Reliability and Validity of Visual Estimation of Femoral Head Hypoperfusion on Perfusion MRI in Legg-Calve-Perthes Disease

Background Assessing the severity of Legg-Calve-Perthes disease is important for treatment decision-making and determining prognosis. Perfusion magnetic resonance imaging (MRI) scans have improved our ability to assess femoral head hypoperfusion, and the use of an image analysis software called HipVasc (Scottish Rite for Children, Dallas TX) has allowed precise quantification for research purposes. In the clinical setting, using HipVasc software is not practical, and visual estimation is used to assess hypoperfusion. Currently, the reliability of visual estimation is unknown. The purpose of this study was to determine the reliability and validity of visual estimation of hypoperfusion on perfusion MRI and compare the results to measurements obtained with the HipVasc software. Methods Fourteen pediatric orthopaedic surgeons, divided into 4 groups, participated in this study. Each surgeon completed a 30-minute in-person training on HipVasc before proceeding. Stratified by category of severity, 122 MRIs were randomly assigned to 1 of the 4 groups so that each group member measured the same set of ∼30 MRIs. Each surgeon documented their visual estimation of hypoperfusion first, then used HipVasc to measure the hypoperfusion and recorded their results. Interobserver reliability was assessed at the group level by calculating the intraclass correlation coefficient. Results Good correlation and reliability was found between the visual estimate and HipVasc measurements with an overall mean intraclass correlation coefficient of 0.87 and mean Pearson coefficient of 0.90. The mean interobserver reliability of visual estimation was 0.84. Observers had a tendency to overestimate hypoperfusion with visual estimation. Conclusions This study demonstrates the reliability and validity of visual estimation as a clinically applicable method for determining femoral head hypoperfusion. Our study is the first to measure the reliability of visual estimation and HipVasc with a large cohort of clinicians, specifically pediatric orthopaedic surgeons with varied experience in graphical software analysis. We recommend using visual estimation as a clinically applicable method to make treatment decisions, and its accuracy is comparable to HipVasc software analysis. Level of evidence Level III-diagnostic.