Monitoring IV Fluid Bag Levels with the Arduino Portenta H7 and Vision Shield

When a patient is receiving intravenous (IV) fluids, it is essential that air is not introduced into the line, as its presence can create problems such as excessive pressure or even emboli that can put the life in danger. Normally, the level of fluid remaining in the bag is checked periodically by a nurse, but due to staffing issues, this may not be enough. Therefore, Manivannan Sivan designed an automated surveillance system that uses computer vision as well as machine learning to perform this repetitive task instead of a person.

To start the project, Sivan put together a series of images covering three categories with an Arduino Portenta H7 and Vision Shield. In each image, the IV fluid bag is 15 cm away and has either an adequate level, less than 50%, or a low level of remaining fluid. Once trained, his model was able to successfully recognize the correct level about 96% of the time, although additional images taken at different light levels would help improve accuracy even further.

From there, Sivan deployed his model to the Portenta H7 to see how it performed in everyday conditions, with classification taking place once every two seconds. His plan is to take the resulting category and export it to a waiting server on the network so that hospital staff can see the status of the fluid bag in real time on a dashboard.

/

For more details on this project, you can watch Sivan's explainer video below or read his write-up on the Edge Impulse documentation page.