Automatic Extraction of Hiatal Dimensions in 3-D Transperineal Pelvic Ultrasound Recordings

Deep learning-based pelvic levator hiatus segmentation from ultrasound images

“…There are more considerable differences between the training set and the independent test set, and thus, a more solid experimental conclusion can be drawn. In 2021, Williams et al [26] proposed a method to measure levator hiatus area from 3D pelvic ultrasound images, including C-plane extraction and levator hiatus segmentation. Compared with an expert, the mean HDD was 11.26 ± 5.95 mm and the mean area absolute error was 2.66 ± 2.78 cm 2 .…”

Reply

“…Their method follows the official guideline for manual selection of the SMHD. The same group continued their work with a publication that coincided with the submission and revision of our paper 3 . In that study, the authors created a pipeline for automatic extraction of minimal hiatal dimensions, similar to the one presented in our study.…”

“…As mentioned in our original study, our pipeline can perform the full analysis in a few seconds. While we did not investigate the time it takes to perform the manual analysis, Williams et al 3 reported a time of approximately 2.5 min. We would like to emphasize that our study was conducted in a research setting and the software used has not been commercialized yet.…”

“…In that study, the authors created a pipeline for automatic extraction of minimal hiatal dimensions, similar to the one presented in our study. The study by Williams et al 3 also included evaluation of inter-and intraobserver variability in SMHD selection. Similar to our results, their pipeline demonstrated agreement with the manual assessment.…”

“…As mentioned in our original study, our pipeline can perform the full analysis in a few seconds. While we did not investigate the time it takes to perform the manual analysis, Williams et al 3 . reported a time of approximately 2.5 min.…”