Reliability and Agreement in the Cerebellar Tonsil Tip Localization: Two Methods Using the McRae Line Concept in MRI

Abstract

The cerebellar tonsil tip position (TP) is a common parameter used for the radiologic diagnosis of Chiari malformation type 1 (CM1). However, these paramedian structures are usually not properly visualized in the midsagittal section. Such mismatch may be a source of bias in TP measurements based on the McRae line (ML) traced between median craniometric points. This study aims to evaluate the intraoperator and interoperator reliability and agreement of 2 protocols to trace the ML in magnetic resonance imaging (MRI) for the cerebellar tonsil tip localization, adding a 3-dimensional multiplanar reconstruction (MPR) approach to the midsagittal plane. Sixty-two T1-weighted head MRIs were obtained for 32 CM1 patients and 30 controls. Two operators independently applied 2 TP measurement protocols, one considering only the visualization of the sagittal plane and the other using MPR. The intraclass correlation coefficient was used to assess intraoperator and interoperator reliability, and the Bland-Altman graphical method was used to evaluate the agreement between the measurement protocols. The sagittal method significantly underestimated ML and tonsillar herniation when compared with the MPR method. The MPR method provided better reliability of the ML measurement when compared to the sagittal method, but this did not influence the reliability of the TP. Analysis of the Bland-Altman plot showed that the limits of agreement were close to acceptable for the ML, but not for measures of TP. The standardization of the LM tracing by the MPR method improves the acquisition of data regarding the position of the tonsils.