Every year, orthopedic surgeons perform more than 1 million knee replacement surgeries in the United States alone1. Pre- and post-total knee arthroplasty (TKA) lower limb alignment is assessed manually or with the aid of interactive software applications, which can lead to high intra- and inter-reader variability, poor reproducibility and calibration errors. Furthermore, the measurement of angles and lengths on AP standing lower extremity radiographs is a time-consuming task. Currently, the best option to increase standardization in measurements is to continuously train medical professionals involved. Due to time and financial constraints, this is sometimes difficult to implement in hospitals and integrated healthcare systems.
Fully automated measurements of lower extremity radiographs could provide accurate, comprehensive, standardized and reproducible results more efficiently than the current radiological procedure. Graphical reports illustrating all measurements enable explainable artificial intelligence and mitigate the black-box phenomenon.
The radiographic assessment of lower extremities is a standard protocol to estimate limb length discrepancies and knee alignment for the purpose of surgical planning as well as post-operative assessment. Interventions include, among others, open wedge high tibial osteotomy (OWHTO) (2), double-level osteotomy (DLO) (3), and total knee arthroplasty (TKA) (4–6). Each of these use-cases involves different measurements of clinically meaningful lengths and angles and/or definition of loading bearing axes. X-ray assessment of varus or valgus knee alignment is found to be most reliable when using a full leg standing protocol to measure the hip-knee-ankle angle (HKA) (7,8). At the same time, measurements related to load bearing points and axes are critical for surgical planning and follow-up assessment (2,9). Individual phenotypes of femoral and tibial alignment present added complexity to the overall picture of lower limb evaluation (10,11). Measurement of angles and lengths on AP standing lower extremity radiographs is a time-consuming task done with several different interactive software applications. This can lead to intra-reader and inter-reader variability as well as a lack of standardization. IB Lab LAMA’s automated measurements could thus provide accurate, comprehensive, standardized and reproducible results more efficiently than the current radiological procedure.
.png)
IB Lab LAMA™ Version 1.03 is a leg angle measurement assistant powered by artificial intelligence that performs automatic measurements on long leg radiographs with implants, intended to aid medical professionals. This results in acceleration, standardization and improving accuracy in pre- and post-operative assessment of the hip and knee surgeries. LAMA aids in detecting knee alignment deformities by measuring relevant angles on standing AP radiographs of the leg. Additionally, LAMA also aids in detecting leg length discrepancy by providing the following measurements: femur, tibia and full leg length, and the difference between right and left legs on bilateral images.
The standalone performance testing aimed to evaluate the accuracy of the artificial intelligence algorithms that comprise IB Lab LAMA™ compared to expert reads diagnostic angles and lengths in standard AP standing lower extremity radiographs. Two expert readers created a clinical reference standard (“ground truth”) by performing measurements of diagnostic angles and lengths on a pool of AP standing lower extremity radiographs with the aid of commonly used clinical planning software. IB Lab LAMA™ 1.03 measurements were then compared to this ground truth.
The study population consisted of 200 patients (146 female, 53 male, one other; age 69.2±8.9) who were referred for AP standing lower extremity radiograph following total knee arthroplasty (TKA). The patients’ knee systems included the following types of unconstrained and constrained knee implants: ACS (Implantcast® Advanced Coated System), Attune (DePuy Synthes® ATTUNE®), LPS (Zimmer® NexGen® Legacy® Posterior Stabilized), Persona (Zimmer® Persona® The Personalized Knee®), Vanguard (Zimmer® Vanguard®), RHK (Zimmer® NexGen® Rotating Hinge Knee), Attune Revision (DePuy Synthes® ATTUNE® Revision), GenuX (Implantcast® MUTARS® GenuX® MK Revision) and LCCK (Zimmer® NexGen® Legacy® Constrained Condylar Knee (LCCK)).
IB Lab LAMA™ performed within 1° for all angles regarding the mean difference, indicating no substantial fixed bias.
The mean absolute deviation was within 1° for the angle measurements HKA, AMA, lateral femoral component angle (LCFA) and medial tibial component angle (MCTA). For angle measurements JLCA, mLPFA and mLDTA, LAMA™ performed within 2° mean absolute deviation.
The Mechanical Axis Deviation (MAD), Leg length, and Tibia length measurements all performed within 5mm mean absolute deviation. The mean difference was within 2mm for all lengths.
JLCA, LCFA, MCTA and MAD performed exceptionally well when compared with pre-implant images.
IB Lab LAMA™ is the world’s first AI-supported easy-to-use application to perform automated and precise measuring of leg geometry to evaluate lower limb deformities, which promptly produces over 12 radiological measurements that lead to time savings and standardization. With version 1.03, these measurements are also performed on legs with knee and hip implants. Considering there are over four million knee and hip implant surgeries performed in OECD member countries each year, LAMA 1.03 has great potential to significantly reduce the workload and improve the workflow of medical professionals by automating radiological measurements pre-and-post hip/knee surgery.
IB Lab LAMA™ aids in detecting genu varum/valgum by measuring mechanical axis deviation (MAD) and detecting leg length discrepancy by comparing the length of both legs on bilateral images. Detailed analysis of mechanical angles, according to Paley, allows informed decision making on the next steps in treating the patient.
Relevant clinical findings are highlighted in a visual output report by applying the latest international medical standards to enable timely and accurate decision making. The findings are summarized in a graphical output report, attached to the original x-ray image and saved automatically in the PACS system. Optionally, the AI results can be fed as text into a pre-defined RIS template or PDF for accelerated reporting. The AI facilitates the monitoring of disease progression by facilitating the comparison of radiographic disease parameters over time.
