Synopta

Real-time surgical clarity with AI.

Comment

Inspiration

One of the team members, going through an ACL surgery, experienced that doctors have very limited guidance during unusual events occurring in the operation room. Our team was immensely inspired by the experience and wanted to create a solution which provide doctor’s tools and guidance to cater to such event. The end goal of Synopta is to enhance surgical productivity and improve patient outcomes

What it does

Synopta is an AI agent that processes endoscopic image data in real-time frame by frame, providing guidance through a prompt-engineered system with audio capabilities. By providing audio input and output, the user can now interact and use the system while actually performing surgery. We utilized the Cerebras AI API built on top of Lama 3.4, leveraging fast LLM processing speed. Synopta can help provide essential guidance to the doctor’s performing the surgery in real time and can greatly impact in situations where the doctor must take a call. Additionally, with seamless interface the doctor can talk to AI by audio which greatly help the surgeon or doctor gather information instantly. Over 40% of patients experience some sort of complications during surgery. With Synopta, doctors can boost surgical productivity drive better decision-making. Synopta can significantly help millions of doctors during their surgeries on the go and possibly impact millions of lives.

How we built it

For building our intelligent AI agent we levered Cerebras AI due to its fast-processing speeds and scalable architecture. We are processing around 40 frames per second, and so utilizing the Cerebras helps us make inference real-time. We used OpenCV (for live video and image processing), Next.js for the frontend and authentication, and Flask for serving the backend. Our product is built on an architecture which is scalable and flexible that adapts to the needs of the user.

Challenges we ran into

One significant hurdle that we had to overcome was to optimize our prompt engineering system to ensure proper functioning of the product. Additionally, creating a webapp which could facilitate real time analysis of data along with voice communication protocol was challenging for us to integrate.

Accomplishments that we're proud of

• Creating a voice communication protocal via Cerebras AI to facilitate easy transfer of thoughts while creating an interactive experience with the user and the system. • Successfully created a prompt-engineered solution which consistently delivered precis responses on a scalable and robust architecture. • Designed and implemented a real-time image processing web application in order to give guidance during surgical procedures with audio feedback. • Building a robust and flexible system architecture to support cross-sector implementation of the core technology.

What we learned

Throughout the entire project we got familiar with using Cerebra’s API, building OpenCV applications and integrating all the functionalities with a web app.

What's next for Synopta

By building classification models along our prompt-engineered solution, we can successfully detect diseases which are normally difficult to find by human eye. Furthermore, we are also going to extent our use cases in other types of surgical procedures.

Built With

Share this project:

Updates