Idea
An improved way of delivering PDT (Photodymanic Therapy) for surface level lesions. The idea is to create a modified array of LED lights (two types with spectral peak differences of 10 - 15nm), and vary the current through the lights slightly to produce an exitation light custom to each patient. Blue light photos will be taken of the treatment area after each treatment to track progress. The goal is to along with the doctor’s qualitative input, implement a machine learning algorithm to deliver the most ideal spectrum of light to the patient depending on their previous progress, required light penetration depth, and the similarity of physiology with other patients.
How we built it
Web app built using Angular, Node.js, MongoDB Machine learning algorithm implemented using Python Scipy Image processing implemented using Python PIL Hardware built using Arduino
Challenges we ran into
Had to change language from MATLAB to python with less than 24 hours left due to compatibility issues
Accomplishments that we're proud of
Built circuit from scratch Machine learning algorithm built from scratch Built custom imaging algorithm One guy did the whole front end and back end of the web app
What we learned
New languages: syntax, libraries associated with them
What's next for Photodynamic Therapy
Optimization to be done with more data and fine tuning algorithms for edge cases More realistic patient trials
Log in or sign up for Devpost to join the conversation.