To assess the value of double closed-loop management artificial intelligence (AI) in the quality control of standard obstetric ultrasound sections.

The data uploaded by Huazhong University of Science and Technology Union Shenzhen Hospital in the first and second quarters of 2021 (Q1 and Q2, from January to June) on the AI-based ultrasonic quality control system (345 cases including 15475 obstetric ultrasound images) were selected for AI quality control. Targeted learning was carried out according to the quality control results of Q1. Standard rate, basic standard rate, and non-standard rate of the images in the two quarters were calculated and compared. The reasons for the presence of non-standard images in the two quarters were analyzed. The accuracy of AI-based quality control system was tested by double-closed loop management mode of appeal for controversial result toward expert judgement.

The results indicated that the standard section rate increased from 80.76% in Q1 to 83.90% in Q2, while the non-standard section rate decreased from 5.36% to 3.33%. Besides, the rate of the sections that fully met the requirements increased from 14.3% to 42.9%, and the rate of the sections that were not qualified decreased from 23.8% to 0. The differences for the above comparisons were all statistical difference χ²=24.687, 0.050, 1.200, and 5.676, respectively; P<0.001, <0.001, =0.040, =0.071). The double-closed-loop management mode showed that the accuracy of AI-based quality control system was 98.5%.

AI-based quality control mode with double closed-loop management is objective, accurate, efficient, and convenient. Using this system can improve the accuracy of obstetric ultrasound examination, which is of great value in prenatal ultrasonic quality control.