Rapid Detection of Common Failure Modes for Knee Prosthetics

点击翻译按钮获取中文摘要

Abstract The number of knee joint replacement procedures has increased exponentially over the last decade with no indications of slowing down. While sustaining highly successful outcomes, implant failures still occur and the annual incidence of knee revision surgeries remain steady around 10% of primary surgeries. Implant loosening has been cited as the dominant cause of long-term failure. Loosening, however, is only detectable when the bone loss is so great that gaps between the implant and the bone are visible in an x-ray image, by which time the loosening may be quite severe with an immediate risk of fracture of the remaining thickness of the bone around the implant. When a patient has pain in or around their knee implant, surgeons are generally reluctant to perform revision surgery in the absence of a clear diagnosis of the cause of the pain. A tool with high sensitivity and specificity and the ability to detect loosening better than current diagnostics would provide a surgeon with the ability to confirm, or rule out, loosening of the implant during the diagnostic assessment of a patient in pain. Bruin Biometrics LLC proposes to develop a non-invasive device that could detect loosening of an artificial knee implant with greater accuracy than the current diagnostics standard. The final product is envisioned as a strap-on device resembling a knee brace which uses sensors to capture acoustic emission and joint angle information while the test subject performs a series of flexion/extension movements. Multiple passive acoustic sensors are in contact with the skin to pick up naturally occurring acoustic emissions from within the body. Signals from the sensors are filtered, analyzed, and categorized to determine if any of the signals are caused by a loosened implant and the degree of loosening. During the related Phase-1 project, we verified a loose implant will produce distinctive acoustic events relative to well-functioning or otherwise worn implants. More recently, we established the additional uniqueness of these signals against naturally occurring native knee signals using a porcine knee model. Evidently, the characteristics of acoustic signals generated by native tissues, such as ligament, tendons, and muscles, are readily distinguishable from artificial joint signals through frequency domain feature filtering. The work proposed for the phase-2 research will build on this base knowledge and develop and optimize a diagnostic tool capable of capturing, locating, and classifying signals indicative of implant loosening. This work will start with further in vitro testing of knees and implant and progress through cadaveric tests and, finally, testing of the device in conjunction with clinical evaluation of live post-op human patients who are experiencing pain.