Adenomyosis of Uterus, Congenital Abnormalities, Endometriosis, Infertility, Female, Leiomyoma (Uterine Fibroids), Uterine Neoplasms
Conditions
Keywords
reproductive medicine, artificial intelligence, deep learning, endometriosis, endometrial receptivity, adenomyosis, uterine leiomyoma
Brief summary
The goal of this observational study is to develop and validate artificial intelligence (AI)-based algorithms that support ultrasound diagnosis of endometriosis, adenomyosis, myometrial masses, uterine malformations, and impaired endometrial receptivity in women aged 18-45 years. The main questions it aims to answer are: Can AI algorithms, applied to standardized transvaginal ultrasound images, accurately detect and classify endometriosis and adenomyosis in real time during routine examination? Can AI-based ultrasound assessment predict the histological dignity of myometrial masses, and the likelihood of successful assisted reproduction (AR/IVF) outcome from endometrial features? Participants attending the Department of Obstetrics and Gynecology, Semmelweis University, with clinical suspicion of endometriosis or adenomyosis, a confirmed myometrial mass scheduled for surgery, a suspected uterine anomaly, or scheduled IVF treatment, will undergo standardized transvaginal ultrasound examination (following the IDEA, MUSA, and IETA protocols) alongside collection of clinical, questionnaire, and, where surgery is performed, histopathological data. Imaging and clinical data will be used to build a database supporting the development and validation of the AI algorithms.
Interventions
None listed
Sponsors
Semmelweis University
Study design
Observational model
COHORT
Time perspective
PROSPECTIVE
Eligibility
Sex/Gender
FEMALE
Age
18 Years to 45 Years
Healthy volunteers
No
Inclusion criteria
* Women aged 18-45 years with clinical suspicion of endometriosis or adenomyosis, in whom the diagnosis of endometriosis is subsequently confirmed, including peritoneal, ovarian, and deep infiltrating endometriosis. The diagnosis is established by imaging, laparoscopy, or laparotomy, based on characteristic intraoperative findings and histological analysis (#Enzian classification). * Women undergoing ultrasound examination for suspected congenital uterine anomaly or other intracavitary uterine pathology. * Women with previously confirmed leiomyoma scheduled for surgical treatment. * Women scheduled for IVF treatment or embryo transfer. * Willingness and capacity to provide written informed consent prior to enrolment.
Exclusion criteria
* TVUS is not technically feasible. * Postmenopausal status. * Pregnancy. * Puerperium (up to 3 months postpartum). * Suspected premalignancy, or presence or history of malignancy. * Chronic comorbidities, including uncontrolled diabetes mellitus, severe cardiovascular or respiratory disease, systemic autoimmune disease, or uncontrolled thyroid dysfunction. * Other conditions, including psychiatric illness or substance use, and any circumstance in which study participation could pose a risk to the patient, or in which enrolment could bias the study results. * For the endometrial receptivity sub-study, confirmed uterine or endometrial pathology other than adenomyosis.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Diagnostic accuracy of the AI algorithm for detection of endometriosis and adenomyosis on transvaginal ultrasound | 3 years | Sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) of the AI-based detection algorithm for endometriosis and adenomyosis, assessed against the reference standard of surgical findings and histopathological confirmation (for participants undergoing surgery) or, where surgery is not performed, expert consensus review of the standardized ultrasound examination. |
Countries
Hungary
Contacts
CONTACTNandor Acs, PhD
Outcome results
None listed