Assessment of hindfoot alignment using MRI and standing hindfoot alignment radiographs (Saltzman view)

Objective To compare the hindfoot alignment measured on standing HAV radiographs (Saltzman view) and on non-weight-bearing coronal MR images. Materials and methods The apparent moment arm was measured on weight-bearing conventional radiographs (Saltzman views) and on MRIs of the ankle in 50 consecutive patients (mean age, 54 years; age range, 18–77 years). The evaluation was performed independently by three readers using analogous reference points for both methods. Positive values were assigned when the deepest point of the calcaneus was lateral to the tibial axis as valgus, negative values as varus. The intertechnique agreement and correlation for the measurements performed with HAV radiographs and MRI were assessed for each reader using the Bland-Altman method and the Pearson correlation coefficient, respectively. The interobserver agreement was assessed using the intraclass correlation coefficient. Results The means of apparent moment arms, with the standard deviation (SD) in parentheses, of three readers were +2.0 (±8.4) mm, +1.5 (±6.6) mm and −1.4 (±8.2) mm on HAV radiographs and +4.6 (±7.4) mm, +6.3 (±5.3) mm and +5.4 (±6.4) mm on MRI. The Bland-Altman analysis found a systematic bias for all three readers, corresponding to an overestimation of measurements with MRI (systematic bias ranging from 2.6 to 4.8 mm). The intertechnique correlation was found moderate to high. The Pearson coefficients for the three readers were 0.75, 0.64 and 0.65. The interobserver agreement among the three readers was 0.72, 0.77 and 0.68 for HAV, MRI and both modalities together, respectively. Conclusion Hindfoot alignment can be estimated on MRI but the correlation between the values on HAV radiographs and MR images is only moderate with a tendency to increased positive values (valgization) on MR images.