Capillary.io is a medical solution designed with three things in mind: objectivity, efficiency and acceleration of the technique while also helping doctors collaborate with each other. It has been created by Borja Gracia (internal medicine) and Eduardo Ramos (software engineer), a work in progress since late 2017.
Finally, we released it to the public in 2020 and we are very happy to share our story with you.
Borja: When I became an internist I tried to specialize in autoimmune diseases. Therefore, I did many capillaroscopies in my daily life. Even when I was finishing, I thought that it was very subjective and dependent on my opinion. Also, I observed that most doctors did not rely on objective data when performing their capillaroscopic reports, and relied on their intuition when observing a few capillaries.
I thought capillaroscopy needed more objective indicators: whole nailfold capillary counting and classification, and capillary size measurement in order to produce accurate reports with more confidence.
The problem was that this consumed a lot of time. Manual counting and inspection of all capillaries was not very practical.
Then, I noticed that at the time, mobile phone camera applications were able to automatically detect faces in photos and even recognize specific people... That's when the idea came to my mind: it should be possible to use a computer to automatically detect capillaries in photos!
I tried to contact engineers to carry out my idea... The truth is that it was very difficult until I was assigned to Calatayud and a colleague knew Eduardo. He loved the idea and we set to work.
Eduardo: We started the project back in October 2017. We did it in our free time and we had our jobs, so first we created a prototype to see if it was viable. Once we saw that it was feasible, we started to collect a lot of capillaroscopy images. For deep learning and artificial intelligence techniques, a lot of data is needed, so Borja had to check all of the images and write down the results one by one so that the computer could understand the data.
After a long time and trying many algorithms and machine learning models... we started to get good results! This was a great milestone of the project, it was finally usable. So I began to build a proper website and web application that would make our system available to everyone.
Useful features kept coming: organizing images in capillaroscopies (for both patients and research), automatic reports, projects, folders and workspaces for collaboration of several doctors... and we even conducted some blind tests to check the correctness of our models against a group of doctors that collaborated with us.
And more recently, we decided to take the project even further and we added new algorithms to automatically measure the capillary loops and limbs, making the overall system even more objective and improving the automatic reports.
Then we thought that uploading images taken with generic microscope applications was not good enough, so we created an application specifically designed to take photos for nailfold capillaroscopy. It had to work with any USB microscope since we wanted everyone to be able to get started immediately with the microscopes that they already owned.
Our goal with Capillary.io is to enable doctors to use nailfold capillaroscopy for their diagnosis and research in the best way possible.
Give Capillary.io an opportunity, we promise you that our tool will make your capillaroscopies better!
To contact us, please send an email to email@example.com.