Abstract presenting algorithm for knee wear determination nominated for Wilhelm Roux Award

The abstract “Accurate Wear Determination in Clinical Radiographs after Total Knee Arthroplasty using 2D-3D Registration and Artificial Intelligence: An In Vivo Validation” submitted to this year’s German Congress of Orthopedic and Trauma Surgery (DKOU) was accepted and nominated for the Wilhelm Roux Award. The work presents an automated algorithm for the accurate and precise measurement of wear in total knee arthroplasty using clinical AP radiographs.

Polyethylene wear poses a major risk for the long-term success of a total knee arthroplasty (TKA). Current clinically available methods using standard routine radiographs provide insufficient precision and are not suitable for the objective and valid evaluation of modern implant materials. Alternative methods, such as the highly precise RadioStereometric Analysis (RSA), require cost-intensive special equipment that is not feasible for use in multicenter studies with large cohorts or post market surveillance studies.

A fully automated algorithm to determine implant wear

With the aim of developing a method that would enable precise wear measurements using clinical AP radiographs, our research team – Tanja Bien and Priyanka Grover under the guidance of Dr. Marcel Dreischarf – joined forced with PD Dr. Christoph Lutter and Prof. Dr. Wolfram Mittelmeier of Rostock University Medical Center as well as Prof. Dr. Dr. Thomas Grupp and Marius Selig of Aesculap AG.

The team analyzed 20 postoperative TKA radiographs using our convolutional neural network-based segmentation algorithm in combination with our 2D-3D registration algorithm. The CAD models of the implants were projected onto the Xray images allowing for an exact calculation of the three-dimensional position of the implants in the joint as well as the exact determination of the component distance. This allowed the team to calculate postoperative changes in this distance over time and, subsequently, wear in the implant material.

Automated assessment of knee implant wear and relative implant positioning
Fig. 1: (A) Standard AP radiograph; (B) Segmentation/localization of implant components; (C–E) Calculation of three-dimensional implant positions; (F) Determination of minimal component distance

Precision and accuracy of this fully automated method are comparable to the current gold standard (RSA). Moreover, as it does not rely on a special and costly radiographic setup, the method can be used to objectively investigate new implant materials and designs.

Wilhelm Roux Award session at DKOU 21

We are very proud to be able to share these findings during one of this year’s DKOU program highlights. The presentation will be held by PD Dr. Christoph Lutter during the Wilhelm Roux Award session on Thursday, 28 October 2021, at 10:30am.

The Wilhelm Roux Award honors the best lecture in the field of basic research at DKOU. The goal of the basic research section is to define common research policy content through a close cooperation with the committees of the German Society for Orthopaedics and Trauma (DGOU), the German Society for Trauma Surgery (DGU) and the German Society for Orthopaedics and Orthopaedic Surgery (DGOOC).

More news

Artificial Intelligence: Whitecloud Award nomination for IMAST abstract

The abstract “Novel AI-based algorithm for automated cervical sagittal balance analysis validated on pre- and postoperative X-rays of 129 patients” submitted to this year’s International Meeting on Advanced Spine Techniques (IMAST) was accepted and is nominated for the Whitecloud Award. The work presents an AI-based algorithm for the automated assessment of cervical sagittal balance parameters, demonstrating excellent reliability and accuracy.