Solve clinical problems with artificial intelligence.
Clarius AI is a platform where you can define your research studies, acquire and label data, export to Deep Learning tools such as TensorFlow, and then validate your models in real time.
How it works
1. Define a clinical problem.
Create a research study and customize parameters, such as the type of study, what you'll be labeling, your research team members, and more.
2. Acquire data.
You can use both prospective or retrospective data. Use anonymized data from your existing ultrasound exams, or scan new exams and add them to your study.
3. Label the data.
Our labeling tools allow you to quickly label large amounts of data. Keyboard hotkeys make it easy to go through sequences of frames.
4. Export your data and train model.
Export your labeled data as a TFRecord to use with your training framework of choice.
5. Validate your model in real-time.
Our Listener API enables you to test your AI model built with TensorFlow as you scan in real time. Scan a patient and see how your model performs on a secondary device for immediate live validation.
6. Deploy your model to the Clarius App.
As AI models get validated, Clarius can make these accessible in the Clarius app for others to use.
“One of the most challenging aspects of AI in healthcare is the ability to capture quality data that focuses on specific clinical needs. We're accelerating AI development by making it easy to acquire and label large amounts of data with our scanners and cloud platform.”
— Dr. Reza Zahiri, Director of Innovation
Clarius AI is currently being used by various universities and research institutes to solve clinical problems. Here are a few partners we're working with now.
Automatic Axillary Artery Detection
Automatic detection of the axillary artery in ultrasound images
SINTEF Medical Technology, Trondheim, Norway
Nerve and Blood Vessel Visulization
Investigators: Erik Smistad, Kaj Fredrik Johansen, Daniel Høyer Iversen, Ingerid Reinertsen
Application: Highlighting nerves and
blood vessels for ultrasound-guided axillary nerve-block procedures
ThinkSono, London, UK
Investigators: Fouad Al-Noor, Sven
Mischkewitz, Antonios Makropoulos
Application: DVT Detection