In order to ascertain which patients carried the diseased SNPs, the BME 100 lab groups tested the patients using a fluorimeter and PCR reactions they created. Overall, there were 16 groups made up of roughly 5 students each. In those groups, typically two students were at the bench working on the actual experiment while the remaining members processed the images through a software called ImageJ and organized the data in an Excel spreadsheet. The students working on the experiment were in charge of pipetting the liquid onto the slides, adding the SYBR green dye, and taking the photos. The students on ImageJ divvied up the photos and then "split the channels", separating the red, blue, and green components oof the pictures. When the photographs were uploaded and split, the red and blue photos were discarded (due to the SYBR green dye being used)leaving only the green to be analyzed. From analyzing the green photos, the amount of flourescense was ascertained and recorded. From that data, averages and standard deviation were used to calculate the plausibility of the patients carrying the diseased SNP. To reduce error, the students working on the experiment carefully made sure to pipette the same amount of liquid and dye every time. Alternatively, their ImageJ colleagues analyzed three versions of the same image, with 14 original images in total. All in all, the collective data from the 10:30 BME 100 class showed relatively successful results identifying the carriers of the disease. Overall, there were 32 patients analyzed by the BME students, and only two failed to yield results/had blank results. Its safe to say the experiment was a success.
What Bayes Statistics Imply about This Diagnostic Approach
Reliability depends on sensitivity and specificity of the tests. The fact that the probabilities of both calculations are high implies the tests are highly reliable. Calculation 1 describes the sensitivity of the system to classify the person will have the disease SNP if their results for the disease were positive. Calculation 2 describes the specificity of the system’ s ability to detect that the person will not have the disease SNP because they originally tested negative. Since both of these calculation are close to 100% the reliability is high.
The tests predicted that 100% of the patients will not develop the disease given a negative final test conclusion. They detected close to 100% of the patients who test negative did not actually develop the disease. This concludes their prediction for calculation 4 was reliable. Calculation 3,however, only predicted about half of the patients who tested positive will actually develop the disease, but calculation 1 shows almost 100% of the patients who test positive actually developed the disease. Therefore, calculation 3 is not reliable.
Based on our results, we determined that in trying to accurately predict the probability of a patient developing the disease SNP, the best method forward is determining specificity, which means detecting someone who doesn’t have the disease will test negative for the disease, is more reliable than determining the sensitivity, which means someone who does have the disease will also test positive for the disease SNP.
Some possible sources of error can come from the camera that was used, the process of pipetting, and too much exposure to light.The biggest source of error came from taking images of the droplets. The angle and distance of the camera were not held constant which can lead to different areas and concentrations found on ImageJ. The focus of the camera was also extremely hard to keep consistent, and this can also lead to ImageJ calculating inaccurate values for each droplet. Along with the errors of the phone images, the exposure of the flash of the phone when the picture was taken could have affected the sybr green solution, since it is very sensitive to light. If it was exposed to too much light, there would not be as much green glow, and therefore the concentration calculation end up lower than they actually are. The pipetting could have also caused some error with the results. Not picking up a consistent amount of solution while pipetting could have caused the droplets to be different sizes, resulting in a wide range of areas in ImageJ. All these errors would result in different initial values, therefore resulting in less accurate Bayes values. The wrong bayes values can then lead false predictions and detections of the patients disease results.
Intro to Computer-Aided Design
3D Modeling
Our teamed decided to go with solid works and we had a good experience using it. Some members had already had some experience with solid works, so designing the prototype was easy for them and made it easy for them to help other members as needed. It was also useful to watch YouTube video on how to use solid works since it is a well know software. Since solid works as an assembly application, we were able to divide the design up into different parts so one person did not have to design the whole fluorimeter on their computer. We decided one person would create the phone holder, one person creates the box, and one person create the fluorimeter. Once everyone did their part, we were able to download all our parts to one computer and use the assemble feature on solid works to put all our pieces together to complete the design.
Our Design
Our design is made up of 3 main parts: the phone holder, the box with an adjustable stand, and the slide holder with a fluorescent light. Our phone holder is an improvement from the original phone holder because it is not only compatible for any type of phone, but it also keeps the phone perfectly still at a constant, controlled distance. This saves the user time by allowing them to not worry about the angle and distance the phone is at during trial.This phone holder will be attached to the outside of the box. This box has new design components that will also improves the fluorimeter process. In side the box there is an adjustable stand that secures the side holder on the top of the platform. We designed this stand so that it is accessible to all kinds of phones, according to the height on where the camera is located. Since the phone holder is keeping the phone still and accurate, the adjustable legs inside the box will be the piece that is moving, adjusting to the phone camera. We chose to add this new component to the fluorimeter process so that each picture will be taken at the same distance, with the same accuracy, and at the same height.This will then allow for a more accurate reading when the photos are processed through ImageJ.
Feature 1: Consumables
The consumables included in the kit
SnapFluo phone holder
SnapFluo box
SnapFluo fluorescent light and slide holder
SYBR Green solution
3 glass slides
The SYBR Green Solution we provide is the best on the market. Any costumer's camera will be able to focus on the droplet because the green solution works best with our product. The 3 glass slides we provide fit perfectly in our design and have extra grip strength to prevent the droplet from losing its form when the platform is being moved.
Feature 2: Hardware - PCR Machine & Fluorimeter
The PCR machine is a very complex and well thought out design, so there was no need to make any changes. It got the job done in a timely manner without any complications. The fluorimeter on the other hand, caused the most problems due to the overall, blatant flaws.
During our experience with the fluorimeter we faced many issues that had a detrimental effect on our results. We were not able to have the camera consistently focused on the droplet during each trail. This lead to the issue of changing the distance our phone was placed at, in order to get a focused picture. Another issue that came up was figuring out which side of the slide to use for the droplet. So our design fixes all the issues we encountered. We came up with a new phone holder that keeps the phone still, prevents light exposure, and focuses on the droplet. Our adjustable stand saves time by making the angle of the image perfect.
BME 100 Fall 2017
