Inspiration

Came across state of art portable ultrasound device usable on smartphones. We wanted to build a full solution by analyzing the ultrasound images produced by this device and removing the user's need to go to the hospital.

What it does

Analyzes the image captured by the portable ultrasound device and calculates the probability of a cardiac health complication present within the user.

How I built it

Used a machine learning framework called Tensorflow and Python.

Challenges I ran into

First time dealing with machine learning, big learning curve, numerous small bugs, had to sample numerous frameworks before settling with Tensorflow. Biggest challenge: not enough data to train algorithm.

Accomplishments that I'm proud of

Got the neural net/image classifier working.

What I learned

Machine learning.

What's next for Lil' Ultrasound

Improve the accuracy of the algorithm by collecting more data. Expanding the scope of the algorithm to other applicable body parts such as the liver, kidneys, etc.

Share this project:

Updates