OR Medical Device Recognition for Sterilization-Safe Usage Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing medical device tracking systems require manual logging or augmentation with tracking tags that are not compatible with autoclave sterilization, making them inefficient and impractical for reusable medical devices.
Innovation Solution
A system utilizing a capture device with a camera and machine learning model to automatically recognize medical devices in an operating room, transmitting data to a base station for recognition and analysis by an analytics server, which generates usage data without requiring RFID tags or other modifications to the devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual logging is used to track medical device usage, then device usage can be recorded, but the process is inefficient and labor-intensive
Solution Approach 1:
The patent replaces manual logging (mechanical human operation) with an automated image recognition system using machine learning models. The system captures images of medical devices, automatically identifies them through ML algorithms, and logs usage data without human intervention, thereby improving efficiency and eliminating time loss associated with manual tracking.
2Extent of automation
If tracking tags (RFID, retro-reflective IR) are attached to medical devices, then automated tracking is enabled, but the tags are not compatible with autoclave sterilization
Solution Approach 1:
The patent extracts the tracking functionality from physical tags attached to devices and replaces it with an image-based recognition system. Instead of requiring tags on each device, the system captures images and identifies devices through machine learning, eliminating the sterilization compatibility issue while maintaining automated tracking capability.
Solution Approach 2:
The patent creates a digital representation (image) of the medical device instead of using a physical tag. The machine learning model recognizes devices by analyzing visual features in images, effectively copying the identification function from physical tags to digital image processing, thereby avoiding sterilization problems.
3Measurement precision
If medical devices are augmented with embedded or attached tracking tags, then device identification is improved, but the devices require modification which increases complexity and cost
Solution Approach 1:
The patent replaces physical augmentation of devices with tags or embedded identifiers with a non-invasive image recognition system. The machine learning model identifies devices based on their visual appearance in captured images, eliminating the need to modify devices while maintaining accurate identification capability.
Data Source
AI summary
A system for medical device usage monitoring includes one or more capture devices, one or more base station devices, and an analytics server. The capture devices each capture image data indicative of a monitored location at a timestamp in an operating room and transmit the image data to a base station device. Each base station device recognizes medical devices in the image capture data with a trained machine learning model and transmits recognition data to the analytics server. The analytics server generates medical device usage data over time based on the recognition data. The analytics server may infer usage based on the recognition data. The analytics server may analyze the medical device usage data and generate analytical usage data. Methods associated with the system are also disclosed.