RESUMO
OBJECTIVES: To develop a semi-automated method to assess puborectalis muscle echogenicity on three-dimensional/four-dimensional (3D/4D) volume transperineal ultrasound images using 4D View and Matlab® software and evaluate its intra- and interobserver reliability. METHOD: The data of 23 women in their first trimester were included. 3D/4D volume datasets were obtained at rest. Two inexperienced observers were trained by an experienced observer to construct tomographic ultrasound images (TUI) from the original data and to delineate all structures. Puborectalis muscle area (PMA) and the mean echogenicity of the puborectalis muscle (MEP) were calculated offline. Intra- and interobserver reliability were determined by intraclass correlation coefficients (ICC) and their 95% CIs. RESULTS: The development of a semi-automated method to calculate puborectalis area and echogenicity is described in detail. PMA and MEP measurements in pregnant women demonstrated almost perfect intraobserver reliability for both inexperienced observers, with ICC values ranging from 0.88 to 0.99. The interobserver reliability showed ICCs of 0.63 for PMA and almost perfect ICC values, of 0.96-0.98, for echogenicity. The majority of intraobserver mismatch between two delineations of PMA occurred near the borders. CONCLUSIONS: Matlab software can be used to provide reliable measurements of the area and echogenicity of the puborectalis muscle. As the latter can be used to assess structural changes in the puborectalis muscle, it appears a promising new tool for studying pelvic floor structural anatomy.
