Annie's Summer Blog
My links:
15 July
Today we started with our biweekly tactile meeting. I showed the group some of the silicone samples I have made to attach to the DIGIT sensor. The general recommendation is to do some research on the different fabrication methods of the different sensors and why this is (for example why does DIGIT not use a reflective coating while GelSight does). I also shared with the group the adventures I have had getting the ATI to work on Linux, as it seemed the issue was the medical devices computer being out of date to install the right packages for the national instruments drivers. So, I tried to install the drivers again on the Linux laptop (after getting the wifi issue solved again through Mark's help of finding a wifi USB thing). Unfortunately I ran into the same issues and the download of the specific package I need: ni-daqmx would not download. For the near future, I believe I will keep my current set up of using the Linux laptop to collect DIGIT data and using the medical devices computer to collect ATI data and communicate between the two over wifi.
My other task for today was to figure out an optimal solution for coating the transparent silicone gels with a reflective coating and/or white opaque ecoflex. I explored a couple methods today:
- Coating with just ecoflex: this ends up as translucent in thin layers like the DIGIT sensor, however, I feel like I did not get as crisp of an image when compared to the preliminary images I got on Friday with the metallic powder coat
- Reflective coatingt: Right now I am using the Rustoleum aluminum finish paint, but my issue is with the paint crackling on the surface of the gel allowing light to come through in random patterns. I tried to cover it with ecoflex immediately to not have it crackle and unfortunately that did not work. My next idea is to mix the spray paint into the ecoflex itself kind of like what I did with the metallic powder and hopefully it will be more flexible and not crackle that way.
I also had a chance to speak with Danielle Panelli, a MFM specialist at Stanford. She feels the most valuable contribution to my project would be to differentiate between patients who deliver preterm who display symptoms of dilation and short cervices and who do not deliver preterm with these same symptoms. Overall, she thinks there is merit to a project like mine and knowing more about preterm labor and its progression is certainly important.
Update: mixing the spray paint with the ecoflex was not a good idea, it got clumpy and weird... oops
Update: the spray paint and metallic coatings I have right now are just too sparkly, so I think I need to either wait for the mirror finish one or possibly borrow a can from Monroe Kennedy's lab who also use it for DenseTact
12 July
Today I started off by doing some testing on coating my solaris gels with a reflective layer. We do not have any of the spray paint I wanted to use in lab, so I took an adventure to the loft and found some Rustoleum Metallic Aluminum paint. Julia recommended use the Rustoleum Mirror Finish paint, and while I am waiting for my purchase approval to go through, I wanted to try this. Julia also gifted me some Meyspring Silver Lining aluminum metallic powder that I mixed with some Ecoflex to try as well.
I was only able to try the sample of metallic ecoflex which was "too metallic" when placed on the DIGIT. It sparkled when hit with the LEDs which obstructed view.
I also had an issue when running the Linux laptop today as it would not connect to wifi. After spending some time with Seonghong, we were able to figure out that "secure boot" was enabled, and when we disabled it the wifi status magically fixed itself.
On Monday I will do some more experimentation with the final assembly of my new and improved DIGIT gels and hopefully start my experimentation.
11 July
Today I started by calling the NSF to get my fellowship sorted out, which of course was not frustrating at all! I guess stay tuned for updates here.
I also got some good-looking molds from my resurfaced DIGIT daughter molds, and I am attempting to make some good samples with these. I made three molds to speed up production of the gels and am going to try a few different things with them. One of them I will cast as normal, one I will put the acrylic plate on the back while it is still curing to try and bond it, and the last one I am going to cure without applying heat to see if that improves the clarity.
On the software side of things I tried and failed to install a National Instruments driver on the medical devices computer, and seemingly the issue is that our version of Ubuntu is out of date. I think the easiest thing to do moving forward is to actually use the medical devices computer to run MATLAB and send commands to the Arduino via serial and connect separately to the laptop running Linux with Python to sen commands to the DIGIT sensor to take frames at specific intervals.
Update: I was finally able to et in touch with the NSF people who sent me a rather snarky email back stating that I will have another opportunity to update my status in August, so thanks I guess.
10 July
Today, I spent the first part of my day organizing all my thoughts for this project and also getting the "story" right for what the timeline of research has been and development of methods for PTL prediction and prevention. In between this, I was also attempting to again cast silicone into the DIGIT-shape using solaris and the original DIGIT gel as a daughter mold. I did have more success with this today. First, I covered the original DIGIT gel in a thin layer of Mold Star to smooth out the surface (it got cracked and scratched from repeated handling). Then, I sprayed this surface with a generous coat of mold release and stuck it face-down in a pot of degassed mold star. The mold came out pretty good and I made some preliminary samples with solaris in this mold. In my first attempt, the top part of the gel got stuck in the mold although I applied mold release, so on the second attempt I applied a "very" generous coat of mold release which allowed the gel to slip out, however, it has a surface texture from the excess drops of mold release. I will try again tomorrow to find the right balance and to also make more molds to speed up the process.
In terms of my clerical tasks, I need to call the NSF GRFP people and see if they can help me after a month of ghosting. I meant to do this today but got distracted, so tomorrow morning it is.
9 July
I got back to Stanford this morning from my extended 4th of July weekend in LA with my family. I played a lot of golf including the US Women's Amateur Qualifier, where I shot 73 (+1) which did not make the cut, but nonetheless I am proud of myself for keeping up my skill and putting myself out there. Onto new adventures! Today I am focused on fixing my DIGIT gel molds so they will have smooth surfaces. Unfortunately my molds last week had air bubbles and I did ruin one by forgetting to put in mold release, oops! My current plan is to glue the corner of the original DIGIT gel back together using silicone since it ripped. Then I will spray it with mold release (for real this time) and try to press it into my existing mold with a little bit of silicone to create a fine layer to cover the air bubbles. I will also try to make a new mold entirely.
I also did some preliminary research on IVF embryo implantation mostly because I was just curious. It turns out that implanting the embryo in the precise location is a little bit of guess work on abdominal ultrasound guidance.
Lastly, I attempted to replace the batteries in all the multimeters but was only successful with two and I did my financial training so I can make purchases.
Update: "gluing" the gel back together with silicone did not work Update: new mold did not form properly (there is a piece missing in the original gel that the silicone filled in) Update: resurfacing of air bubble mold kind of didnt work either. The silicone just stuck to the original sample rather than the mold, so I tried again but instead removing the original gel from the mold this time after pressing it in to get the excess silicone out. It sort of worked. My silicone sample actually turned out pretty good except one of the top corners ripped off and took the protective covering on the mold with it. I am going to try this same method with the solaris instead of the mold star, since I have found that solaris sticks to the mold star pretty well.
My next idea is to resurface a figit gel and use that as my new daughter mold. The idea here is to rely on gravity to smooth out the surface
2 July
This morning I spoke to Yasser El-Sayed, the head of the division of MFM at Stanford. The summary of the conversation is linked in interview notes. The TL;DR is he doesn't think tat doing my thesis in preterm labor prevention has very much clinical impact or translation and encouraged me to explore other avenues. Then, this afternoon I spoke to Yair Blumenfeld who reminded me I am in this for an academic endeavor and what I can learn from this is valuable regardless of the clinical outcome. He seems to believe there is value to a project like this and encouraged me to talk to more OB/GYN faculty to find other applications for a device of this nature. He also encouraged me to reframe the needs statement of my project to: alternative sensing for prediction of preterm labor in cases of long cervices at 20 weeks. Anyway, I will continue to do more learning and talking to people as well as continue to develop my technical skills. Today I also attempted to make some DIGIT-compatible gels with Solaris using the DIGIT gel to make a daughter mold. This afternoon I leave for LA to go home for the 4th of July and my golf tournament.
1 July
Today I was able to control y stepper motor in Arduino through MATLAB over serial. My next step is getting the ATI mini to run on linux so I can pair it with the DIGIT sensor. I tried to run the DIGIT on windows with no luck. I need to find out what to download to the PC so I can use the National Instruments driver with the ATI. I also started helping some of the summer students get more familiar with Arduino through Tinker CAD.
26 June
It looks like I forgot to update my blog on yesterday. On Tuesday the 25th we did the second installment of the lab cleanup which took most of the day. I also 3D printed some parts to go on my camera and ATI Mini for my experimental setup. I have to do the purchasing training in order to buy anything else for my project (and I need some LED lights too). Today I had Jazmin make an Arduino code to run a stepper motor for a particular distance and for a particular speed. Mark and Hao helped her with this task as well as there was a power supply issue as well as a grounding issue in the circuit. Overall, we got it worked out and the motor driving correctly. I also experimented with some filtering to help out the ATI force readings, and using a Butterworth low pass filter seems like the best option so far. My next steps are to complete the tactile sensor on the hardware side and get it properly attached to the ATI and stepper motor and to work on time syncing for the various parts of the test setup (stepper motor, ATI data, video data).
24 June
Basically no research got done today with the lab cleanup taking place. But we completely reorganized the medical devices corner, yay!
21 June
Today we started off with an impromptu meeting to discuss projects for the SURFs and SURIs for the summer. I also had a meeting with Lauren Shluzas who was on Materna Medical. SHe gave me some great advice on people to reach out to who might be interested in my project and said that there are probably funding opportunities within Stanford Biodesign for my project. Later in the summer I will reach back out to her for more advice. Then, I went ahead and continued making parts and printing for my test set up with the ATI Mini. My gelatin from yesterday did set, however, I forgot to grease the mold so it turned into goo when I tried to take it out. I think I need to not only grease the mold but make a flexible version of my cervix phantom mold for easier release of these softer hydrogels. I also talked with Julia about DIGIT Pinki for a little bit and I am going to try to make my own macroscale version to play around with then get to the smaller version later.
20 June
Today I made a 3D printed attachment for the ATI mini to sit on the one axis stage so we can do some controlled force and displacement testing on a phantom. There is also something up with the Bambu printer because my first try failed printing and my second and third tries had rough finishes under the supports. The CAD model of my part is here and the CAD file for the ATI Mini 45 is here. For my phantom, I whipped up some 10 wt% gelatin and put it in a cervix mold I made last quarter. However, I think I messed it up and I am not confident it will gel but we will see after it sits overnight. If I am going to keep making these hydrogel phantoms, I will either need to do it in another lab or acquire some stir bars, a celcius hot plate, weigh boats, and a mg scale.
19 June
Today I spoke to Luis Garcia about CERVA. The notes of this interview can be found in Interview notes. Overall, he seemed enthusiastic about a project in this space and there is a definite need for more devices that can give us better prediction of preterm labor. One interesting thing that I did not know before speaking to him is that there's essentially no treatment to stop preterm labor once it starts (besides controversial progesterone). He also seemed adamant that there is a need to update the cerclage, and I agree with him on that. I had my one on one meeting with Mark today also and we discussed the cerclage idea and essentially we came to the conclusion that Ali's cuff sensor with a biocompatible adhesive would probably do the trick. I also got the ATI Mini up and running on the sensor bench computer (yay!) and was able to find the correct calibration constants. My files for this are under "Annie" on that computer.
18 June
After getting back from the dentist, I spent the first part of my day participating in the lab clean up. I then went to refine my interview questions for physicians considering I have my first scheduled meeting tomorrow. I then went on to beef up my literature review. I focused on finding any and all new tech related to this problem and aside from a slew of biomarker kits out there (usually used to measure fetal fibronectin or placental alpha microglobulin-1) the two main techniques out there today to measure biomechanical properties of the cervix are elastography and aspiration. I found that there is a product out there from Samsung Korea called E-Cervix (ElastoScan for Cervix). Furthermore, there is Pregnolia on the aspiration side. I would like to do a risk and benefits comparison of each of these techniques and dive a little more into how they fundamentally work so I can find the gaps in this space.
17 June
I spent most of the day organizing a list of OB/GYN (and specifically MFM) specialists at Stanford because I have some outstanding questions I need to ask more people based on the brainstorming conversation last week. The list of organized faculty and who I reached out to can be found here. I also created a broad and unrefined list of questions I want to ask the people I reached out to here. Lastly, I attempted to hook up and run the ATI Mini with little luck today, but will try again tomorrow.
14 June
Coming off of the brainstorm session yesterday I am deep in the ambiguity of this project, so I decided to list out what the ideal characteristics of a solution to this problem would be. The greater scientific community thinks there is value in determining the mechanical properties of the cervix throughout pregnancy. However, one of the biggest concerns I have seen is that people are incredibly weary of anything actually touching the cervix! It is impossible to get mechanical properties without manipulating the cervix in some capacity, but I wonder if we could be more clever about this. "Cervical ripening" is the first or one of the earliest stages of labor. There is evidence that suggest the cervix ripens before it shortens, effaces, and dilates. But, what is the underlying mechanism for this ripening? Ultimately, collagen architecture breaks down allowing the tissue to be softer and stretchier. There is also evidence of increased water content in the cervix that helps disrupt that architecture. While the actual pathophysiology of this is unknown (at least based on my limited search) these are real things within the tissue that we could see.
I wanted to see if we could actually image collagen in a handheld way in vivo and didn't find anything in the literature. It seems that the best microscopy technique for imaging collagen is second harmonic generation (SHG). Collagen fibers are quite small (nm scale) so a setup like this would probably be really challenging and expensive.
Next, I wanted to see if we could image water content, and it turns out we can through Near Infared techniques. Water absorbs light at around 1000nm, and people have looked into in vivo imaging of water content in biological tissues. Basically, people rely on photoacoustic imaging where the water absorbs light, gets discreetly heated, then thermal expansion causes an acoustic response which can be read by an ultrasound transducer (or possibly an optical fiber with FBG to make it ultra small??). Yan et al showed this in excised tissue samples.
Anyway, I think there might be something here but I need some help flushing it out and understanding the solution space better.
13 June
It is officially summer, woohoo! With my ME310 spring report officially in the books I can finally do research again. Today I am focusing my efforts on a literature review. I started with the question, "What is the state of the art technology for determining prediction of preterm birth?" but ultimately decided to pivot to a more fundamental level. I am getting a better picture of the causes and prevalence of preterm labor and working my way up to tech.
In the afternoon, we did a brainstorming session with Ali, Julia, EmJ and Mark. We talked a lot about the problem and the general conclusion is that we need to learn more. The point of this conversation was to see if DIGIT Pinki can actually provide a solution to the problem at hand or if it is actually more of a "tech push" onto this problem. Ultimately, we have to think a little bit more about this. One interesting thing that I want to look into more is the use of optics to infer mechanical properties of the cervix.