This prospective observational validation study aims to evaluate the accuracy and reliability of artificial intelligence-based video analysis for quantifying anterior knee laxity. Participants with unilateral anterior cruciate ligament (ACL) injury and ACL-intact controls will undergo bilateral knee examination using a Rolimeter by two independent orthopedic surgeons. Three consecutive measurements will be obtained for each knee and examiner. Each examination maneuver will be video recorded using a standardized acquisition protocol including a metric reference of known dimensions. The videos will subsequently be analyzed independently by a multimodal artificial intelligence system blinded to clinical diagnosis, MRI findings, injured side, examiner measurements, and contralateral knee measurements. AI-derived anterior tibial translation and side-to-side difference will be compared with Rolimeter measurements. The primary objective is to assess agreement between AI-derived and Rolimeter-derived side-to-side difference. Secondary objectives include inter-examiner reliability, AI repeatability and the diagnostic accuracy of AI-derived measurements for identifying ACL injury.
Each examiner will perform three consecutive Rolimeter measurements on both knees. Individual maneuvers will be video recorded so that the Rolimeter and AI measurements refer to the same mechanical examination. Videos will be pseudonymized and analyzed independently. The AI system will have no access to participant group allocation, MRI findings, injured side, Rolimeter measurements, or results from other videos belonging to the same participant. A standardized metric reference positioned in the plane of the examined knee will be visible in each recording to permit estimation of displacement in millimeters. AI measurements will not be used for clinical decision-making.
Study Type
OBSERVATIONAL
Enrollment
70
Ospedale Vito Fazzi
Lecce, Italy
Agreement Between AI-Derived and Arthrometer-Derived Side-to-Side Difference in Anterior Tibial Translation
Agreement between the side-to-side difference (SSD, mm) in anterior tibial translation estimated by artificial intelligence from standardized knee examination videos and the corresponding SSD obtained with an arthrometer. Agreement will be assessed using intraclass correlation coefficient (ICC, absolute agreement) and Bland-Altman analysis, including mean bias and 95% limits of agreement.
Time frame: Day 1 (index study examination)
Measurement Error of AI-Derived Anterior Tibial Translation Compared With Arthrometer Measurements
The difference between AI-derived and arthrometer-derived measurements of anterior tibial translation will be quantified in millimeters using mean absolute error (MAE), root mean square error (RMSE), mean bias, and 95% limits of agreement.
Time frame: Day 1 (index study examination)
Inter-Examiner Reliability of Arthrometer Measurements
Reliability of anterior tibial translation and side-to-side difference measurements obtained independently by two orthopedic examiners using an arthrometer. Inter-examiner reliability will be assessed using intraclass correlation coefficients with 95% confidence intervals and Bland-Altman analysis.
Time frame: Day 1 (index study examination)
Agreement Between AI Measurements Derived From Examinations Performed by Different Examiners
Agreement between AI-derived anterior tibial translation and side-to-side difference obtained from videos of knee examinations performed independently by the two examiners. This outcome will assess the robustness of AI-derived measurements with respect to examiner-related variability.
Time frame: Day 1 (index study examination)
Diagnostic Accuracy of AI-Derived Side-to-Side Difference for ACL Injury
The ability of AI-derived side-to-side difference to discriminate participants with complete anterior cruciate ligament injury from ACL-intact controls will be assessed using receiver operating characteristic (ROC) curve analysis. Area under the curve (AUC), sensitivity, specificity, positive predictive value, and negative predictive value will be calculated.
Time frame: Day 1 (index study examination)
diagnostic Accuracy of Arthrometer-Derived Side-to-Side Difference for ACL Injury
The diagnostic performance of arthrometer-derived side-to-side difference for identifying complete anterior cruciate ligament injury will be evaluated using ROC curve analysis, including AUC, sensitivity, specificity, positive predictive value, and negative predictive value.
Time frame: Day 1 (index study examination)
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.