{"paper_id":"3941d0d8-c889-4dd6-909b-f0f177606716","body_text":"ARTICLE OPEN\nNoninvasive imaging of 4D electrical activation patterns of\nuterine peristalsis during normal menstrual cycles\nSicheng Wang 1,2,3,7, Kelsey Anderson 3,4,7, Stephanie Pizzella 2,3, Haonan Xu 2,3, Wenshang Wang 1, Zichao Wen 2,3, Yuan Nan 2,3,\nJosephine Lau 2,3, Qing Wang 5, Valerie Ratts 3,4 ✉ and Yong Wang 2,3,5,6 ✉\nUterine peristalsis, characterized by spontaneous slow-wave contractions of the subendometrial layer of the uterine myometrium,\noccurs throughout the menstrual cycle. Disruptions in peristalsis patterns may occur in women experiencing abnormal uterine\nbleeding, endometriosis, and infertility. Current tools to measure uterine peristalsis in humans have limitations that hamper their\nresearch or clinical utility. Here, we developed an electrophysiological imaging system with wearable electrical sensors to\nnoninvasively quantify the four-dimensional electrical activation pattern during human uterine peristalsis with high spatial and\ntemporal resolution and coverage. To demonstrate capabilities of this new imaging system, we enrolled 26 pre-menopausal\nparticipants ages 18 –40 years old with regular menstrual cycles (25 –34 days in length) and normal gynecologic anatomy.\nParticipants initially underwent magnetic resonance imaging to obtain the body-uterus geometry. Then, during each of the four\nphases of a single menstrual cycle, patients underwent noninvasive electrical recording for 30 min. We measured uterine peristalsis\nfrequency, direction, duration, magnitude, and power and found that these parameters differed in different phases of the menstrual\ncycle. Future use of this uterine peristalsis imaging system may advance our understanding of the role of uterine peristalsis in\nmaintaining normal uterine function and contributing to gynecological pathophysiology.\nnpj Women’s Health             (2024) 2:1 ; https://doi.org/10.1038/s44294-023-00003-x\nINTRODUCTION\nDuring each of the four phases of the human menstrual cycle, the\nsubendometrial layer of the uterine myometrium generates slow,\nlow-magnitude, spontaneous contractions termed uterine peri-\nstalsis\n1–10. During the menses phase, peristalsis waves primarily\npropagate from the fundus to the cervix (F –C), aiding the\nexpulsion of blood and tissue. In contrast, during the peri-\novulatory phase, peristalsis waves predominantly propagate from\nthe cervix toward the fundus (C –F) and are thought to help\ntransport sperm toward the fallopian tubes.\nUsing transvaginal ultrasound (TVUS), Ijland et al. detected\nuterine peristalsis waves traveling C –F and others traveling\nF–C11,12. TVUS has a few limitations. First, TVUS requires insertion\nof a vaginal probe, which can be uncomfortable for the\nparticipant. Second, the quality of TVUS images may be limited\nby the position and orientation of the uterus. Third, traditional\nvisual inspection of TVUS video is sometimes subjective and\noperator dependent, causing limited interobserver agreement in\ndetermining peristalsis direction 13. Therefore, TVUS may result in\nincomplete findings14–19. Recent developments of quantitative\nvideo analysis like strain analysis 20,21 and speckle tracking 22–24 can\ngenerate quantitative measures.\nOther methods that have been used to study uterine peristalsis\nalso all have signi ficant limitations. Intrauterine pressure catheters\nare invasive, and a catheter placed inside the uterus could alter\nperistalsis patterns. Hysterosalpingography cannot be used to\nmeasure peristalsis amplitude or frequency, and radiation\nexposure limits the imaging duration. Cine magnetic resonance\nimaging (MRI)\n25–28 can be used to detect uterine peristalsis by\nacquiring sequential images over time and playing the MRI frames\nat elevated speed 25. However, extended cine MRI is expensive,\ntime-consuming, and operator-dependent, and it cannot reveal\nthe initiation and termination sites of uterine peristalsis. Moreover,\nall these modalities can be uncomfortable for the participant and\ncannot be used for long-term observation.\nAn alternative technique for evaluating uterine peristalsis is to\nrecord the slow-wave electrical signals that drive contractions. For\nexample, Kuijsters et al.\n7 measured spontaneous electrical signals\nin ex vivo human uteri. In another study, electrodes were placed\ninside the nonpregnant uterine cavity to directly measure\nelectrical activity on the uterine surface\n29. Additionally, Sammali\net al. 30 used transabdominal electromyography to measure\nuterine electrical activity from electrodes placed on the body\nsurface. However, this method only captures high-frequency\nelectrical signals from a small abdominal area, so it cannot\ncharacterize the spatial patterns of peristalsis on the uterine\nsurface. A new method is needed to noninvasively and\nquantitatively de fine the detailed features of human uterine\nperistalsis over the entire uterus for an extended period of time.\nWe recently developed an electrophysiological imaging system\ncalled electromyometrial imaging (EMMI)\n31–34 to quantitate the\nelectrical activity underlying uterine contractions during labor.\nHerein, we adapted EMMI to develop a new uterine peristalsis\nimaging (UPI) system and used it to longitudinally image the four-\ndimensional (4D) electrical activation patterns of uterine peristalsis\nover each phase of the menstrual cycle in healthy, nonpregnant\nparticipants with normal menstrual cycles. With UPI, we provide\nquantitative evidence that uterine peristalsis changes in\n1Department of Electrical and Systems Engineering, Washington University in St. Louis, St. Louis, MO 63130, USA. 2Center for Reproductive Health Sciences, Washington\nUniversity in St. Louis, St. Louis, MO 63108, USA. 3Department of Obstetrics & Gynecology, Washington University in St. Louis, St. Louis, MO 63108, USA. 4Division of Reproductive\nEndocrinology & Infertility, Washington University in St. Louis, St. Louis, MO 63108, USA. 5Mallinckrodt Institute of Radiology, Washington University in St. Louis, St. Louis, MO\n63110, USA. 6Department of Biomedical Engineering, Washington University in St. Louis, St. Louis, MO 63130, USA. 7These authors contributed equally: Sicheng Wang, Kelsey\nAnderson. ✉email: valerie.ratts@wustl.edu; wangyong@wustl.edu\nwww.nature.com/npjwomenshealth\n1234567890():,;\n\nfrequency, direction, duration, magnitude, and power throughout\nthe menstrual cycle.\nRESULTS\nStudy participants\nWe enrolled 26 non-pregnant pre-menopausal women between\n18 and 40 years of age who were not using hormonal medication\nand had regular menstrual cycles between 25 and 34 days in\nlength. Demographic and gynecologic history data of enrolled\nparticipants are shown in Table 1.\nUterine peristalsis imaging (UPI) system\nThe UPI system is illustrated in Fig. 1. First while applying MRI-\ncompatible fiducial markers around the abdomen and lower back,\na participant undergoes a one-time, fast, anatomical MRI scan (Fig.\n1A) to acquire the patient-speci fic uterus-body surface geometry\n(Fig. 1B, C). Second, at each phase of the menstrual cycle,\ncustomized wearable pin-type electrode patches are applied to\nthe same locations on the body surface as the MRI fiducial markers\n(Fig. 1D). Body surface electrical signals (Fig. 1E) are recorded for\n20–30 min, and electrical signals (Fig. 1F) are recorded with a\nband-pass filter (0.01–0.05 Hz)\n19,35,36. Third, UPI software is used to\nmap electrical signals to each point on the entire three-\ndimensional (3D) uterine surface (Fig. 1G, H). These electrical\nsignals are used to derive activation sequences (Fig. 1I).\n4D spatial-temporal quanti fication of uterine peristalsis\npatterns\nUterine isochrone maps (Fig. 2A) were generated according to the\nactivation sequence (Fig. 1I). Uterine surface data were analyzed to\ngenerate uterine magnitude maps (Fig. 2B). Based on uterine\nisochrone maps, we de fined the peristalsis wave direction (C –F,\nF–C, or others) and locations of the initiation and termination sites\n(cervix area, fundus area, and other areas) for each peristalsis\nwave. We then summarized the data for all peristalsis waves\nrecorded during the entire imaging session (Fig. 2C–E) and\ngenerated distribution probability maps of the initiation (Fig. 2F)\nand termination (Fig. 2G) sites for C –F and F –C peristalses for the\nentire imaging session. See detailed descriptions in the Methods\nsection.\nUPI findings in a healthy participant with regular\nmenstrual cycles\nIn Fig. 3, we present representative uterine peristalsis waves of a\n32-year-old healthy participant. During the menses phase, 40.4%\nof waves traversed from near the fundus toward the cervix (F –C),\nand 25.5% traversed from near the cervix toward the fundus (C –F)\n(Fig. 3A). During the proliferative phase, 42.7% of waves were F –C\nand 36.5% were C –F (Fig. 3B). During the ovulatory phase, 45.2%\nof waves were C –F, and 33.3% were F –C (Fig. 3C). In the secretory\nphase, 41.9% of waves were C –F, and 32.6% were F –C (Fig. 3D). In\nall cases in which we were able to determine the direction of\nperistalsis in TVUS images ( n = 88/94, 93.6% of waves), the\ndirection of peristalsis imaged by UPI matched the direction\nobserved by TVUS.\nUterine peristalsis wave features differ by menstrual cycle\nphase in healthy participants\nWe used the UPI system to image uterine peristalsis in 26 healthy\nnonpregnant females with regular menstrual cycles. Electrical\nrecording was conducted during each menstrual cycle phase, and\nall data for each phase from all 26 participants were compiled. We\nfirst plotted the frequency of peristalsis waves by days in a\nstandard 28-day menstrual cycle (Fig.\n4A). On average, frequency\npeaked around day 15 (during the periovulatory phase). Next, we\nplotted the average peristalsis wave frequency (Fig. 4B) during\neach phase and saw that the highest frequency (3.13 [2.72, 3.54])\noccurred in the periovulatory phase, and the lowest frequency\n(1.43 [1.21, 1.67]) occurred in the menses phase. Plotting the\naverage direction ratios revealed that the most common wave\ndirections were C –Fo rF –C (Fig. 4C–E). Plotting the average\nmagnitude (Fig. 4F–H) revealed that the C –F and F –C waves had\nhigher magnitudes than the other direction waves. The highest\nmagnitude C –F waves were in the periovulatory phase (Fig. 4F),\nand the highest magnitude F –C waves were in the menses phase\n(Fig. 4G). Finally, we plotted average power (Fig. 4I–K) and found\nthat the highest power C –F waves were during the periovulatory\nphase (Fig. 4I) and that the highest power F –C waves were during\nthe menses phase (Fig. 4J).\nAsymmetric C–F peristalsis direction correlates with dominant\nfollicle in the periovulation phase\nWe noticed that C –F peristalsis waves during the peri-ovulatory\nphase tended to move preferentially toward one fallopian tube. In\nFig. 5A–C, we present data from one participant whose dominant\nfollicle was on the right side. Figure 5A displays a T2-weighted\nanatomical scan focusing on the pelvic region and reproductive\nTable 1. Demographics of enrolled patients ( N = 26) with regular menstrual cycles.\nAge, years 27.75 (25.96 –29.54)\nBody mass index, kg/m 2 29.99 (26.82 –33.18)\nRace, n (%)\nWhite 14 (53.8%)\nBlack 9 (34.6%)\nAsian 2 (7.7%)\nOther 1 (3.8%)\nCycle length, days 28.35 (27.46 –29.24)\nPhase Menses Proliferative Ovulatory Secretory\nEstradiol (pg/mL) 33.7 (27.0 –40.4) 93.9 (76.7 –111.1) 155.9 (119.3 –192.6) 118.9 (99.7 –138.2)\nProgesterone (ng/mL) 0.28 (0.21 –0.35) 0.20 (0.20 –0.21) 2.00 (1.20 –2.80) 8.34 (6.52 –10.16)\nEndometrial thickness (mm) 3.50 (2.66 –4.34) 6.79 (5.94 –7.64) 8.92 (7.56 –10.28) 9.43 (8.28 –10.58)\nMean value and 95% con fidence interval are shown for each numerical variable.\nS. Wang et al.\n2\nnpj Women’s Health (2024)     1 \n1234567890():,;\n\norgans, and Fig. 5B shows the reconstructed 3D uterus geometry.\nFigure 5C illustrates sequential isochrone maps of asymmetric C –F\nperistalses, with red indicating the initiation sites and blue\nindicating the termination sites. All three of these peristalsis\nwaves traveled toward the right follicle.\nFor 15 participants, we were able to determine by TVUS which\novary had a dominant follicle. The clinical characteristics and\nmeasurements of these patients are summarized in Table 2. For\neach of these participants, we created maps illustrating the\nprobability of C–F peristalsis waves during the periovulatory phase\nterminating in a particular region. Figure 5D exhibits the\ntermination probability maps of five participants with right-sided\ndominant follicles. Figure 5E displays the termination probability\nmap of ten participants with left-sided dominant follicles. In most\ncases, the most common termination sites (red in the maps) were\non the side closest to the ovary containing the dominant follicle.\nDISCUSSION\nThe UPI system presented here can noninvasively and objectively\nprovide detailed information about the electrical activation\npatterns of uterine peristalsis at high spatial and temporal\nresolution. In all phases of the menstrual cycle, the most common\ndirections of uterine peristalsis waves were C –F and F –C, but we\nFig. 1 Schematic of uterine peristalsis imaging system. Anatomical MRI ( A) followed by segmentation ( B) yields a patient-speci fic geometry\nof the body surface, uterus surface, and fallopian tubes ( C). Yellow dots indicate positions of MRI-compatible markers. D Electrode patches are\nplaced on the participant ’s abdomen and back in the same positions as the markers. Raw ( E) and filtered ( F) electrical signals (bandwidth:\n0.01–0.05 Hz). Uterine surface electrical signals from ( G) one uterine surface point near the cervical region (white plus sign in ( I)) and ( H) one\nuterine surface point around the fundal region (white asterisks in ( I)). Red dots denote the points of steepest negative slope to represent the\nactivation times during peristalsis cycles. I Detailed activation sequence of one complete uterine peristalsis wave initiated near the fundus and\nterminated near the cervix. Blue indicates inactive uterine regions, and red indicates active uterine regions.\nS. Wang et al.\n3\nnpj Women’s Health (2024)     1 \n\nalso observed other complex wave patterns as have been noted\npreviously12,13. In these normal participants, the predominant\nperistalsis pattern in menses was F –C. Others have seen this\npattern in TVUS and postulated that it facilitates expulsion of\nmenstrual and endometrial tissue while protecting against\nascending pathogens\n37. Inefficient F–C peristalsis waves may lead\nto accumulation of endometrial tissues in the uterine cavity and\nincrease the risk of developing endometriosis.\nWe observed no dominant contraction pattern during the\nproliferative phase. Prior studies have shown that, in the late\nproliferative phase, the predominant peristalsis direction switches\nfrom F –Ct oC –F. We only performed electrical recording on\nparticipants at one time in the proliferative phase and thus likely\nincluded data from both before and after this switch 38.\nIn the peri-ovulatory phase, the predominant peristalsis pattern\nwas C –F. Kunz et al. used serial hysterosalpingography to follow\nlabeled macrospheres the size of sperm and observed that they\nwere transported from the cervix into the uterus and fallopian\ntubes\n39, suggesting that the C –F peristalsis pattern facilitates the\ntransport of sperm toward the oocyte. Consistent with this idea,\nwe observed that C –F peristalsis waves commonly traveled in the\ndirection of the dominant follicle.\nFinally, in the secretory phase, we observed both C –F and F –C\nperistalsis waves, with neither predominating. This is consistent\nFig. 2 Quantification of uterine peristalsis waves. A Uterine isochrone maps from the same uterine peristalsis wave as in Fig. 1I. Warm and\ncool colors represent early and late activation, respectively. The white arrow depicts the peristalsis wave propagation direction. B Uterine\nmagnitude map from the same uterine peristalsis wave in Fig. 1I, showing the magnitude distribution over the entire 3D uterine surface in one\nperistalsis wave. C Distribution of uterine peristalsis directions (C –F, F–C, others) and initiation ( D) and termination ( E) sites (cervix, fundus, and\nother areas) analyzed from the entire electrical recording. Initiation ( F) and termination ( G) site distribution probability maps of C –F and F –C\nuterine peristalses for the entire imaging session (see details in Materials and Methods).\nS. Wang et al.\n4\nnpj Women’s Health (2024)     1 \n\nwith work by Fanchin et al. 40 evaluating peristalsis waves at the\ntime of embryo transfer after in vitro fertilization.\nUPI has several technical advantages over other modalities used\nto image uterine peristalsis. First, UPI is noninvasive, which is\noptimal for long-duration uterine monitoring. Whereas we\ncontinuously recorded for up to 30 min, researchers using other\nmodalities usually recorded for 5 min and at most 15 min. We are\ncurrently developing wearable electrode sensors, which would\nallow researchers to monitor uterine peristalsis for hours or days.\nLonger recordings will likely yield more accurate assessment of\nuterine peristalsis waves and permit identi fication of more\ncomplex wave patterns. Second, modalities using invasive\nmonitoring may perturb peristalsis. For example, the TVUS probe\ntouches the cervix, which could alter peristalsis. Third, UPI\nprovides high spatial and temporal resolution and coverage and\ncan characterize the complex patterns of electrical activation\nacross the entire uterus. Thus, we are able to objectively measure\nthe initiation sites, direction, frequency, and duration of uterine\nperistalsis waves. Finally, UPI data re flect participant-speci fic\nuterine-body anatomy uterine peristalsis patterns.\nOne limitation of our study is that a few enrolled participants\nmay have had undiagnosed gynecologic pathology. We mini-\nmized this possibility by collecting detailed obstetric and\ngynecologic history from prospective participants during the\nscreening process and excluding anyone with a history of uterine\nanomalies, infertility, ovulatory dysfunction, medication use,\nendometriosis, or documented fibroids greater than 3 cm. Never-\ntheless, future studies could expand the sample size and enroll\nmore patients with recent gynecologic exams.\nA second limitation is that the inverse calculations in the UPI\nsoftware assume that the medium is homogeneous between the\nuterine surface and abdominal surface without any primary\nelectrical source. However, the uterus in a non-pregnant person\nis near the bladder and bowel, both of which contain smooth\nmuscles that generate slow-wave signals. Electrical signals from\nthese organs are unlikely to interfere with our measurements for\nthree reasons. First, the rhythmic phasic contractions are 0.33 Hz in\nhuman stomach, about 0.083 Hz in duodenum, and\n0.125–0.167 Hz in ileum. As our band-pass filter selected activity\nbetween 0.01 and 0.05 Hz, none of these signals affected our\nanalysis. Second, the detrusor muscle in the human bladder\ngenerates low-amplitude rhythmic contractions at\n0.033 ± 0.008 Hz. Although this is within the frequency range for\nuterine peristalsis, participants were asked to empty their bladders\nbefore the study. Thus, the detrusor muscle was likely relaxed in\nour participants. Third, the electrical amplitude of slow-wave\ncontractions in human bladder is low. Moreover, we validated over\n90% of uterine peristalsis wave directions in simultaneously\ncollected TVUS videos in each participant.\nWe are pursuing multiple avenues to further develop UPI. We\naim to replace the current short anatomical MRI with a low-cost\n3D ultrasound measurement to generate patient-speci fic body-\nuterus geometry. We also aim to re fine the electrode placement to\nimprove imaging accuracy and develop automatic or semi-\nautomatic peristalsis wave classi fication. Finally, we are developing\nlow-cost wearable electrode sensors\n41,42. Together, these technical\nadvances will reduce the cost of the UPI system and make it easier\nto implement in a variety of settings.\nFig. 3 Representative uterine peristalsis imaging (UPI) outcome in one healthy participant with normal menstrual cycles during four\nphases of the menstrual cycle. Dominant F–C uterine peristalsis pattern during the ( A) menses phase and ( B) proliferative phase. C –F uterine\nperistalsis patterns during the ( C) peri-ovulatory phase and ( D) secretory phase. In each panel, A represents the anterior view and P represents\nthe posterior view of the uterus. White asterisks denote initiation sites, white plus signs denote termination sites, red arrows indicate F –C\nperistalsis waves, and green arrows indicate C –F peristalsis waves.\nS. Wang et al.\n5\nnpj Women’s Health (2024)     1 \n\nIn future larger studies, UPI can be used to establish reference\nranges for features of uterine peristalsis in normal menstrual\ncycles. These normal reference ranges could be used to identify\npatients with abnormal gynecological conditions such as endo-\nmetriosis, ovulatory dysfunction, uterine anomalies, abnormal\nuterine bleeding, and infertility. Moreover, with the detailed 4D\nelectrical activation patterns imaged by UPI, it will be possible to\nlongitudinally evaluate the effects of various clinical interventions\nand optimize treatment plans for individual patients. In the long\nterm, UPI may facilitate development of nonpharmaceutical\nstrategies to electrically correct abnormal uterine peristalsis\nunderlying gynecological conditions such as endometriosis. This\napproach would be similar to how cardiac pacemakers are now\nused to treat heart conditions.\nMETHODS\nEthics and participant enrollment\nThis study was performed in the Division of Reproductive\nEndocrinology & Infertility and Center for Outpatient Health at\nWashington University School of Medicine. This study was\napproved by the Washington University Institutional Review\nBoard, and all participants signed informed consent documents.\nParticipants were included if they were female at birth, between\nthe ages of 18 and 37 years, and had regular, predictable\nmenstrual cycles every 24 –35 days. Potential participants were\nexcluded if they were post-menopausal, pregnant, or breastfeed-\ning; had a uterine anomaly; had exposure to medications known\nto affect uterine contractility (e.g., magnesium, opioids, beta\nantagonists, nifedipine); were non-English speaking; had abdom-\ninal circumference >55 cm; had MRI contraindications (pacemaker,\nmetal implants, etc.); or had documented or self-reported histories\nof infertility, ovulatory dysfunction, or endometriosis. Each\nparticipant was imaged with the UPI system four times during\none menstrual cycle, once each during menses, early proliferative,\nlate proliferative (peri-ovulatory), and secretory phases.\nDetermination of phase in the menstrual cycle\nPatients were determined to be in one of four menstrual phases\n(menses, early proliferative, late proliferative, and secretory) by\nusing a combination of patient-reported bleeding, cycle length,\nultrasound findings, ovulation predictor kit (Clearblue, Geneva,\nSwitzerland) results, and hormonal measurements. Serum\n(5–10 ml) was collected and sent to the Core Laboratory for\nClinical Studies at Washington University in St. Louis to measure\nconcentrations of estradiol, progesterone, and testosterone. The\nmenses phase was assigned when a patient reported bleeding.\nThe early proliferative phase was assigned after the patient had\nstopped bleeding, ultrasound demonstrated early follicular activity\n(largest average follicle dimension <16 mm), serum estradiol\n<200 pg/ml, and serum progesterone <3 ng/ml. The late prolif-\nerative (peri-ovulatory) phase was de fined by a positive result on\nan ovulation predictor kit, serum estradiol >200 pg/ml, serum\nprogesterone <3 ng/ml, and/or a dominant follicle on ultrasound\n(largest average follicular dimension ≥16 mm). The secretory\nphase was assigned when serum progesterone was >3 ng/ml.\nFig. 4 Quantitative analysis of uterine peristalsis during each phase of the menstrual cycle. A Uterine peristalsis frequency plotted by day\nin a 28-day menstrual cycle. Red curve was fitted by using Gaussian distribution. B Peristalsis frequency during the four phases.\nC, D, E Direction ratio of peristalsis waves in C –F, F–C, and other directions. F, G, H Magnitude of uterine peristalsis waves in C –F , F-C, and other\ndirections. I, J, K Power of peristalsis waves in C –F, F–C, and other directions. N = 26 participants in each phase. Error bars in ( A) depict the 95%\nconfidence interval. Column bars in ( B–K) are shown as mean values with 95% con fidence interval. Kruskal-Wallis test was performed to\nanalyze the difference of each UPI parameter between menstrual cycle phases. P-values less than 0.05 are marked.\nS. Wang et al.\n6\nnpj Women’s Health (2024)     1 \n\nUterine peristalsis imaging (UPI) procedure\nFirst, a woman underwent a one-time, anatomical (T1W sequence)\n3T Siemens Prisma MRI scan (~10 min) to acquire the patient-\nspecific uterus-body surface geometry while wearing up to 10\npatches containing up to 128 MRI-compatible fiducial markers\naround the abdomen and lower back (Fig. 1A). 3D uterus\nsegmentation was performed in T1-weighted anatomical images\nusing the software Amira 6.2.0 with the standard image\nprocessing procedure. Uterus and body geometry were gener-\nated, and electrode locations were identi fied on the MRI\nanatomical images (Fig. 1 B, C). Second, customized BioSemi\npin-type electrode patches were applied to the same locations on\nthe body surface as the MRI fiducial markers. An ADC box was\nused to record the body surface electrical signals (Fig. 1D, E) for\n20 min. Third, the participant underwent another 10-min electrical\nrecording while simultaneously undergoing transvaginal ultra-\nsound (TVUS). TVUS scans of the uterus were performed by a\nsonographer holding the transducer probe while the patient was\nlying in a lithotomy position. After obaining anatomical images to\ninform menstrual phase determination, cine clips were recorded\non a GE Voluson S8 ultrasound machine. The duration of each clip\nwas 20 s and 30 –35 clips were acquired, on average.\nFig. 5 Uterine peristalsis patterns during periovulation in participants with identi fied dominant follicles. A T2-weighted anatomical image\nof one participant (r1) with right dominant follicle. Blue segment is the uterine cavity and red segments are the follicles. B Reconstructed 3D\nuterine geometry from the participant in ( A) with two interstitial portions marked by red stars. C Three selected sequential isochrones maps\nfrom the participant in ( A). Red indicates the initiation site, blue indicates the termination site, and the orange arrow indicates the peristalsis\nwave direction. Termination probability maps of five participants with right dominant follicle ( D), and ten with left dominant follicle ( E). All\nimages show anterior views of the uterus.\nS. Wang et al.\n7\nnpj Women’s Health (2024)     1 \n\nHigh-density electrode patches\nSpecifically, the distance between electrodes is less than 1 cm,\nwhereas it is more than 2 cm in the patches used for pregnant\nwomen. The electrode spatial density used to image the mild\nuterine peristalsis of the small nonpregnant uterus is more than\nfour times higher than the version used to image the labor\ncontractions from the large pregnant uterus. These high-density\nelectrodes patches were placed around the participant ’s abodo-\nmen (front, back, and sides) to comprehensively capture the\nelectrical signals from the uterine peristalses during menstrual\ncycles.\nTo enable the same patch placement across multiple visits, we\nemployed multiple anatomical landmarks on the patient body\nsurface to place adhesive rulers on patients ’ abdomen surface. The\nsame electrode patches are placed at the same location along the\nrulers. As shown in Supplementary Fig. 1, four color-marked rulers\nare applied to the patient ’s front and back to establish the spatial\ncoordinate system. There are a total of ten patches: four on the\nback, four on the front, and one on each side. Patches A1, A2, B1,\nB2 are placed on the patient ’s lower abdominal area. Patches C1\nand C2 are placed on the patient ’s left and right sides of the\nabdomen. Patches D1, D2, D3, and D4 are placed on the patient ’s\nlower back. Speci fically, we placed two adhesive rulers on the\nfront (Supplementary Fig. 1A). One red ruler ran vertically along\nthe umbilicus with the inferior end of the ruler ending at the\nsuperior edge of the patient ’s public symphysis. The inferior edge\nof the lower stomach patches (B1 & B2) are placed in line with the\ninferior end of the red vertical ruler. The patches are placed with\nthe medial edge running along the red vertical ruler. Then, the\ngreen horizontal ruler is placed along the superior edge of the\nlower patches. The upper patches (A1 & A2) are placed above the\nhorizontal ruler. Again, the patches are placed with the medial\nedge running along the vertical ruler. The side patches (C1 & C2)\nare placed with the inferior edge aligned with the top of the\nhorizontal ruler from the front. Similarly, we placed two adhesive\nrulers on the back (Supplementary Fig. 1C). One organe ruler runs\nvertically along the spine with the inferior end of the ruler ending\nat the patient ’s coccyx. The purple horizontal ruler is placed at the\ntop of the patient ’s hip bones. The patches (D1, D2, D3, and D4)\nare placed above the horizontal ruler. The medial patches are\nplaced along the edge of the vertical ruler and the lateral patches\n(C1 & C2) are placed along the edge of the medial patches. The\ncoordinates of each patch are also recorded in the longitudinal\nstudies to further ensure the consistent placement of patches\nacross multiple visits and avoid its in fluence on UPI results.\nSignal processing\nIn experiments in which electrical signals were directly measured\non the human nonpregnant uterine surface, the median uterine\nperistalsis frequency was 0.039 Hz in the proliferative phase and\n0.020 Hz in the secretory phase\n29. In TVUS and cine MRI studies 1,\nthe frequency of uterine peristalsis is between 0.33 and 6\ncontractions per minute throughout the cycle. Therefore, we\nselected a frequency band between 0.01 and 0.05 Hz 19,35,36 to\nminimize the high-frequency artifacts not correlated with\nelectrical activities of uterine peristalsis. The body surface electrical\nsignals were processed with a band-pass filter to generate wave\nelectrical signals (peristalsis waves) over the entire abdomen\nsurface (Fig. 1F).\nInverse computation in UPI\nWith the electro-quasi-static assumption of the bioelectric field,\nthe inverse computation combines the patient-speci fic uterus-\nabdomen surface and electrical potentials measured on the\nabdominal surface to reconstruct the potential distribution over\nthe entire 3D uterine surface. We assume that the medium is\nhomogeneous between the uterine surface and abdominal\nsurface without any primary electrical source. Then, the inverse\nproblem could be mathematically described by the Cauchy\nproblem for Laplace ’s Eq. ( 1) with boundary conditions (2, 3) on\nthe abdominal surface.\n∇\n2ϕ xðÞ ¼ 0 (1)\nDirichlet (2) and Neumann (3) conditions for the abdominal\nsurface potentials are:\nϕ xðÞ ¼ ϕAxðÞ ; x 2 ΓA (2)\n∂ϕ xðÞ\n∂n ¼ 0; x 2 ΓA (3)\nHere, n is the normal vector on the abdominal surface at\nlocation x and ΓA represents abdominal surface. ϕAxðÞ is the\npotential measured on the abdominal surface and ϕ xðÞ is the\npotential on the uterine surface.\nTable 2. Clinical characteristics and measurements in 15 participants with dominant follicles detected on ultrasound.\nPatient ID Age, years Body mass index, kg/m 2 Gravidity Cycle length, days Dominant follicle Diameters of largest average follicle (mm)\nr1 32 48.7 0 28 Right 22.2, 18.5\nr2 32 35.93 2 28 Right 19.9, 17.8\nr3 32 24.51 0 28 Right 23.5, 15.3\nr4 36 28.01 4 28 Right 21.0, 13.6\nr5 25 27.96 0 28 Right 19.9, 19.1\nl1 35 28.3 0 28 Left 19.9, 19.1\nl2 25 26.73 0 27 Left 26.6, 22.6\nl3 31 19.79 1 28 Left 23.8, 17.0\nl4 30 38.4 0 31 Left 14.4, 18.7\nl5 23 22.33 0 28 Left 23.8, 20.8\nl6 26 26.93 0 28 Left 20.8, 18.6\nl7 24 24.2 0 26 Left 22.0, 20.7\nl8 29 34.67 3 27 Left 18.1, 12.4\nl9 33 27.64 0 25 Left 11.4, 21.2\nl10 21 22.24 0 27 Left 22.4, 18.1\nS. Wang et al.\n8\nnpj Women’s Health (2024)     1 \n\nAs a mesh-free method robust to noise, a method of\nfundamental solutions 43 was deployed to discretize the Laplace ’s\nequation and boundary conditions, which is accurate for solving\nthe bioelectric field inverse problem in both electrocardiographic\nimaging\n43 and EMMI 31,33,34 systems. This problem cannot be\nsolved directly as it is an ill-posed inverse problem. Therefore,\nTikhonov-based inverse computation with a fixed regularization\nvalue of 0.01 was used to obtain the solution.\nΦ\nA ¼ A Φ U (4)\nHere, Φ A is a M * T matrix of measuring surface potentials, Φ U is\na N * T matrix of uterine surface potentials, where M is the number\nof measuring electrodes applied on the abdominal surface and N\nis the number of discrete points on the uterine surface, and T is\nthe number of recording time points. A is a M * N linear transform\nmatrix encoding the relationship between abdominal surface\npotential Φ A and uterine surface potential Φ U.\nUPI data processing\nThe inverse computation described above was employed to\ncompute the uterine surface electrical signals (Fig. 1G, H) on the\n3D uterine surface. The times when the uterine surface electrical\nsignals at various uterine surface areas reached the steepest\nnegative slope\n44–48 were extracted and de fined as electrical\nactivation times at those uterine areas during peristalsis waves\n(red dots in Fig. 1G, H). During each peristalsis wave, sequential\ntime frames were generated as the activation sequences (Fig. 1I)\nto re flect the detailed 4D spatial-temporal activation patterns of\nthe uterine peristalsis. Within each time frame, the red region\nindicated electrically activated myometrium areas currently\nexperiencing peristalsis, and the blue region indicated inactive\nareas of the uterus. The isochrone map was generated as a color-\ncoded 3D map to summarize the electrical activation sequence\n(Fig. 1I). In the isochrone map, warm and cool colors denote\nregions of the uterus that activated early and late, respectively,\nduring the peristalsis wave.\nInspection of uterine peristalsis direction\nUterine peristalsis direction was categorized according to the\nwave classi fication system proposed by Gestel et al.\n12 (Table 3). A\ncustomized UPI post-analysis software with graphical user inter-\nface was developed in MATLAB (R2021b) to visualize each\nperistaltic wave. The software first detected the uterine peristalsis\nwaves according to the electrical activations and recorded the\nstart and end times. Next, five independent observers visually\ninspected the electrical activation sequences and isochrone maps\nto de fine the direction, initiation, and termination sites of each\nuterine peristalsis wave. They got the same training on how to\nunderstand and read UPI images and passed the test using a\nsample dataset before reviewing the data in this work. The\ninterobserver agreement evaluation was performed in 1240\nuterine peristalses imaged using UPI from the first five participants\nacross four phases in this study, the interobserver agreement was\nsubstantial with 0.93 intraclass correlation coef ficients (ICC) in\nclassifying UP directions, which indicates the great reliability and\nrobustness across different observers. Two observers ( A, B) had\nbeen intensively involved in research on ultrasound and MRI of\nnonpregnant uterus and are familiar with the topic. The other\nthree observers ( C, D, E) were biomedical engineers. All observers\nreceived the same instructions on how to assess the endometrial\nwaves. All UPI activation movies were masked for patients ’ name,\ndemographics, OBGYN history, and menstrual phase. All UPI\nvideos were independently inspected by observers C, D, and E.I f\nthe observers disagreed on the direction or initiation or\ntermination site of a uterine wave, A and B examined the movie\nand made the final call. Next, the software automatically\ncalculated the duration, magnitude, and power of each uterine\nperistalsis wave. Finally, we performed statistical analysis of the\nuterine peristalsis wave frequency, direction ratio, and mean value\nof duration, magnitude, and power.\nInspection of TVUS images\nThree registered sonographers were invovled in the independent\nreviews (without knowledge of the UPI results) of the TVUS\nrecordings to determine the uterine peristalsis direction. They are\nspecialized in the field of OBGYN, with experience in watching\nTVUS images, and one of them had previously performed visual\ninspection of contractions for research. Only segments where all\nthree sonographers reached a 100% agreement on the UP\ndirection were utilized to validate our UPI findings acquired\nsimultaneously during TVUS.\nElectrophysiological characterization and quanti fication\nThree UPI electrophysiological indices were de fined to qualita-\ntively and quantitatively describe uteirne peristalsis patterns.\nDuration (Sec.) was de fined as the duration of a complete\nperistalsis wave measured in the isochrone map (Fig. 2A) of the\nuterine peristalsis wave. Magnitude (mV) was de fined as the\naverage peak amplitude of electrical potential over the uterine\nregion experiencing activation during the entire peristalsis wave. A\nmagnitude map (Fig. 2B) was developed to present the magnitude\ndistribution over the entire 3D uterine surface in one peristalsis.\nPower (mV*sec) was de fined as the product of magnitude and\nduration for each uterine peristalsis wave.\nSpatial and temporal analysis of human uterine peristalsis\nFrequency was determined by counting the number of uterine\nperistalsis waves detected during the recording session and\ndividing it by the total imaging time. To analyze the compositions\nof uterine peristalsis propagation direction, initiation, and\ntermination sites (Fig. 2C–E), we calculated the number of\nperistalsis waves with speci fic propagation directions (F –C, C –F,\nor other) and initiation and termination sites (cervical, fundal, or\nother regions). These counts were then divided by the total\nTable 3. Endometrial wave classi fication system initially proposed by Ijland 11 and revised by Gestel 12.\nWave type Wave symbol De finition\nCervix–Fundus C –F Wave propagates from cervix to fundus\nFundus–Cervix F –C Wave propagates from fundus to cervix\nOthers Alternating Wave propagates from cervix to fundus with an alternating wave from fundus to cervix\nRecoiling Wave propagates from cervix to fundus followed by a re flective wave toward cervix\nStanding Visible wave with no propagation toward cervix or fundus\nOpposing Wave start at cervical and fundal uterine regions simultaneously\nRandom Waves start at multiple sites on the uterus\nS. Wang et al.\n9\nnpj Women’s Health (2024)     1 \n\nnumber of peristalsis waves observed during the 30-min electrical\nmapping session.\nThe direction ratio (Fig. 2C) represents the percentage of\nperistalsis waves occurring in each direction out of the total\nnumber of peristalsis waves observed. Initiation and termination\nsites were identi fied as the regions where uterine peristalsis\nstarted and ended, respectively, based on the activation\nsequences. These sites were categorized into three groups:\nCervical region, Fundal region, and Other regions. The initiation\n(termination) site composition (Fig. 2D–E) denotes the percentage\nof peristalses initiated (terminated) in each uterine region\n(cervical, fundal, or other region) throughout the entire recording.\nTo determine the initiation or termination probability for each\npoint in each direction, we calculated the relative frequency\nmeasurement (ranging from 0 to 1) by dividing the number of\nperistalsis waves initiated or terminated at a speci fic point by the\ntotal number of peristalsis waves in that direction during the\nimaging session. This allowed us to generate spatial probability\nmaps of initiation or termination sites (Fig. 2F–G) on the uterine\nsurface.\nStatistical analysis\nBaseline demographic and OBGYN history characteristics of\npatients were summarized by using frequencies and percentages\nfor categorical variables and means (95% con fidence interval) for\nOBGYN history, ovarian follicles, and hormone measurements.\nThe primary outcomes of each uterine peristalsis wave were one\nqualitative variable (direction) and three quantitative variables:\nduration (sec), magnitude (mV), and power (mV*sec). UPI-indexed\nparameters were calculated according to directions (C –F, F–C, and\nothers) using the mean value of UPI measurements for each\npatient in each visit within the standard 30-min time window.\nKruskal-Wallis test was performed to analyze the difference of\neach UPI parameter between menstrual cycle phases. P < 0.05 was\nconsidered statistically signi ficant.\nDATA AVAILABILITY\nThe data reported in this work are available from the corresponding author upon\nreasonable request.\nReceived: 21 July 2023; Accepted: 6 December 2023;\nREFERENCES\n1. Kuijsters, N. P. M. et al. Uterine peristalsis and fertility: current knowledge and\nfuture perspectives: a review and meta-analysis. Reprod. BioMed. Online 35,5 0–71\n(2017).\n2. de Vries, K., Lyons, E. A., Ballard, G., Levi, C. S. & Lindsay, D. J. Contractions of the\ninner third of the myometrium. Am. J. Obstet. Gynecol. 162, 679 –682 (1990).\n3. Lyons, E. A. et al. Characterization of subendometrial myometrial contractions\nthroughout the menstrual cycle in normal fertile women. Fertil. Steril. 55, 771–774\n(1991).\n4. Bulletti, C. et al. Uterine contractility during the menstrual cycle. Hum. Reprod. 15,\n81–89 (2000).\n5. Kunz, G. & Leyendecker, G. Uterine peristaltic activity during the menstrual cycle:\ncharacterization, regulation, function and dysfunction. Reprod. Biomed. Online 4,\n5–9 (2002).\n6. Kuijsters, N. P. M. et al. Validation of electrohysterography for uterine peristalsis in\nnon-pregnant uteri. Fertil. Steril. 100, S383 (2013).\n7. Kuijsters, N. P. M. et al. Propagation of spontaneous electrical activity in the\nex vivo human uterus. Pflugers Arch. Eur. J. Physiol. 472, 1065 –1078 (2020).\n8. Bulletti, C. et al. Abnormal uterine contractility in nonpregnant women. in Annals\nof the New York Academy of Sciences 828 223–229 (1997).\n9. Bulletti, C. & de Ziegler, D. Uterine contractility and embryo implantation. Curr.\nOpin. Obstet. Gynecol. 17, 265 –276 (2005).\n10. Duquette, R. A. et al. Vimentin-positive, c-KIT-negative interstitial cells in human\nand rat uterus: a role in pacemaking? Biol. Reprod. 72, 276 –283 (2005).\n11. Ijland, M. M. et al. Endometrial wavelike movements during the menstrual cycle.\nFertil. Steril. 65, 746 –749 (1996).\n12. van Gestel, I., IJland, M. M., Evers, J. L. H. & Hoogland, H. J. Complex endometrial\nwave-patterns in IVF. Fertil. Steril. 88, 612 –615 (2007).\n13. Kuijsters, N. P. M. et al. Visual inspection of transvaginal ultrasound videos to\ncharacterize uterine peristalsis: an inter-observer agreement study. J. Ultrasound\n23,3 7 –44 (2019).\n14. Van Gestel, I., Ijland, M. M., Hoogland, H. J. & Evers, J. L. H. Endometrial wave-like\nactivity in the non-pregnant uterus. https://doi.org/10.1093/humupd/dmg011.\n15. Ijland, M. M., Volovics, L., Evers, J. L. H., Hoogland, H. J. & Dunselman, G. A. J.\nRelation between endometrial wavelike activity and fecundability in spontaneous\ncycles. Fertil. Steril. 67\n, 492 –496 (1997).\n16. Gora, S., Elad, D. & Jaffa, A. J. Objective analysis of vaginal ultrasound video clips\nfor exploring uterine peristalsis post vaginal and cesarean section deliveries.\nReprod. Sci. 25, 899 –908 (2018).\n17. A, N. et al. Uterine peristalsis: comparison of transvaginal ultrasound and two\ndifferent sequences of cine MR imaging. J. Magn. Reson. Imaging 20, 463 –469\n(2004).\n18. Forman, E. J. et al. Single embryo transfer with comprehensive chromosome\nscreening results in improved ongoing pregnancy rates and decreased mis-\ncarriage rates. Hum. Reprod. 27, 1217 –1222 (2012).\n19. Meirzon, D., Jaffa, A. J., Gordon, Z. & Elad, D. A new method for analysis of non-\npregnant uterine peristalsis using transvaginal ultrasound. Ultrasound Obstet.\nGynecol. 38, 217 –224 (2011).\n20. Huang, Y. et al. Characterization of uterine peristaltic waves by ultrasound strain\nanalysis. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 69, 2050 –2060 (2022).\n21. Huang, Y. Ultrasound-based strain mapping for quantitative characterization of\nuterine peristalsis outside pregnancy. at https://research.tue.nl/en/publications/\nultrasound-based-strain-mapping-for-quantitative-characterization (2020).\n22. Sammali, F. et al. Dedicated ultrasound speckle tracking for quantitative analysis\nof uterine motion outside pregnancy. IEEE Trans. Ultrason. Ferroelectr. Freq. Con-\ntrol 66, 581 –590 (2019).\n23. Blank, C. et al. Assessment of uterine activity during IVF by quantitative ultra-\nsound imaging: a pilot study. Reprod. Biomed. Online 41, 1045 –1053 (2020).\n24. Rees, C. O. et al. Uterine contractile activity in healthy women throughout the\nmenstrual cycle measured using a novel quantitative two-dimensional transva-\nginal ultrasound speckle tracking method. Reprod. Biomed. Online 46, 115 –122\n(2023).\n25. A, K. et al. Cine MR imaging of uterine peristalsis in patients with endometriosis.\nEur. Radiol. 17, 1813 –1819 (2007).\n26. Nakai, A. et al. Uterine peristalsis shown on cine MR imaging using ultrafast\nsequence. J. Magn. Reson. Imaging 18, 726 –733 (2003).\n27. S, L. et al. Optimized approach to cine MRI of uterine peristalsis. J. Magn. Reson.\nImaging 44, 1397 –1404 (2016).\n28. Shitano, F. et al. Evaluation of uterine peristalsis using cine MRI on the coronal\nplane in comparison with the sagittal plane. Acta radiol 57, 122 –127 (2016).\n29. Georgiou, J. et al. The electrouterograph: a novel tool for assessing uterine\ncontractions of non-pregnant women. IEEE Open J. Eng. Med. Biol. 3,3 4 –\n40\n(2022).\n30. Sammali, F., Kuijsters, N. P. M., Schoot, B. C., Mischi, M. & Rabotti, C. Feasibility of\ntransabdominal electrohysterography for analysis of uterine activity in non-\npregnant women. Reprod. Sci. 25, 1124 –1133 (2018).\n31. Wu, W. et al. Noninvasive high-resolution electromyometrial imaging of uterine\ncontractions in a translational sheep model. Sci. Transl. Med. 11, eaau1428 (2019).\n32. Wang, H. et al. Accuracy of electromyometrial imaging of uterine contractions in\nclinical environment. Comput. Biol. Med. 116, 103543 (2020).\n33. Wang, H. & Wang, Y. Spatial-dependent regularization to solve the inverse pro-\nblem in electromyometrial imaging. Med. Biol. Eng. Comput. 58, 1651 –1665\n(2020).\n34. Cahill, A. G. et al. Analysis of electrophysiological activation of the uterus during\nhuman labor contractions. JAMA Netw. Open 5, e2214707 –e2214707 (2022).\n35. Zhang, Y. et al. Analysis of in vivo uterine peristalsis in the non-pregnant female\nmouse. Interface Focus 9, 20180082 (2019).\n36. Eytan, O. et al. Characteristics of uterine peristalsis in spontaneous and induced\ncycles. Fertil. Steril. 76, 337 –341 (2001).\n37. van Gestel, I., Ijland, M. M., Hoogland, H. J. & Evers, J. L. H. Endometrial wave-like\nactivity in the non-pregnant uterus. Hum. Reprod. Update 9, 131 –138 (2003).\n38. Ijland, M. M., Hoogland, H. J., Dunselman, G. A. J., Lo, C. R. & Evers, J. L. H.\nEndometrial wave direction switch and the outcome of in vitro fertilization. Fertil.\nSteril. 71, 476 –481 (1999).\n39. Kunz, G., Beil, D., Deininger, H., Wildt, L. & Leyendecker, G. The dynamics of rapid\nsperm transport through the female genital tract: Evidence from vaginal\nS. Wang et al.\n10\nnpj Women’s Health (2024)     1 \n\nsonography of uterine peristalsis and hysterosalpingoscintigraphy. Hum. Reprod.\n11, 627 –632 (1996).\n40. Fanchin, R. et al. Uterine contractions at the time of embryo transfer alter\npregnancy rates after in-vitro fertilization. Hum. Reprod. 13, 1968 –1974 (1998).\n41. Lo, L. W. et al. An Inkjet-Printed PEDOT:PSS-Based Stretchable Conductor for\nWearable Health Monitoring Device Applications. ACS Appl. Mater. Interfaces 13,\n21693–21702 (2021).\n42. Lo, L. W. et al. Stretchable sponge electrodes for long-term and motion-artifact-\ntolerant recording of high-quality electrophysiologic signals. ACS Nano 16,\n11792–11801 (2022).\n43. Wang, Y. & Rudy, Y. Application of the method of fundamental solutions to\npotential-based inverse electrocardiography. Ann. Biomed. Eng. 34, 1272 –1288\n(2006).\n44. Lammers, W. J. E. P., Ver Donck, L., Stephen, B., Smets, D. & Schuurkes, J. A. J. Focal\nactivities and re-entrant propagations as mechanisms of gastric tachyar-\nrhythmias. Gastroenterology 135, 1601 –1611 (2008).\n45. Han, H., Cheng, L. K. & Paskaranandavadivel, N. High-resolution in vivo mono-\nphasic gastric slow waves to quantify activation and recovery pro files. Neuro-\ngastroenterol. Motil. 34, e14422 (2022).\n46. Paskaranandavadivel, N., Ogrady, G. & Cheng, L. K. Time delay mapping of\nhigh-resolution gastric slow wave activity. IEEE Trans. Biomed. Eng. 64,1 6 6\n(2017).\n47. Erickson, J. C. et al. Automated gastric slow wave cycle partitioning and visuali-\nzation for high-resolution activation time maps. Ann. Biomed. Eng. 39, 469 –483\n(2011).\n48. O ’Grady, G. et al. Origin and propagation of human gastric slow-wave activity\ndefined by high-resolution mapping. Am. J. Physiol. - Gastrointest. Liver Physiol.\n299, 585 –592 (2010).\nACKNOWLEDGEMENTS\nWe thank the participants for their involvement in the research program. We thank\nSharon Achilles, MD., Ph.D. for her critical guidance, comments, and advice across this\nstudy. We thank Deborah Frank, Ph.D., for editing the manuscript; Madison Copeland\nfor managing and coordinating the study; Bri McNeil and Marlene Kouakam for\nexplaining the study to patients and obtaining consent; and Nilay Jakati for helping\nwith the patient experiments.\nAUTHOR CONTRIBUTIONS\nS.W. and Y.W. designed the experiments and developed the UPI software. K.A., S.P.,\nQ.W., V.R., and Y.W. contributed to the study design and guided the clinical studies.\nQ.W. and Y.W. developed the MRI sequence. S.W., K.A., S.P., and H.X. conducted\nhuman experiments. S.W. and H.X. segmented the MR images, S.P. reviewed the TVUS\nimages. S.W. analyzed the data. S.W., K.A., S.P., V.R., Y.W. contributed to the\nmanuscript writing.\nCOMPETING INTERESTS\nThe authors declare no competing non- financial interests but the following\ncompeting financial interests: Y.W. is a scienti fic consultant for Medtronic, EP\nsolution, and has research fundings from NIH, Bill & Melinda Gates Foundation, and\nBurroughs Wellcome Fund.\nADDITIONAL INFORMATION\nSupplementary information The online version contains supplementary material\navailable at https://doi.org/10.1038/s44294-023-00003-x.\nCorrespondence and requests for materials should be addressed to Valerie Ratts or\nYong Wang.\nReprints and permission information is available at http://www.nature.com/\nreprints\nPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims\nin published maps and institutional af filiations.\nOpen Access This article is licensed under a Creative Commons\nAttribution 4.0 International License, which permits use, sharing,\nadaptation, distribution and reproduction in any medium or format, as long as you give\nappropriate credit to the original author(s) and the source, provide a link to the Creative\nCommons license, and indicate if changes were made. The images or other third party\nmaterial in this article are included in the article ’s Creative Commons license, unless\nindicated otherwise in a credit line to the material. If material is not included in the\narticle’s Creative Commons license and your intended use is not permitted by statutory\nregulation or exceeds the permitted use, you will need to obtain permission directly\nfrom the copyright holder. To view a copy of this license, visit http://\ncreativecommons.org/licenses/by/4.0/.\n© The Author(s) 2024\nS. Wang et al.\n11\nnpj Women’s Health (2024)     1","source_license":"CC0","license_restricted":false}