution.

The 7 trials are done so that we can confirm that the results we are getting are most accurate and that we can compare it to already existing research. Due to some alignment issues during the course of the past week, I was only working with approximately 0.7 mW of power.

This first picture below depicts 7 trials of distilled water in the Quartz cuvette. As it can be seen, all the trials are pretty much the same, except the B trace, which was the second trial. But, this is why we are doing this and why we want this kind of data. After this data was taken, the next step will be to analyze it in the Matlab. Ezequiel will take the average of all 7 trials and we should have the plot for distilled water in a Quartz cuvette for all future testings. Also, since the Matlab program is quite sophisticated and precise, we will be able to see exactly how large the ripples on the plot are. In the future, we hope to have such good alignment that we don’t have ripples or that we have them really small, which will allow us to analyze data at the top level and with most accuracy and precision.

Distilled water in the Quartz cuvette.

The data collected from these 7 trials looks something like this:

The trace data for each trace and for each step is recorded. Since there is a setting on the spectrometer how many sampling points you want for your plot, I selected 500. This would give us enough data, but not too much, to process it thoroughly and get conclusive results.

The next thing tested was distilled water in a Glass cuvette. The picture below shows the spectra of distilled water in a Glass cuvette. All of this data was saved also with 500 sampling points and plugged into a Matlab program, which Ezequiel will talk about later.

One good thing to note for this spectral plot is that we are able to retain the shape of the spectra for free space, distilled water in a Quartz cuvette and distilled water in a Glass cuvette. Given the ripples on this spectra as well, I can’t say that the alignment is perfect. However, work can be done, and by doing these tests, we are making steps in the right direction.

Distilled water in Glass cuvette.

Next thing tested was the 100 mg/dL glucose solution in the Quartz cuvette, 7 trials as well. The reason we are doing all of this testing in both the Quartz and Glass cuvette is because, aside from calibration and getting reference of the system, we are trying to conclude which material of cuvette is best for our research. According to Dr. Li, the Glass cuvette would be the best, because there is least power attenuation through it. However, because this is a research based project, we have to test everything and get some results and discuss it before we make any big decisions and conclusions.

All of that being said, below is the picture of the absorption spectra of a 100 mg/dL glucose solution in a Quartz cuvette.

100 mg/dL glucose solution in Quartz cuvette.

By looking at this absorption spectra, I can conclude it is pretty consistent. I am aware that Matlab will show major ripples, but this is good data for the progress report and calibration.

Last but not least, here’s the spectra for the 100 mg/dL glucose solution. The spectra itself looks a little weird at the beginning, but the overall idea of the shape is kept. That is very good.

100 mg/dL glucose solution in Glass cuvette.

This spectra shows some more noise at the beginning and the end of the plot. There are also more discrepancies at the very beginning and then the data becomes more consistent in the middle and at the end of the spectra. This was also done 7 times. The average will be taken for reference.

Next week’s plan:

• Keep taking glucose data.

• Get more reference. Maybe using multiple glucose solutions?

• Change glucose concentrations for it to match human ones and take data.

• Take trials of DI water and glucose solutions in one plot.

• Work more on stability to avoid ripples as much possible.

Ezequiel Partida

Topics covered: Tamara’s Data to Matlab and Cuvette Observations

Materials used: Optical Setup of the Non-Invasive system, Matlab

This week, I utilized the data that Tamara acquired with a similar Matlab program as last week. This week, Tamara got 7 trials of each type of data, Spectra of Quartz Cuvette with water, Quartz Cuvette with 100 g/dL of Glucose, Glass Cuvette with water, and Glass Cuvette with 100 g/dL of Glucose. The purpose of this data was to find a more stabilized and accurate depiction of an absorption spectrum of glucose at 100 mg/dL for either quartz or glass cuvette (whichever yields the more power retention), and to compare our results to any research in our optical range. Additionally, I was to observe any power loss between the 2 cuvettes used.

First, I received 4 .CSV files from Tamara, each containing 7 trials of their corresponding data. With Matlab, I parsed out the individual data and plotted them in separate figures, as shown below.

Then, I got the averages of all the 7 trials from each of the data and plotted those in the same plot in order to view the differences of power. The results are as follows.

On first view, it can be seen that the 2 spectra with the quartz cuvette are significantly better than the glass cuvette spectra with regards to power. By zooming in to the approximate peaks, it can be seen that the quartz cuvette water spectrum peaks at about -60 dBm, while the glass cuvette water spectrum peaks at about -70 dBm. When looking at the 100 mg/dL glucose spectra, the dBm’s drop from about -60 dBm to -75 dBm.

From this observation alone, it is notable that the glass cuvette absorbs more of the input light source than the quartz cuvette. Additionally, this week, there were 7 total trials, which were all notably consistent. Therefore, it was concluded by our work this week that quartz cuvettes yield a higher output power. However, when looking at the information of the spectra, it is obvious that there are very little discrepancies with the quartz cuvette spectra. That is, for glucose solution and normal DI water, the differences are basically not there. Even on the zoomed in plot, the only differences in the spectra are in phase. However, for the glass cuvette, there is a significant drop in power for the spectra of glucose solution. There is more Power Output for the quartz cuvettes. However, there is more spectral discrepancies for the glass cuvettes.

In order to view the absorption spectrum for this solution for the 2 cuvette spectra, I performed a subtraction and plotted them. Then, I gave the absorption spectra to Jonathan to compare to other research.

Jonathan De Rouen

Topics Covered: Glucose Spectrum Analysis & Research

Materials Used: Research

This week I had taken the liberty to do research on new methods of introducing the solution into our optical system. I looked for any type of specialized device or method but could only find the method of using a pipette to introduce our liquid solutions. However I do believe a simple water pump device can be used with our cuvettes to be implemented in order to introduce our device. However since our main focus was to analyze the absorption spectra I continued to do research on the spectrum we are expecting to see at the wavelength of light. Research from the Department of physics in Sungkyunkwan University can give us some idea of the spectrum we may see at larger scale of concentrations than the ones we are doing analysis with. However this should yield only a concentration loss. However due to their data we can now compare our data to their analysis of the glucose spectrum, from them subtracting the water spectrum from their glucose spectrum.

With figure in the smallest frame being the pure absorption spectra. There original glucose in aqueous solution is shown below to support the idea the as the concentrations increase there will be a minor loss in the absorption of the spectrum.

Other sources have there own spectrum of glucose in aqueous solutions as shown below at 100 mg/ dL.

From: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3675860/

However from the prior graph we can compare it to our data from the trying to acquire pure glucose absorption data.

So far this data is still inconclusive and does not correlate with what our research has shown so it can only be said to be inconclusive as of now.

For next week:

1. Get accurate data;

2. request to meet with Dr. Li and research the name of a pumping system for improving calibration

3. Design a simple liquid pump system that can be used

4. Get all the Glucose members on a GUI so it can make gathering data faster.