Endometrial Receptivity Prediction During in Vitro Fertilization Using Artificial Intelligence

Infertility, Infertility (IVF Patients)

assisted reproduction, Artificial intelligence (AI), Endometrial receptivity, Embryo implantation, Ultrasound imaging

The investigators plan to use artificial intelligence to analyse vaginal ultrasound images of the uterine lining (endometrium) taken during routine IVF treatment, which may predict implantation success during IVF treatment. Participation in the study is voluntary, involves no additional testing or intervention beyond routine procedures, and consent can be withdrawn verbally or in writing at any time without cause or adverse consequences. Over a three-year period, the trial is expected to enrol approximately 1,500 patients between the ages of 18 and 40 who are indicated for IVF treatment and who volunteer for treatment. Patients enrolled in the study will not be required to attend more clinic visits during treatment than they would otherwise have to. During the trial, certain patient-specific data (age, indication for treatment, body mass index), stimulation-specific data (duration of stimulation, type and dose of drug, endometrial thickness), ultrasound scans and outcome-specific data (treatment failure, biochemical pregnancy, clinical pregnancy) will be collected. The data will be stored in a secure database. The data collected during the study will only be accessible to the professionals involved in the study and no information, including personal data, will be disclosed to third parties.

Infertility affects one in six people globally over the duration of their reproductive lives. For successful IVF treatment, three essential things are needed: a euploid embryo, receptive endometrium and in case of frozen embryos, the exact timing of the transfer. Several methods are available for evaluating embryos, such as: microscopic evaluation (e.g. Gardner grade), genetic tests (PGT), biochemical tests (non-invasive PGT). Furthermore, morphological and morphokinetic evaluation is possible using time-lapse video surveillance systems that are evaluated using artificial intelligence. The most common method for testing the receptivity of the endometrium is the determination of the largest diameter measured in the mid-sagittal plane. A thickness between 7 and 10 mm is considered the cut-off for successful implantation. Further studies were conducted on the level of the serum and endometrial progesterone, the volume, echogenicity, peristalsis of the endometrium, the flow conditions in it and the success of implantation. The predictive value of hysteroscopy, 3D ultrasound and the role of ultrasound elastography to predict pregnancy outcome has been investigated. The above-mentioned tests for endometrial receptivity have not been widely used in practice, and there is currently no method for predicting endometrial receptivity in everyday practice. As the quality of imaging has improved, the focus has shifted to analyzing the pattern of the endometrium, the latest research using artificial intelligence. The aim of this study is to investigate whether segmentation and analysis of the endometrium using artificial intelligence in vaginal ultrasound images taken during stimulation and on the day of transfer can help to more accurately determine the receptivity of the endometrium. The significance of this is that in case of poor implantation chances it is possible to freeze the embryo(s) and have the opportunity to implant the embryo in a subsequent natural or hormone replacement therapy cycle, possibly with better chances.

No related papers found yet.

None listed

Hungary