Jazzmine's Project Blog
Here, Jazzmine will provide weekly updates on her research progress, detailing her specific contributions to the research.
09/14/2016 As we started to settle into our new routines for the fall semester, I met with Professor Stubbs to go over our plan for the next steps in initiating our project. We are waiting on a few odds and ends to get sorted before purchasing our equipment – the Emotive Insight Headset, Emotive Pure - EEG software, and the Raspberry Pi 3 kit – which we anticipate ordering in the next few days. I have also started working on the submission forms for the Institutional Review Board (IRB) proposal so that we can ask volunteer subjects to play the game we create. I anticipate submitting our proposal next week. Along with the IRB proposal, Professor Stubbs has offered to help us draft a release form that our subjects can read and sign before participating in the game for our research. I will also use Kali Linux from a laptop from which I was conducting security testing for our school's network in an independent study this past year. Jazzmine and I will discuss the details and scope of the game she will create, and we will confirm it with Professor Stubbs at our next meeting.
This week we are aiming to get some of the basics for the project set up. Amal has sent in a request for a linux laptop with Kali installed to being the security testing. Jazzmine has downloaded Visual Studios and started on a skeleton for the game. Together, we are working on IRB proposal and gathering materials needed. We also brainstormed a few potential ways to go about the project. First, intercepting the EEG data from the BCI and sending different data to the game. And second, sending data directly to the game tricking it into believing it is another user
Our goals for next week are to figure out a way to access the data from the BCI with Kali. As well as completing the basic gane. Finally, we hope to look over past IRB propsoals in order to complete the proposal for this project.
09/28/2016 This week, we have made more progress in assembling all the pieces of our project together. Amal now has the laptop running Kali Linux, and has started doing research into wireless attack tools in preparation for the exploitation testing of the BCI. She has found a number of bluetooth-specific attack tools on Kali that she will continue exploring.
Professor Stubbs ordered our equipment - the EEG raw software, BCI headset, and the Raspberry Pi 3; the equipment should arrive if not at the end of this week, then early next week, and we will begin experimenting with as soon as they arrive! The software needs to be installed on one of our lab computers - we have the activation code. Professor Stubbs will look into a "research room" to see if it is available for our use, and we will also submit a ticket to our Technology Department for permission to install Visual Studios and the EEG software on the lab computer.
We have also finished drafting our IRB proposal and consent form for our future participants. We will take the next few days to just polish it before finally submitting it. Jazzmine has been working on the details of the "game" that we will have our participants play; it will be more akin to simply watching a series of images flash across a computer screen and capturing the subjects' brainwaves as they view the images. Jazzmine and Amal have decided on meeting 1:1 on Thursday evenings to do our own additional planning and research.
10/05/2016 Good news: our equipment arrived! We now have the BCI headset and the Raspberry Pi 3. Jazzmine and Amal will meet this Thursday evening to play with the BCI headset, and review and analyze the game for our participants that Jazzmine has set up. We are currently sorting out available time slots for the media lab in which we will conduct most of our research. We are anticipating using it at least two evenings every week. The ticket we submitted to Technology for permission to install the EEG software and Visual Studios is still in progress.
Lastly, we just need to obtain signatures to the appropriate documentation for our IRB proposal, and then we will hand it in to the Human Protections Administrator for approval. Jazzmine will also send Professor Stubbs and Amal links to tutorial videos on using the EEG software.
This week I focused getting started with the Emotiv Insight. I began by downloading a few of the Emotiv applications. Most were visualizing the input from the channels and the others were trainging games to get the user accustomed to using the headset. Initially, Amal and I set aside time to actually test out the apps but the headset is unresponsive. There are a few tips on the Emtoiv community boards that may help alleviate the problem.
While waiting for the Emotiv Insight to begin working properly, I have started EEG Lab tutorials and downloaded it to my personal machine. Amal is working on getting approval to install it on the lab computer. Also, I have been researching other BCI projects to see what aspects of their projects could benefit this one.
Following feedback from the IRB proposal, Amal and I started a draft of edits to make. Because I attended the Grace Hopper Conference, Amal did most of the lab work while I enjoyed the conference. During the travel to and from Houston, I was able to work on some EEG analysis tutorials as well as edit some of my code for the headset.
This week we focused on working with the BCI headset. The headset is not as user friendly as anticipated. It would not turn on without being charged for 24 hours. Also, the headset reccomends having a three finger distance between the hairline and the first EEG node. Unfortunetly, both Amal and I have small forheads so it was difficult to get the headset positioned properly.
After posistioning the headset correctly, we tried to connect the headset to the Emotiv EEG software. After an hour of trying, we decided to try to connect to the Emotiv iOS app on our Iphones. The app allowed us to accurately view the Emotiv at work. It was intreresting to see that when we were talking about graduation and becomeing adults the app recorded Amal's stress level rising.
This week we again focused on our IRB proposal and getting the Emotiv full integrated. Amal is taking the lead with the IRB. During our meetings we both braingstormed ideas on how to more accurately convey the eligibility requirements. Also, editing the participation email to be less technical.
With the training from last week, we were able to quickly set up the headset and connect it to my phone. The problems with only using the iphone app is it visualizes the data but doesn't analyze it and it drains my battery extremely quickly. Amal and I decided to try to connected it to any devices we could since the lab computer still wasn't connecting. Eventually we were able to connect the headset to my macbook. We tested out a few "fun" apps from the Emotiv center. We still haven't figured out why the lab Mac wouldn't connect but for now we will be using my macbook.
Amal and I spent most of this week working on the IRB proposal. We made final changes to the meat of the proposal as well as adding a recruiment email and flyer. The flyer included a photo from the emotiv website and the email included the photo and an emotiv produced video on proper use of the headset.
This week was a bit frustrating. We finally got the emotiv EEG working and connecting to the headset on my laptop only to find out it couldn't run the code I had written in Eclipse. After reseraching, the problem was in versions of Eclispe and Java and potentially some OS missmatch as I wrritten the code on a windows machine. Unfortunetly, the BCi will not connect to my PC. I tried several times to download the community code and hand type my changes but it was not successful.
Edit: The Emotiv uses smart BTLE bluetooth technology which is optional and many older model computers do not have it included. The iMac in the lab is older and couldn't connect but my 2015 macbook could easily connect. The other problem was I had unchecked all the channels which made it seem like the BCI wasn't connecting. Once I checked all the boxes the BCI connected instantly.
The code I was trying to run was written on a PC machine using a different version of Eclipse. In order to inlcude the Emotiv packages, the lab computer must have at leave Java 1.7. I do not have the authority to update Java but will send in a ticket to technology.
It's Thanksgiving so we didn't get much done but on the brightside I was able to run my code on my macbook and connect the BCI and our IRB proposal was accepted.
This week I was able to capture some data from myself and another student. Now, I am working on filtering the data and processing it. So far, it seems like a high pass butterworth filter would be the best choice. Also, I am taking inspiration from published papers on how to better adapt the GUI for audiances not familar with BCIs.
Edit: The software uses a smaller GUI window that does not resize with different screen sizes. The majority of my testing has been done on a large iMac so a smaller window was not a problem. When I tested with other people I used my 11 inch Macbook and it was extremly difficult to see. In the training sessions I hadn't labeled the meaning of the directions because I knew that focus was tied to the X axis and medication to the Y but for a new user the training session was not helpful without the labels.
For the last meeting before winter break, Amal and I are making some changes to our project and creating expectations for break. Also, I am editing my past blogs to include more technical details.
This week, I am contemplating recoding into c++/c. The main reason being there is far more information available online compared to java. Right now I am having an issue where the BCI is varying the amount of readings per minute. This lead to inaccuracies where I edited the algorithm to "get" the correct data before realizing that the readings were off.
After researching possible solutions, the forums offered a suggestion in C# to add multiple threads to avoid the long seconds that are causing the increased readings per minute. There is also a bare bones reader on github in c++ that I am thinking about cloning and altering to fit the project. In the details page, I can already see that the authors have already fixed the increased readings per minute issue.
12/8/2016 - 1/20/2017
Amal and I will update next week with our progress from break.
The first week back was a bit stressful. Planning out weekly meeting became a problem. Amal and I both had unintetional changes in our schedules that made it difficult to nail down a time that works for us and Prof Stubbs. Without a meeting we worked independetly.
I created a tear off flyer advertising participants as well as a self scheduler allowing each participant to see a project description, answer some eligibility questions, view the consent form and if they would like schedule a text time.
Given that my birthday was this week, I would say I was less productive than expected. After meeting with Amal, I edited the flyer to match the same wording as Amal's recruitment email. And, I have added a script for when the experiments begin. With about two weeks until testing begins, I believe I will have to spend extra time in the lab to ensure that I have worked out all the errors.
During our group meeting we discussed applying for a few conferences. Amal would like to apply to a conferene in Maine while I would prefer to continue working on the project more. But, we have both agreed to apply for the undergraduate conference at Simmons which has a deadline of March 31.
I have been finding it difficult to identify and solve errors and there are only a few dicussion boards available. My goal is to work more collaboratively with Amal and see if she can better spot the errors. Also, given the success of the recruiment email we will not be using the flyer. Our goal was to have 15 people sign up and in less than a week we already have 11. The goal in the short term is to do a dry run the first week of February with Prof Stubbs and Veilluex to work out any problems before the real tests on the second and third weeks of February.
For this week, we are planning to get everything ready to test next week. The program I wrote is still a bit buggy. I've contacted a graduate student I found online who has experience in BCIs and specifically Emotiv brand. Many of the problems were due to me hard coding functions that Emotiv already created. Once I substituted my functions for Emotiv ones the code was working much smoother.
In preperation for testing Amal and I created a Prepp account with the psychology department allowing us to officially register a study. I also created a You Can Book Me site for participants to do the screening questions and scedule a time to come in and it will automatically sync to my google calender.
The goal for next week is to have some pre-testing with proffesor stubbs and actual testing with participants. Right now we have about 14 people signed up to complete the study.
Last week, we asked for additional funding because the program was taking in 9 readings per person and we have 14 people signed up with only 50 readings allowed per month. I am working on getting it down to one reading total per person to avoid upgrading our current subscription. With a snow day today, it is delaying our progress a bit as Amal and I are unable to get to the lab where we work on Thursdays.
I have been able to successfully complete the program with a GUI instead of a console output which was used before. The major problem being the BCI will not connect to my PC where I programmed it. Amal suggested upgrading my computers but it didn't help. Eventually, we decided to download Visual Studios on my Mac and transfer the code there. Unfortunately Emotiv's C# SDK is only set up for PC since Visual studios on Mac is a recent update.
The goal for next week is to either translate the code to java and use the Mac for testing or get the BCI to connect to the PC. Also, Amal and I need to come up with a back up plan for participants signed up for Monday since it looks like we will be getting a foot of snow.
As expected, we did have a snow day on Monday. This caused a major delay as we planned to meet with our participants for the first day of the study. Fortuntly, it allowed Amal and I more time to work on the program. I had a difficult time translating all my code to Java and instead am using both computers to complete the study. One will take in the data from the BCI and the other will take in data from the computer and process it.
I spent an entire day pre-testing on students in the multicultural lounge. From this I was able to make some minor cosmetic changes like making the text larger and disabling visibiltiy on useless buttons to avoid confusion. The time per trial was changed from 30 secounds to 10 seconds 3 times to avoid boredom. The script was changed to emphasize that participants should remain still at all times as many people had a tendency to bob their heads while singing or move too much during deep breathing. Also, we will be pre-defining the actions. During one of the pre-tests an international student didn't understand what "fetch" was and I had to stop the test explain it and restart. Lastly, no one liked the game I made. They all said it was "dumb" and many were more interested in Emotiv's preset games so we removed mine and let the users play Emotiv's. Since this was all pretesting, we did not keep any one's data except one person who did sign a consent form and was testing the completed verson.
The goal for this week is to test on more participants and get EEG Lab to download so we can analyze the data and prepare for phase II. Also, Amal and I intend to purchase our tickets to the NSBE conference in Kansas City. We have decided to attend this conference over the conference in Maine because it is a larger conference with more proffesional connections.
This week we started officially testing. Origianlly we had 13 people signed up and confirmed but due to a snow day we lost 4 participants. They are currently being rescedhuled for another session. On the first real day of testing only 2 of the 4 expected participants showed up. One person told us they had trouble finding the room because the google invite did not have the location in the location place instead in the title. To prevent this from happening again Amal sent out reminder emails the night before clarifying the location.
Amal and I have gotten into a good rhythm with testing. I explain the consent form and more about the project and get the software up while Amal sets up the BCI on the participants head. We found this works best because am not comfortable being in other's personal space and it's much easier for me to come up with small talk when the BCI takes a while to get nodes connected.
The goal for next week is to being hopfully gain more particiapnts and then analyze the data using BCI 2000. It is an open source software that can filter and perform complex math on CSV EEG data taken from BCIs. I will spend the rest of this week learning BCI 2000 and EEG lab.
This week we continued our trials on participants. Because of the snow days and cancelations we are still recruiting more particpants for post spring break. Since I only have to run the program, I've started at looking into analyzing the data.
Most programs to analyze data use MatLab and Simmons does not have a subscription so I have been looking into open source softwares. There is program called BCI 2000 that has an already compiled version of a BCI data analyzer that is completely free. I have submitted an application for an account and am waiting to here back.
In the mean time, I am looking through Github to see any other programs that can analyze EEG data. The main problem right now is that many of them used the Epoc which has 14 channels and 512 readings per minute and we have the Insight with 5 channels and 128 readings per minute.
We are continuing the testing that we started before spring break. Unfortunatly, the testing has been more difficult. THe BCI really only works on individuals with short or thin hair and the majority of our perticipants do not have that. So, we are having to settle with 3 out of 5 nodes having full connectivity. Also, the BCI has stopped holding a charge. It will last anywhere from 60 secs to 10 minutes but setting up the nodes takes about 15 minutes plus 2.5 minutes for the data collection.
Amal has contacted Emotiv about the issue but has nto recieved a response. Online it says to keep the BCI charging at all times which we do. But, today it was blinking instead of a constant yellow light indicating that it is charging. Also, the documents are not saved in a way that makes it easier to process the data. The unique ID and time stamps are combined making a long string of numbers that is hard to identify which user it belongs to. I am in the process of matching the time stamps to the schedules and renaming the documents to process faster in BCI 2000.
We are also starting our application to the Undergraduate Symposium. It is due on March 31 but we are aiming to have it completed by next meeting to avoid any last minute struggles while we are away at the NSBE conference.
The Emotiv is still not working and the customer care has still not responded. Also, we submitted our abstract to be considered for the Undergraduate Symposium at Simmons. For the NSBE conference, Amal is fully booked but I am still looking to find a flight back from MCI.
The program is analyzing the data and showing similarity levels but some are too similar. Meaning more than one person can log into each others accounts. I am working to see if this is a flaw in my program or if they are true results.
Amal and I are planning to go to the NSBE conference in Kansas City. Because We are going to be away we are just going to work on getting the emotiv to charge properly and double checking schedueles.
While Amal was at the conference, I was able to get the BCI working. I contacted Emotiv several times only to find out that we were one day from our warranty so they could not help me. The forums provided serveral helpful ideas. I am not sure which one or combo caused the BCI to work. I let the battery completely drain then recharged it to full. There was an external and registered updates that I downloaded to the BCI. Somehow it started working again.
With the BCI working we were able to continue our research testing. Because of the time the BCI stopped working we had to reschedule a few of our participants to later in the week and next week. Additionally, our poster has been accepted to the Undergraduate Symposium and chosen as the keynote presenters for the conference.
This week, we had most of our partiipants come in after the BCI broke earlier and many people forgot that they had an appointment because it had been severAL weeks since their last appointment. I have started analyzing the data and there is a lot of noise. I am currently researching how to tell the difference between extraneious readings and geniune readings. Also, some of the readings are mathematically identical so I am not sure if it is a practical error or truth.
I will be out of the country for the next week so I am trying to get as much work done as possible that requires internet. Also, during the last few participants my computer no longer connected to the BCI after I upgraded my Mac. I had to set up the program on the lab computer to connect to the BCI. Amal and I decidd to only do the keynote presentation instead of the keynote presentation and a poster.
Since we have our keynote presentation coming up we spent most of the time this week working ont he presentaiton. I worked on getting the results analyzed using MATLAB and EEGlab. Then making the results simple enough to present. Amal and I spent a few days practicing and making sure that our presentation is less than 15 minutes.
The results we have were not as good as we had hoped and we came up with ways we would like to add changes in case we were to continue the project. But, we decided to cut that part out of the presentation because it was too long. Originally, we had a live demonstration but the BCI was being tempermental and we wanted to avoid an hiccups we created a video.
Back to the top