Purpose: To study the interobserver agreement for the analysis of lesion filling following cementoplasty of an acetabular osteolytic lesion, and investigate how subjective analysis compares to volumetric analysis.
Materials and methods: A total of 21 acetabular osteolysis were retrospectively analysed on pre- and immediate post-cementoplasty CT-scans by two senior interventional radiologists and one resident using a 4 grade scale to quantify lesion filling (Fsubjective): F ≤ 25%, 25% < F ≤ 50%, 50% < F ≤ 75% or F > 75%. Volumetric analysis (Fvolumetric) was performed with the delineation of the osteolysis and the cement using regions of interest. The interobserver agreement for Fsubjective was evaluated using the Fleiss' Kappa test for the 4 grade scale and for a simplified 2 grades scale (F ≤ 50% and F > 50%). The performance of Fsubjective versus Fvolumetric (considered as the gold standard) was then evaluated for each reader using the calculation of accuracy and error to reference for the 4 grades scale and accuracy, sensitivity, specificity, positive predictive value and negative predictive value for the 2 grade scale.
Results: Interobserver agreement was considered as very low (< 0.2) for the 4 grade scale analysis and as low (> 0.2 and < 0.4) for the 2 grade scales analysis with kappa factors of 0.196 and 0.36 respectively. Compared to volumetric analysis, the overall accuracy of the 4- and 2- grade scales were 36.5% and 60% respectively.
Conclusion: Subjective evaluation of cement filling of an acetabular osteolytic lesion is associated with poor interobserver agreement and overestimation of the percentage compared to volumetric analysis.
Keywords: Acetabulum; Cementoplasty; Lesion filling; Volumetry.