Week of 01/15/2018 - Progress Report 8

Tamara Jovanovic

Topics covered: Redesign of the cuvette stand to be 3D printed, materials research

Materials used: SolidWorks Software

Upon some delivery issues that occurred during the past week, we had to purchase another rotational stage. This new option for the rotational stage is from Newport Laboratories, as it is more reliable and will be accessible to us much faster than the original choice. Thanks to this and our project advisor, Dr. Asghari, we will be able to continue on our project according to the previously established timeline.

The picture of the rotational stage is shown below.

Now that the rotational stage is delivered, the measurements for the cuvette stand designed in SolidWorks will be confirmed and the 3D printing will begin.

As for the cuvette stand, because of the change in height and base of the rotational stage, certain design changes are made. We want to be able to screw the 3D stand on the rotational stage using the screws that are provided to us in the photonics lab. This is why there are 2 holes (the design might involve 4 holes in the end, depending on the physical structure of the rotational stage. This will be assessed when the rotational stage is delivered.) at each end of the stand. According to the measurements, those holes will be right on top of the screw holes of the rotational stage and it will be screwed in for stability. The movements to the whole system will be made by the knobs attached on the rotational stage itself. The movements will only need to be made in order to maximize the angle of refraction, and since these changes will need to be made in the smallest possible increments to find the most appropriate angle, it will be very helpful that we will have the rotational stage. That way we can track the change and observe glucose behavior at the best possible angle.

The new design of the cuvette stand is shown below.

Back side of the stand.

Once the rotational stage is delivered, the 3D stand will be printed. The printer used to print this will most likely be MakerBot Replicator 2, found in the lab, pictured below. The material which will be used will most likely be nylon (Polyamide). Nylon is strong and flexible type of plastic which is exactly what we need for our cuvette stand. If used in 3D printing, it produces products that, aside from being very tough, are also really hard to damage, as an abundance of common chemicals are unable to damage it. It is also very cost efficient because it is a material which is widely used in many other industries and it costs only about $18/kg. It does require high temperature (about 250 degrees Celsius) to be used in 3D printing, but the printer we have access to is definitely equipped to operate at such high temperatures and produce good quality products.

For next week’s progress, the cuvette stand will be printed since the rotational stage will be available to us early next week. The material (nylon) for printing will be acquired and I will run some tests before the actual printing of the stand occurs. Hopefully that won’t take a long time so if these tasks are achieved, the rotational stage and newly printed 3D stand will be implemented in our optical system and we will start working on maximizing the refraction angle of the laser through an actual cuvette and getting some baseline data.

Ezequiel Partida

Topics covered: Last week follow-up, use past research to test RMSE% and correlation methods

Materials used: Matlab, Numerous Research

Last week, I tested our implemented RMSE% method to see how it applied to certain data. I came to the conclusion that RMSE% works best to find overall discrepancies in data. In contrast, correlation yields a deterministic constant for overall sameness of shape. Therefore, it seems smart to use correlation to find whether 2 datasets are similar shape, and then use RMSE% to see how different they truly are. In order to test this assumption, I looked over past research on glucose absorption and applied both methods to people’s resulting data to see if I got similar results as them.

Note: Glucose absorption seems to differ from water absorption by smooth gradual changes, so these research data will be good test cases.

Data #1: Spectral response of glucose, Prof. Dr. Ingeborg Beckers, Andor.com

First,visual inspection was done on the plot above in order to pick a few points and plot a rough sketch, since the website of the research data did not provide excel or text format of the data. After a rough sketch was plotted, interpolation was performed in order to create long enough arrays to perform analyses on. After this was done, the rough plots from my Matlab scripts is shown below.

I know the interpolated data is not exactly the same as the one from Prof. Beckers’, however on the research paper, error analysis was not performed. All that was said of the data was that there was a significant difference in water and glucose absorption through their method, in vivo monitoring, for the given wavelength range. Therefore, by using our analysis methods, that conclusion was supported. Our analysis methods on this interpolated data is shown below.

As shown above, the correlation method shows that the water and glucose spectra have the same shape (0.9989 almost 1), even if the interpolated data was not exactly the same as the original research data. Also, the RMSE% shows a 20% error inthe 2 datasets, which shows significant enough difference between water and glucose in the NIR region. This is good since our equipment has high resolution and range, so we can be confident that out 1600 nm region will show significant results.

Data #2: A Preliminary Investigation into the Design of an Implantable Optical Blood Glucose Sensor, A. Trabelsi1, M. Boukadoum , M. Siaj, American Journal of Biomedical Engineering 2011; 1(2): 62-67

The data shown below shows absorption spectra of different concentrations of glucose in NIR region (around 1500-1690 nm) through a procedure using electrically pumped vertical-cavity surface-emitting laser (VCSEL) diodes. This data shows more gradual changes with minimal change with each different concentration (around 0.0001 units per concentration).

Using visual plotting and interpolation again, the following Matlab plots for the top red and black lines are shown below.

Using our analysis methods, the following results are shown.

In the report of the research, it is stated that the R^2 correlation coefficient between each concentration was about .95. Here, our method gave 0.999, however the interpolated data is not accurate and our method is a better correlation that R^2. However, the RMSE% gives insignificant error, which could be a cause of the small units or because our interpolated data is not accurate to show the actual gradual changes. Nonetheless, this could be fixed once we gather our own glucose data with different methods. The good part is that correlation once again proves that we can distinguish between shapes of data.

Next week, the rotational stage will be available, so we will begin to calibrate our optical set up with the stage in order to align our system better and hopefully start gathering glucose data. This week, not much could have been accomplished since the stage was not here and our CAD 3D cuvette stand had already been designed and we have no real data to start analysis with. However, this should hopefully be fixed this following week.

Jonathan De Rouen

Topics Covered: Further Exploring Connection problems with new Computer and GPIB cable

Materials used: GPIB cable, Agilent IO (Key Sight connection expert), MatLab

This week I focused on implementing our GUI onto the new computer system in the optics laboratory. We were using a newer version of the Agilent IO software called Key Sight and it would not recognize the spectrometer. Trying to connect the GPIB cable I encountered this error message from the computer.

My first thought was to reinstall Keysight Connection expert to try to reinstall the package. However I still would receive the same error on the computer. I also tried to install the driver for the 82357B USB/GPIB Interface, the computer would then recognize the cable as a device but it still would not be discovered by the Keysight connection interface. Next I tried to use the same interface we have been using on a previous computer. Due to the installation disk being lost, we tried to copy the files directly from the computer and installing it onto this computer. However the application would still not run. I considered removing the entire program from the computer and reinstalling it to try and recover this lost discovery service, but I did not want to interfere with the remote connection for another groups User Interface.

Review

Topics Covered: Further Exploring Connection problems with new Computer and GPIB cable Materials used: GPIB cable, Agilent IO (Key Sight connection expert), MatLab This week I focused on implementing our GUI onto the new computer system in the optics laboratory. We were using a newer version of the Agilent IO software called Key Sight and it would not recognize the spectrometer. Trying to connect the GPIB cable I encountered this error message from the computer.

My first thought was to reinstall Keysight Connection expert to try to reinstall the package. However I still would receive the same error on the computer. I also tried to install the driver for the 82357B USB/GPIB Interface, the computer would then recognize the cable as a device but it still would not be discovered by the Keysight connection interface. Next I tried to use the same interface we have been using on a previous computer. Due to the installation disk being lost, we tried to copy the files directly from the computer and installing it onto this computer. However the application would still not run. I considered removing the entire program from the computer and reinstalling it to try and recover this lost discovery service, but I did not want to interfere with the remote connection for another groups User Interface. Next week I hope to work with Dr. Asghari and see if he may have a copy of the original installation disk for our interface connection. If not it would be to look into the ethernet connection again explored from the first semester. Since the part has arrived I also would like to assist with the data capturing as well as developing identification algorithms for the project.