SurgiScan

(For: TRACK: Patient Safety Technology Challenge (2) )

Inspiration

Surgical errors, such as accidentally leaving tools inside patients, pose significant health risks and lead to severe malpractice issues. We wanted to leverage computer vision to create a solution that enhances patient safety by reducing these preventable mistakes.

What it does

SurgiScan uses computer vision technology to detect and track surgical tools during procedures. It provides real-time monitoring to ensure that no tool is left inside a patient, improving safety and preventing malpractice.

How we built it

We built SurgiScan using the YOLOv5 (You Only Look Once) object detection model, training it on a custom dataset of surgical tools. We utilized OpenCV for image processing and developed a framework to track each tool during its use, providing visual feedback to the surgical team.

Challenges we ran into

Training models with YOLOv5 was extremely challenging, especially when working with custom datasets for specific surgical tools. Finding the right balance between accuracy and speed required a lot of experimentation and fine-tuning.

Accomplishments that we're proud of

We successfully trained a computer vision model that accurately detects and tracks surgical tools. This milestone is a step forward in using technology to enhance surgical safety and reduce medical errors.

What we learned

We gained hands-on experience in building and training object detection models, specifically YOLOv5. We also learned about the complexities of computer vision in real-world applications and the importance of having diverse and high-quality datasets.

What's next for SurgiScan

We plan to improve our detection model's accuracy and expand its tool recognition capabilities. Our next steps include real-world testing in simulated surgical environments and integrating the software into hospital systems to ensure seamless adoption.

Built With

• numpy
• opencv
• python
• yolo
• yolov5