Data collection system for surgical instrument localization

Unmet Need

The operating room represents the largest cost and profit center for health systems across the United States. Despite its importance, very little data is gathered describing open surgery. Surgical technique is more of an art than a science, with residents receiving predominantly peer-to-peer tutoring with subjective and variable feedback. Resulting significant variation in cost and patient outcomes motivate the development of sensing technologies to identify methods of surgery.

Technology

RFID systems are currently the preferred method for tracking surgical instruments during operations. RFID can determine which instruments are used when, but understanding how instruments move is still a challenge. Imprecision in RFID communication parameters make positional localization impossible without machine learning and large training datasets. Duke inventors have developed a system for generating large artificial datasets that can be used to pretrain algorithms that predict location from RFID parameters. This technology is intended to be used to generate large, labeled training datasets to inform RFID tracking systems used in surgical operation. Specifically, the present invention describes a positioning robot holding an RFID-tagged instrument that will move the instrument randomly to generate a diverse dataset of different positions with concurrent RFID sensing. A computer links instrument localization vectors with corresponding RFID parameters in a database for use in supervised training and testing.

Other Applications

This technology collets data used to pre-train ML models, which locate instruments used in surgeries based on responses of their RFID-tags. Further, the type of datasets generated by this model can be used to understand how surgeons operate and correlate the usage of specific tools with specific surgery outcomes for surgical residents. With adjustments, the robotic system can also generate datasets for localization leveraging other sensor systems.

Advantages

• This robotic system can generate training datasets quickly and cost-efficiently for RFID-based instrument localization compared to other data generation methods.
• The dataset generated by the robotic system is easily adjustable based on application-specific criteria.
• With adjustments, the robotic system can be used to generate cost efficient and large datasets for other localization purposes.