Currently, it is possible to screen for Down syndrome during pregnancy. However, the screening measure used – ultrasound imagery – has less than 80% detection accuracy using common indicators, making it difficult to provide a concrete and truthful diagnosis. Other screening measures, such as amniocentesis or fetal umbilical venipuncture, can be used. But these measures are invasive and may be harder on the mother. According to Medical XPress, a research team from the Chinese Academy of Sciences is working to develop an effective, but non-invasive, way to screen for Down syndrome during the first trimester using artificial intelligence (AI).
Developing the Screening Tool
To begin, the research team created a Convolutional Neural Network. MathWorks explains that a Convolutional Neural Network is:
a network architecture for deep learning which learns directly from data, eliminating the need for manual feature extraction. CNNs are particularly useful for finding patterns in images to recognize objects, faces, and scenes [and] can also be quite effective for classifying non-image data such as audio, time series, and signal data.
Next, the researchers created various layers within this network which can identify and detect various shapes and edges. They utilized a class activation map to explain the model’s focus and discriminative features. Finally, the researchers trained and validated the model using 2D ultrasound images of fetal faces between gestation ages 11-14 weeks from 822 fetuses.
Ultimately, the research team determined that the model was able to improve screening measures, increasing screening accuracy by over 15%.
Interested in learning more? Take a look at the data in JAMA Network Open.
What is Down Syndrome?
Also known as Trisomy 21, Down syndrome is a congenital genetic disorder in which someone is born with an extra copy of chromosome 21. Around 95% of people with this condition have a complete extra copy. 3% have an extra copy attached to another chromosome, and 2% have an extra copy in only some cells. Doctors are not sure what causes this extra chromosome. However, one risk factor is having a child after age 35. Down syndrome affects 1 in every 700 babies in the United States. Symptoms and characteristics can vary widely from person-to-person. Potential characteristics can include:
- Short stature
- A short neck, small ears, flattened face, and almond-shaped eyes
- White spots on the iris
- Small hands and feet
- Hypotonia (low muscle tone)
- Loose joints
- Hearing loss
- Obstructive sleep apnea
- Mild-to-moderate cognitive impairment
