We offer technologies and solutions based on Artificial Intelligence algorithms for medical images and data. Comparing the source datasets to a wide range of reference data models and deciding which ones are most relevant during image and/or data searches.

Medical images recognition technology we used is based on DML (Deep Machine Learning) and it perfectly suits initial diagnose disease, analyse patterns and their opinions and verify markers based on standard (reference) data sets.

Machine learning is currently going through an incredible phase. There is no lack of scientific breakthroughs, practical use-cases, and, of course, the ‘hype’ around it. It is almost impossible to escape news about exciting deep learning applications – from autonomous cars to virtual medical assistants.

Still, major companies are reluctant to use deep learning techniques in their products. And that is because of their major drawback – lack of interpretability.

  • Would you trust a credit score system without means to understands its decision process?
  • Would you trust a doctor that was taught by enigmatic matrix algebra within a week?

The current generation of deep learning techniques might not be able to rock the world in those cases yet.

But it does not need to. We cannot – and don’t have to replace a human. Our goal is to do what our current technology does best – help in tedious manual labor. Namely, in image recognition.

As such, we were able to implement best solution approaches from the following sources.

Our solutions rely on the state-of-the-art level modern deep learning architechtures, such as U-Net variations:

Сases such as the ones mentioned in Projects page have shown us that with the right amount of data and some computational resources one can achieve great results even without any medical understanding. We’re looking forward to seeing our approach find its new practical applications.