Related Experiment Video
Updated: Sep 7, 2025
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
2.0K
Development and Validation of a Deep Learning Model to Screen for Trisomy 21 During the First Trimester From Nuchal
Liwen Zhang1, Di Dong1,2, Yongqing Sun3
1CAS Key Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Sciences, Beijing, China.
JAMA Network Open
|June 21, 2022
Summary
This study developed a deep learning model for first-trimester trisomy 21 screening using ultrasound images. The model demonstrated high accuracy in detecting trisomy 21, outperforming traditional methods.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Healthcare
- Prenatal Diagnostics
Background:
- First-trimester screening for trisomy 21 is crucial for reproductive decision-making.
- Current screening methods can be improved for accuracy and early detection.
Purpose of the Study:
- To develop and validate a deep learning model for screening fetuses with trisomy 21.
- To assess the model's performance using ultrasonographic images.
Main Methods:
- A convolutional neural network was developed using 2D fetal facial images from singleton pregnancies (11-14 weeks gestation).
- Data augmentation and random selection for training and validation sets were employed.
- Model performance was evaluated using receiver operating characteristic curves, AUC, sensitivity, and specificity.
Main Results:
- The deep learning model achieved high accuracy in trisomy 21 screening.
- Area Under the Curve (AUC) for the model was 0.98 in the training set and 0.95 in the validation set.
- The model outperformed traditional screening methods based on nuchal translucency (NT) and maternal age.
Conclusions:
- The developed deep learning model accurately screens fetuses for trisomy 21.
- This model shows potential as a tool for universal primary screening of trisomy 21.
- Early and accurate screening facilitates informed reproductive choices.
