Aims

The purpose of this study was to develop a convolutional neural network (CNN) for fracture detection, classification, and identification of greater tuberosity displacement ≥ 1 cm, neck-shaft angle (NSA) ≤ 100°, shaft translation, and articular fracture involvement, on plain radiographs.

Varus malalignment after total knee replacement is associated with a poor outcome. Our aim was to determine whether the same was true for medial unicompartmental knee replacement (UKR). The anatomical leg alignment was measured prospectively using a long-arm goniometer in 160 knees with an Oxford UKR. Patients were then grouped according to their mechanical leg alignment as neutral (5° to 10° of valgus), mild varus (0° to 4° of valgus) and marked varus (> 0° of varus). The groups were compared at five years in terms of absolute and change in the Oxford Knee score, American Knee Society score and the incidence of radiolucent lines. Post-operatively, 29 (18%) patients had mild varus and 13 (8%) had marked varus. The mean American Knee Society score worsened significantly (p < 0.001) with increasing varus. This difference disappeared if a three-point deduction for each degree of malalignment was removed. No other score deteriorated with increasing varus, and the frequency of occurrence of radiolucent lines was the same in each group. We therefore conclude that after Oxford UKR, about 25% of patients have varus alignment, but that this does not compromise their clinical or radiological outcome. Following UKR the deductions for malalignment in the American Knee Society score are not justified