Quantitative analysis of uterine peristalsis in women with adenomyosis using optical flow for motion quantification

Quantitative analysis of uterine peristalsis in women with adenomyosis using optical flow for motion quantification Sania Latif [1,2,*], Shaheer U. Saeed [3,4], Yipeng Hu [3,4], Lucrezia De Braud [1,2], Ephia Yasmin [1,2], Ertan Saridogan [1,2], Dimitrios Mavrelos [1,2]. [1] Reproductive Medicine Unit, University College London Hospital[2]Institute for Women’s Health, University College London[3] Hawke’s Institute, University College London[4] Department of Medical Physics & Biomedical Engineering, University College London[5] Centre for Bioengineering & School of Engineering and Materials Science, Queen Mary University of London * [e-mail] Description: This repository contains data presented in the manscript available at: [link to be added after publication] The naming convention is as follows:X_Y_Z X: A adenomyosis / N normalY: B baseline / O ovarian stimulation / E embryo transferZ: freq frequency / amp aplitude / emj coordination at emj / myo coordination in myometrium Abstract: Research question: Do women with adenomyosis have altered uterine peristalsis and can we develop an objective and reproducible tool to quantify uterine peristalsis on transvaginal sonography (TVS)? Design: Prospective observational cohort study in 66 women with moderate/severe adenomyosis (n=26) or a normal uterus (n=40) undergoing TVS during assisted conception. TVS recordings (n=143) were performed at baseline (BL), day 6-10 of ovarian stimulation (OS) and day of embryo transfer (ET) by operator A, and then repeated by operator A, B or C). Image data were stored offline for optical flow analysis using 3D Slicer software (Ver. 4.11, slicer.org). Pixel-level displacements for points at the endometrial-myometrial junction (EMJ) were used to compute peristalsis frequency and amplitude by fitting to the 2D-wave equation. Fit error was taken as a measure of wave coordination. Inter-class correlation coefficient (ICC) was calculated to assess inter- and intra-observer agreement. Results: Women with adenomyosis had a significantly higher fit error at all timepoints during the IVF cycle (BL (+ 0.725, p<0.001), OS (+ 0.612, p = 0.004), ET (+ 0.627, p<0.001), indicating lower uterine peristalsis coordination compared to women with a normal uterus. Inter- and intra-observer agreement for each wave feature (frequency, amplitude and coordination of uterine peristalsis and myometrial contractility) was found to be moderate to good (intra-observer ICC 0.62-0.75, inter-observer ICC 0.58-0.79). Conclusion: The proposed approach for wave motion quantification offers an objective and reproducible measuring tool to quantify uterine peristalsis using TVS, with potential for wide application across assisted conception and gynaecology settings.