Although it is a tedious procedure, early diagnosis of the fetus is highly critical to increasing the chances of survival. The reasons for newborn death can vary. However, congenital heart problems make up about 20 percent of cases.
Masaaki Komatsu at RIKEN AIP and a group of researchers developed an AI model that can define a healthy fetal heart in response to those high mortality rates.
Dig Deeper
Robot Uses Deep Learning, Ultrasound Imaging to Draw Blood AI Method Helps Further Outcomes-Based Research for Cardiovascular Disease
Initially, the model was exposed to thousands of ultrasound images and defined each as healthy or unhealthy. The decisions of the AI model then engaged in another round of deep learning, which further enhanced accuracy. Providers used the tool to view the statistics on a chart, allowing them to distinguish differences between cases relating to the heart or blood vessels.
The next step involved experts, fellows, and residents providing diagnoses, both with and without the assistance of the AI model.
Researchers concluded that the AI-based decision charts led all groups to a more accurate diagnosis. Specifically, diagnoses for fellows and residents were 7 and 13 percent more accurate.
“Our study suggests that even with widespread use of AI assistance, an examiner’s expertise will still be a key factor in future medical examinations,” Komatsu said in the press release. “In addition to future clinical applications, our findings show maximum benefit from this technology could be achieved by also using it as part of resident training and education.”
There have been various studies involving AI assisting cardiac care and diagnosis in the past.
Recently, doctors at the Robert Wood Johnson University Hospital implemented a robotic tele-cardiac ultrasound system to enhance video communication for remote ultrasounds. The plan was successful, providing benefits such as making up for a shortage of ultrasound technologists.
Physicians at Cedars-Sinai also recently developed a new AI algorithm that assists in detecting heart disease. Hypertrophic cardiomyopathy and cardiac amyloidosis are frequently overlooked diseases; however, the new system can identify the thickness of heart walls and the size of heart chambers. Researchers trained this algorithm using over 34,000 ultrasound videos.
Another study published in Cardiovascular Research developed a machine learning technique to predict patients’ long-term risk of a heart attack. Researchers did this by combining coronary artery calcium scoring with non-contrast computed tomography, which indicates the accumulation of cholesterol within artery walls.
By 