Abstract
Problem
The identification of unknown orthopaedic implants is a crucial step in the pre-operative planning for revision joint arthroplasty. Compatibility of implant components and instrumentation for implant removal is specific based on the manufacturer and model of the implant. The inability to identify an implant correctly can lead to increased case complexity, procedure time, procedure cost and bone loss for the patient. The number of revision joint arthroplasty cases worldwide and the number implants available on the market are growing rapidly, leading to greater difficulty in identifying unknown implants.
Solution
The solution is a machine-learning based mobile platform which allows for instant identification of the manufacturer and model of any implant based only on the x-ray image. As more surgeons and implant representatives use the platform, the model should continue to improve in accuracy and number of implants recognized until the algorithm reaches its theoretical maximum of 99% accuracy.
Market
Multiple organizations have created small libraries of implant images to assist surgeons with manual identification of unknown implants based on the x-ray, however no automated implant identification system exists to date. One of the most financially successful implant identification tools on the market is a textbook of hip implants which sells for a per unit cost of $200. Several free web-based resources also act as libraries for the manual identification of a limited number of arthroplasty implants. A number of academic and private organizations are working on the development of an automated system for implant identification, however none are available to the public.
Product
Implant Identifier is mobile application which uses machine-learning to instantly detect the model and manufacturer of any common arthroplasty implant, based only on x-ray. The beta version offers a large library of implants for manual identification and is currently available for free download on iOS and Android. Its purpose is to further develop the model to its maximal theoretical accuracy, prior to official release. The beta version of the application currently has over 15,000 registered users worldwide and has the largest publicly available arthroplasty library available on the market. Over 200,000 implant images have been submitted by users to date.
Timescales
The product was initially released in the form of a closed beta which became available to invited guests around 18 months ago. The current version is an open beta which can be downloaded and used by any individual. It was released roughly 12 months ago. The final rendition of the application will allow for free manual identification using the implant library, as well as subscription-based automated implant identification. The implementation, testing and release of this final subscription product is projected to be completed by Q3 2022.
Funding
A small number of early investors have funded the initial research and development of the beta product; however, another round of investment will be beneficial in the final evolution of the product. This additional investment round will allow for completion of development of the identification algorithm, product dissemination, customer support, and lasting sustainability of the venture.
