Discrepancies in MR- and CT-Based Femoral Version Measurements Despite Strong Correlations

Abstract

Purpose

To determine the correlation and classification consistency of femoral version measurements between MR and CT using four commonly used measurement methods.

Methods

A retrospective study was performed on femoroacetabular impingement (FAI) patients who received preoperative CT and MR imaging assessment of the surgical hip and ipsilateral distal femur. Femoral version was measured using the Murphy method, the oblique method, the Reikerås method and the Lee method. Intra- and inter-rater agreements were calculated. Linear regression and Bland-Altman analysis were performed for measurements using different imaging modalities and measurement methods. Femoral version measurements within the lower quartile, the middle 2 quartiles and the upper quartile were classified into different groups based on their percentile within the sample population. Classification consistency rates between modalities and methods were calculated and compared.

Results

Fifty-three patients (39.4 ± 9.1 years; 32 female) were included for analysis. Intra- and inter-rater reliability were high for all modalities and methods (intra-rater ICC range, 0.963-0.993; inter-rater ICC range, 0.871-0.960). MR- and CT-based femoral version measurements showed strong correlations for all methods, with the Lee method demonstrating the strongest association (r = 0.904), while the oblique method exhibited the lowest correlation (r = 0.684) (all P < .001). MR-based measurements were smaller than CT-based measurements with mean differences ranging from 4.5° to 10.3°. Classification consistency between MR and CT ranged from 51% to 74%, whereas the consistency between different measurement methods ranged from 68% to 85%.

Conclusions

While strong correlations were observed between MR- and CT-based femoral version measurements, MR-based measurements were significantly smaller than their CT counterparts. Classification consistency between the modalities was moderate to high. Measurements between different methods showed strong correlations with high consistency rates.

Level of Evidence

Level III, retrospective case series.