AI image recognition: checking surgical instruments in the operating theatre tray

What can AI-based image recognition do for the reprocessing of medical devices - and why do we need such a solution? The shortage of skilled labour in the healthcare sector not only affects medical staff, but also many service providers who ensure the smooth operation of clinics and hospitals. The demand for qualified personnel is particularly high in the so-called reprocessing units for medical devices, or AEMP for short.

This is where the instruments required for each operation are cleaned, sorted, packaged and sterilised by hand in advance. At the Charité in Berlin alone, around 14 million surgical instruments are processed every year under the strictest hygiene and quality standards.

Most AEMPs have a zero-defect policy, as any problems that occur here have a direct impact on patient treatment. The staff at the packing stations must therefore ensure that all the instruments required for an operation are fully contained in the so-called operating theatre trays. Not an easy task with up to 160 scalpels, scissors, clamps and other instruments that need to be packed as efficiently as possible into such a tray.

AI-based image recognition developed for surgical instruments

A camera system based on AI technologies will support staff in the future: It is called Cir.Log and is currently being developed by researchers at Fraunhofer IPK.

The camera is designed to recognise and track surgical instruments using machine learning algorithms, without markers and based solely on their appearance. It should reliably localise and inspect different surgical instruments,

• which instruments are actually in a tray,
• which are still missing and also
• Identify instruments that are not part of the sieve
.

Cir.Log will work like a barcode scanner, but without the barcode. Expensive and time-consuming application of barcodes, data matrix codes or RFID chips, as is currently used for tracking surgical instruments, would thus become superfluous. Thanks to its compact design, the camera system can be used on standard packing tables to save space and can be easily installed or retrofitted in any AEMP.

Lower costs and greater process reliability thanks to AI-based image recognition

„We are convinced that our solution offers great added value for hospitals and clinics because it not only saves time and costs, but also improves process reliability“, says Jan Lehr, research associate at Fraunhofer IPK. Cir.Log facilitates the familiarisation of new employees and enables significantly faster packing times right from the start, especially for unskilled or new staff.

„We estimate that experienced employees can work 30 per cent more effectively with Cir.Log. The training time for new staff is reduced by 65 per cent," says Lehr. The camera system also provides digital documentation for every packing process and thus contributes to quality assurance in the AEMP.

Prototypes of Cir.Log are already in use, including at the Charité Campus Benjamin Franklin in Berlin. The aim of the research team at Fraunhofer IPK is to further develop the camera system until it is ready for the market and then to sell it as a spin-off of the institute. To this end, the researchers are currently working on a business plan as part of the Exist research transfer programme of the Federal Ministry of Economics and Climate Protection (BMWK) and are preparing to set up the company. The BMWK is funding the project with around 1 million euros.

In order to adapt the technology even better to their needs, the Fraunhofer researchers are currently conducting a survey among interested AEMPs. The requirements of the survey participants will be incorporated into further development.

Article from "medizin & technik" from 16 January 2024