For many years, achieving a neutral coronal Hip-Knee-Ankle angle (HKA) measured on radiographs has been considered a factor of success for total knee arthroplasty (TKA). Lower limb HKA is influenced by the acquisition conditions, and static HKA (sHKA) may not be representative of the dynamic loading that occurs during gait. The primary aim of the study was to see if the sHKA is predictive of the dynamic HKA (dHKA). A secondary aim was to document to what degree the dHKA changes throughout gait.

We analysed the 3-D knee kinematics during gait of a cohort of 90 healthy individuals (165 knees) with the KneeKG™ system. dHKA was calculated and compared with sHKA values. Knees were considered “Stable” if the dHKA remained positive or negative – i.e. in valgus or varus – for greater than 95% of the corresponding phase and “Changer” otherwise. Patient characteristics of the Stable and Changer knees were compared to find contributing factors.

The dHKA absolute variation during gait was 10.9±5.3° [2 .4° – 28.3°] for the whole cohort. The variation was greater for the varus knees (10.3±4.8° [2.4° – 26.3°]), than for the valgus knees (12.8±6.1° [2.9° – 28.3°], p=0.008). We found a low to moderate correlation (r = 0.266 to 0.553, p < 0 .001) between sHKA and the dHKA values for varus knees and no correlation valgus knees. Twenty two percent (36/165) of the knees demonstrated a switch in the dHKA (Changer). Proportion of Changer knees was 15% for varus sHKA versus 39% for valgus sHKA (p < 0.001).

Lower limb radiographic measures of coronal alignment have limited value for predicting dynamic measures of alignment during gait.

Purpose: Gait analysis has become an innovative approach to assess the biomechanical adaptations due to an ACL injury. However, interpreting the large amount of data collected often requires an expert. Therefore, there is a need to develop an automatic method capable to distinguish kinetic pattern of an ACL deficient patients from an asymptomatic population.

Method: 26 ACL deficient patients and 30 asymptomatic participants took part in a treadmill gait analysis. 3D ground reaction forces (vertical, medio-lateral and anterior-posterior) were collected using the ADAL 3D treadmill. Features were extracted from the 3D ground reaction forces as a function of time and then classified by the nearest neighbour rule using a wavelet decomposition method. The classification method was tested on our data base of 56 participants.

Results: The proposed classification method obtained an accuracy of 90%. The classification accuracy per class was higher for the ACL deficient group allowing classifying correctly 25 out of 26 ACL deficient patient. 25 out of the 30 asymptomatic participants were properly classified.

Conclusion: This study shows that an automatic objective computer method could be used in a clinical setting to help diagnose an anterior cruciate ligament injury during a gait analysis evaluation. Future studies should apply this method on a larger database including data from patients with other musculoskeletal pathologies to help diagnose other injuries.