Enhancing medical device cleanliness with real-time computer vision and borescopic AI

Business Need

Stringent health regulations requiring 100% assurance of device cleanliness.

Challenges

Training the model to recognize tiny, faint contaminants like soap residues and water droplets.
Ensuring the system performed in real-time with minimal latency.

Solution

AI Real-Time Object Detection for Borescopes
Using YOLOv9, the company fine-tuned a Computer Vision model capable of real-time identification of contaminants observed by borescope cameras. The solution was integrated with an application (accessible via iPads or laptops) to provide inspectors with a live view of the medical device and highlight any detected contaminants.

Results

• Improved Accuracy: Reduced inspection errors by 87%, ensuring contaminants are consistently detected.
• Enhanced Efficiency: Accelerated inspection processes by 45%, allowing for more devices to be checked within the same timeframe.
• Regulatory Compliance: 5+ New Processes compliance with regulation - Provided logged inspection data for audit and compliance purposes.
• User-Friendly Experience: Intuitive real-time visualization allowed inspectors to act promptly, reducing cognitive load and potential oversights.