Study Comparing Two Image Acquisition Modalities for Second-trimester Pregnancy Screening Ultrasound (Echo-IA)

Clinique Rive Gauche

Status: Enrolling

Conditions: Artifical Intelligence; Echography Ultrasound

Treatments: Procedure: Standard Ultrasound; Procedure: US with AI

Study type: Interventional

Funder types: Other

Identifiers

NCT07286591

Echo-IA

Details and patient eligibility

About

The second-trimester morphology ultrasound is a key examination in obstetric monitoring that aims to assess fetal growth, identify any structural abnormalities, and inspect anexes such as placenta, umbilical cord, cervix,...

Several studies suggest that a significant proportion of fetal malformations can be detected during this time frame if a complete morphological analysis is performed. However, the reliability of the screening depends on the quality of the equipment, the operator's level of expertise, and adherence to protocols that define the necessary scans.

In France, since the first reports of the National Technical Committee on Prenatal Screening Ultrasound (2005), particular attention has been paid to standardizing practices. More recently, the French National Conference on Obstetric and Fetal Ultrasound (CNEOF) published new recommendations (2022, revised in 2023) including the development of reference silhouettes for the second-trimester examination, proposing 26 views (22 required and 4 additional). However, the CNEOF does not formalize quality criteria for evaluating the conformity of these images; this task has been taken over by the French College of Fetal Ultrasound (CFEF), which has established a scoring and validation grid for each fetal slice (see CFEF 2022 document).

In parallel, artificial intelligence (AI) is gradually becoming established as a decision support and automation tool in medical imaging, particularly in ultrasound. Deep learning algorithms are capable of identifying anatomical structures, positioning measurement markers, and selecting the most optimal slice, reducing inter-operator variability and streamlining workflow. In the field of obstetric ultrasound, some companies have launched systems capable of detecting or annotating fetal structures in real time, potentially improving diagnostic reliability and reproducibility. Samsung has developed a system called Live View Assist, available on its latest generation ultrasound scanners, which uses AI to automatically recognize and freeze the required fetal slices in real time.

The tool also offers automated validation: if the detected slice conforms to the expected standards, it is directly checked off on a checklist. This innovation promises time savings, a reduced risk of missing certain complex slices, and improved standardization.

However, there is little data, particularly in France, regarding to the actual performance of this tool in a routine screening context. Before considering the integration of Live View Assist and AI into daily practice, it is therefore essential to evaluate the quality of the images it acquires, the feasibility of a complete examination assisted by AI, as well as the potential impact on examination time and improvement of the workload for sonographers.

The aim of this study is to evaluate whether the quality of the 20 mandatory images automatically validated by Live View Assist is not inferior to that of the 20 mandatory images acquired and validated manually by an ultrasound technician, according to the CFEF quality criteria based on the silhouettes recommended by the CNEOF.

Enrollment: 50 estimated patients

Sex: Female

Ages: 18+ years old

Volunteers: No Healthy Volunteers

Inclusion criteria

• Women aged 18 or over,
• With a single, viable pregnancy, definitively dated by first-trimester ultrasound, with no known malformations,
• Scheduled for a routine second-trimester screening ultrasound, i.e., between 20 weeks + 0 days and 24 weeks + 6 days,
• Having given their informed consent,
• Affiliated with the social security system or a beneficiary of such a plan.

Exclusion criteria

• Multiple pregnancy,
• Known fetal malformation,
• Pathological pregnancy,
• Cognitive impairment, or a disorder causing difficulty understanding instructions or answering questionnaires,
• Patient under legal guardianship,
• Patient not covered by health insurance,
• Protected patient: adult under guardianship, curatorship, or other legal protection, deprived of liberty by judicial or administrative decision

Trial design

Primary purpose: Screening

Allocation: Randomized

Interventional model: Parallel Assignment

Masking: None (Open label)

---

50 participants in 2 patient groups

Standard ultrasound

Active Comparator group

Description:

Manual acquisition: acquisition of the 22 mandatory images and, if possible, the 4 additional images by the sonographer, respecting the standardized procedure derived from the CNEOF recommendations.

Treatment:

Procedure: Standard Ultrasound

Artificial intelligence-assisted ultrasound

Experimental group

Description:

AI-assisted acquisition: using the Live View Assist system to automatically capture the 20 mandatory views in real time, with validation and recording of images as soon as the algorithm detects an image conforming to the reference, and if possible the 2 complementary images.

Treatment:

Procedure: US with AI