Artificial intelligence-based prenatal ultrasound for diagnosing fetal lip and palate during the second trimester

Abstract

Background: Fetal facial anomalies are among the most common congenital conditions, with cleft lip and palate being the most prevalent. This study aimed to evaluate the clinical value of convolutional neural networks in automatically identifying standard ultrasonic cross-sectional images of the fetal lip and palate during the second trimester.

Methods: From September 2021 to December 2022, prenatal sonographers collected dynamic videos of the lip and palate of 700 fetuses at 20-24 weeks of gestation, including 5 standard cross-sectional images, along with nonstandard cross-sectional images and background images. A YOLOv5-based artificial intelligence (AI) model was used as the object detection network. The sonographers manually marked the images of 500 fetuses (450 in the training set and 50 in the validation set). The AI model, a midcareer sonographer, and a junior sonographer were involved in the classification and identification of standard fetal lip and palate ultrasound images in the test set (200 fetuses), and the results were compared with the standard results obtained by the senior prenatal sonographer. The receiver operating characteristic curve was plotted, and the sensitivity, specificity, and accuracy were calculated for the AI model, midcareer sonographer, and junior sonographer.

Results: For the standard coronal section of the nasal lip, the area under the curve (AUC) of the AI model, the midcareer sonographer, and the junior sonographer was 0.971, 0.935, and 0.880, respectively. For the standard midsagittal section of the face, the AUC of the AI model, the midcareer sonographer, and the junior sonographer was 0.988, 0.939, and 0.904, respectively. For the standard upper alveolar ridge section, the AUC of the AI model, the midcareer sonographer, and the junior sonographer was 0.977, 0.840, and 0.824, respectively. For all standard sections, the AI model demonstrated significantly higher AUC values as compared to both the midcareer and junior sonographers (P<0.05).

Conclusions: The AI model demonstrated higher classification efficacy than did the midcareer and junior sonographers and performed more quickly and efficiently.

Keywords: Fetal; artificial intelligence (AI); lips and palate; quality control; standard ultrasonic section.

Figure 1

Standard and nonstandard ultrasonic sections of fetal facial regions. (A) Standard coronal section of the nasal and labial regions. (B) Nonstandard coronal section of the nasal and labial regions. (C) Standard midsagittal section of the facial region. (D) Nonstandard midsagittal section of the facial region. (E) Standard transverse section through the upper dental arch. (F) Nonstandard transverse section through the upper dental arch. (G) Standard oblique coronal section through the lower lip or jaw (showing the soft palate). (H) Nonstandard oblique coronal section through the lower lip or jaw (showing the soft palate). (I) Background image.

Figure 2

The plot of the descending trend of loss and the changing trend of the validation set performance. The detection box loss, foreground loss, and classification loss of the training set all show a decreasing trend, but the loss in the validation set generally becomes flat or no longer decreases after about 50 epochs. Meanwhile, the accuracy, recall rate, and mAP of the validation set also remained flat or no longer increased after about 50 epochs. mAP, mean average precision; Val, validation.

Figure 3

Network architecture diagram. Conv, convolution; CSP, cross stage partial.

Figure 4

ROC of the AI model, midcareer sonographer, and junior sonographer. (A) Nasal-labial coronal section. (B) Midsagittal section of the face. (C) Transverse section of the maxilla. (D) Oblique coronal section through the lower lip or jaw (showing the hard palate). (E) Oblique coronal section through the lower lip or jaw (showing the soft palate). The blue line represents the result from the AI model, the yellow line represents the result from the midcareer sonographer, and the green line represents the result from the junior sonographer. AI, artificial intelligence; ROC, receiver operating characteristic.