Difference between revisions of "Team:Vilnius-Lithuania/Software"

Line 25: Line 25:
 
     <h1>Software part: LipoVision</h1>
 
     <h1>Software part: LipoVision</h1>
 
     <p></p>
 
     <p></p>
−
     <h2>+ Utilizing computer's capability of vision</h2>
+
     <h2>Utilizing computer's capability of vision</h2>
 
     <p></p>
 
     <p></p>
 
     <p>
 
     <p>
Line 31: Line 31:
 
     </p>
 
     </p>
 
     <p></p>
 
     <p></p>
−
     <h2>+ Picking our technology stack</h2>
+
     <h2>Picking our technology stack</h2>
 
     <p></p>
 
     <p></p>
 
     <p>Our outlook for LipoVision included making this tool available on all platforms, preparing it to be easily deployable, more robust than a collection of scripts, and fully extendable. It also had to be simple and intuitive to use for our lab team - running from the console is highly undesirable, so a well supported graphical user interface was also within our aims. We have developed the device compatibility in a modular style, so adding a new one to the supported list is fairly easy.
 
     <p>Our outlook for LipoVision included making this tool available on all platforms, preparing it to be easily deployable, more robust than a collection of scripts, and fully extendable. It also had to be simple and intuitive to use for our lab team - running from the console is highly undesirable, so a well supported graphical user interface was also within our aims. We have developed the device compatibility in a modular style, so adding a new one to the supported list is fairly easy.
Line 46: Line 46:
 
     <p></p>
 
     <p></p>
 
     <h2>
 
     <h2>
−
             + Process
+
             Process
 
     </h2>
 
     </h2>
 
     <p></p>
 
     <p></p>

Revision as of 21:46, 4 November 2018

Software

Computer with Eyes

In pursuit to substantially reduce manual effort in performing microfluidic experiments, we have designed a software called LipoVision. One of the crucial steps for a successful liposome synthesis is the correct preparation of the microfluidic device. A coating procedure is critical for rendering half of the device hydrophilic and half of it hydrophobic. Not only does it require precise attention, the procedure quite often fails because of the human made errors while controlling the infuse rates quickly in a response due to an instability. LipoVision software reduces human labor down to the bare minimum and optimizes the coating procedure entirely. It uses an open standard computer library OpenCV at its’ core, detects the events at the interphase and controls the pumps for the accurate infusion rates according to the situation. The LipoVision software is based on Go and available on all operating systems and is accessible to any custom microfluidic experiment.

Software part: LipoVision

Utilizing computer's capability of vision

The roots of LipoVision can be traced back to the BioHackathon where the idea first took form. BioHackathon provided a splendid environment for the development of ideas, as teams obtained assistance from Lithuanian IT leaders - our event mentors. This year one of the main aspects of our team’s project was the production of liposomes using microfluidical devices. Prior to using the microfluidic devices, they need to prepared by coating the microchannels with PVA, to ensure the post-junction channels are hydrophobic. This step is a very time-consuming process and requires constant attention, while also being particularly prone to failure, which leads to many hours of wasted time. To resolve this problem we created the LipoVision software, which automises the coating process.

Picking our technology stack

Our outlook for LipoVision included making this tool available on all platforms, preparing it to be easily deployable, more robust than a collection of scripts, and fully extendable. It also had to be simple and intuitive to use for our lab team - running from the console is highly undesirable, so a well supported graphical user interface was also within our aims. We have developed the device compatibility in a modular style, so adding a new one to the supported list is fairly easy.

Our objectives laid out - we had to pick our collection of technologies. First and foremost - we had to enable our computers to see for us. Computer vision has advanced to the state of being broadly available and incredibly useful. For our use case we have utilized an Open Source, free to use library OpenCV. This implementation was chosen because of it's portability, support and good documentation. This ensures us that our project is going to be well supported in the long term and that a host of contributors familiar with the technology would currently exist.

Our next pick - a language developed by Google engineers - Go. Outside of Bioinformatics, Go is a language rapidly growing in popularity and use. Its intuitive style and unforgiveness to mistakes guides its users into writing beautiful code. Not surprising, one of it’s forefathers is Ken Thompson, well known for for creating the B language, which in turn is a direct predecessor to C, that we all know and love. An extremely broad standard library and easy management of dependencies reduces the amount of moving parts, making the code more robust. The advantage of having most code being compiled into a single binary was also beneficial, as it simplifies installation and deployment in the field.

Our last choice, having a cross-platform unified user interface, the choice was the GTK+ collection. It works across Linux/OSX/Windows, which perfectly corresponds to our goals..

Process

After setting up syringes and plugging them into the liposome production device the User opens up the application. Usage of LipoVision follows these steps:

  1. Selecting a device from the drop-down. The application will start receiving the stream and showing it on the window, along with a small section at the corner that shows what LipoVision has locked onto.
  2. The user has controls of the camera and all four pumps in the device, he has to adjust the image so that the junction could be recognized. The vertically aligned channels enable the application to recognize danger areas automatically. The user should make sure they are recognized correctly.
  3. Once the recognized region satisfies the user, he has to lock the position in place, and start the coating process.
  4. The last available option is ‘Auto coating’. When all parameters are satisfied and it is enabled, LipoVision will start watching and calculating if the formed bubble is in good shape. When a certain tolerance is crossed, it will increase the air pressure to the channels in the middle, preventing the coating solution from going upstream.

Conclusion

We successfully developed a highly accessible piece of software, that saves large amounts of time by automating an otherwise tedious process, while potentially saving even more time as well as resources by eliminating the risk of failure. This not only removes the chance of requiring to repeat the entire chip preparation process, but also frees up the user due to constant supervision of the process no longer being required.

Example LipoVision video
invert