A Comparative Study of AI Methods for Fetal Diagnostic Accuracy in Ultrasound

A Comparative Study of AI Methods for Fetal Diagnostic Accuracy in Ultrasound

This study serves as a supplemental investigation to the randomized controlled SCAN-AID study (NCT0632187). This study will evaluate and compare the fetal growth estimation outcomes of AI-supported groups, expert sonographers, and control groups using a secondary AI predictive model.

The goal of this study is to compare the effects of two distinct AI methods on fetal ultrasound diagnostic accuracy. It serves as a supplementary investigation to the SCAN-AID study (NCT NCT06232187). This study aims will compare the diagnostic accuracy of two types of AI methods.

From the SCAN-AID study ultrasound novices were randomized into one of three groups with different levels of AI support: control group, AI feedback group 1 where the participants are presented with basic black box AI feedback, and AI feedback group 2 where the participants are presented with a more detailed explainable AI feedback. All the participants are tasked to perform an ultrasound fetal weight estimation (EFW) on pregnant women at gestational age 30-37. The outcomes were than compared to the expert sonographers measurements.

In this study an operator independent AI method that predicts the fetal weight is used on the SCAN-AID ultrasound examinations. .

Pregnant women:

Inclusion Criteria:

• Singelton pregnant.
• Gestational age: 30-38 weeks
• Maternal age < 40 years

Exclusion Criteria:

• Oligo hydramnion
• Severe fetal anomaly e.g. fetal heart anomaly, omphalocele etc.
• Severe fetal macrosomia or growth restriction.