Vision Tracking System for Surgical Tray Equipment
Prepared by Alec Maier, Daniel Brewer & James Miller
Introduction
The Vision Tracking System for Surgical Tray Equipment (VTSTE) is being designed for use in the Sterile Processing Department (SPD) of Lakeland Regional Health. This system will provide technicians with an added layer of verification that items requested for use in an operating room (OR) are in-fact included in a tray, and that trays returning to the SPD from the OR still contain the required items. The purpose of this system is to reduce lost and / or forgotten items used in a surgical operation - mostly unintended retention of foreign objects in a patient.
Our sponsor expects the VTSTE to provide technicians with an added layer of verification that items requested for use in an operating room (an OR) are in-fact included in a tray. The system should also verify that trays returning to the SPD from the OR still contain the required items. The sponsor expects a solution that will reduce lost and / or forgotten items used in a surgical operation. The sponsor also expects the system to lower the required time to organize tools returned to the SPD.
Due to COVID-19, our work inside Lakeland Regional Health was ended. The road-map included training the Mask R-CNN machine learning model on the SPD's daily supply flow of items; however, this quickly became impossible. A shift was made towards collecting data from images available online. We were unable to gather enough real-world information in this manner and defaulted to the COCO dataset found here: http://cocodataset.org/#home. Future work would benefit greatly from returning to the SPD to build a specialized dataset for training the necessary machine learning model on the specifics of the tools seen there.
Results
