Artificial Intelligence Can Detect Early and Accurate Prenatal Diagnosis of Cystic Hygroma

According to a recent study by the University of Ottawa's Faculty of Medicine, artificial intelligence (AI) may recognize rare, fatal diseases and birth abnormalities in prenatal ultrasound imaging and detect cystic hygroma. Dr. Mark Walker, a co-founder of the OMNI Research Group (Obstetrics, Maternal, and Newborn Investigations) at The Ottawa Hospital, and his research team wanted to test how well AI-driven pattern recognition could do the job. Normally, the birth defect can be easily diagnosed prenatally during an ultrasound appointment.

Artificial Intelligence (AI) Early Diagnosis of Cystic Hygroma

The study's objective was to show that deep-learning architecture can help the early and accurate diagnosis of cystic hygroma from first-trimester ultrasound imaging. The study is available in the peer-reviewed open-access journal PLOS ONE.

Cystic hygromas are benign tumors that frequently develop on a newborn's neck and take the form of a fluid-filled sac. A blockage in the lymphatic system, which results in fluid accumulating under the skin, causes the cyst to form. The cyst forms due to a lymphatic system blockage, which causes fluid to accumulate beneath the skin. These cysts can be fatal and may result in stillbirth or miscarriage.

ALSO READ: Nanoparticles Mothers Inhaled During Pregnancy Can Cross the Placenta That Protects the Fetus

Artificial Intelligence (AI) Study Analysis

Images were processed using a DenseNet model to accurately detect cases of cystic hygroma compared to healthy controls using a dataset of over 300 prenatal ultrasounds gathered retroactively at The Ottawa Hospital. Sensitivity, specificity, and other metrics were calculated.

To evaluate the interpretability of the model, gradient class activation heat maps representing pixels in images were also created. The whole model's accuracy was 93%. The results showed that deep-learning algorithms could reach high accuracy in diagnosing cystic hygroma in the first trimester.

The model performed exceptionally well despite a limited number of training photos. Walker said that it is possible that what they showed was that they could utilize the same algorithms for picture classification and identification in the field of ultrasound with high sensitivity and specificity.

The research team at the University of Ottawa has high expectations for where this kind of study could go. The team thinks its method might be used to treat other fetal defects typically detected by ultrasonography with future development, including testing in a sizable multi-site dataset.

According to Dr. Walker, the team plans to create a global consortium to upload obstetric ultrasound images to the cloud. In the end, this might assist doctors in low- and middle-income nations achieve cloud-driven interpretation and diagnosis.

"What we demonstrated was that we are able to use the same tools for image classification and identification with high sensitivity and specificity in the field of ultrasound." Walker said. He thinks their method could be used to treat other fetal anomalies typically detected by ultrasound.

Dr. Walker specializes in high-risk obstetrics and clinical epidemiology. He is a co-founder of the largest maternal and newborn research group in Canada.

Check out more news and information on Technology in Science Times.