Results of the summarizing process for the arXiv paper: 2201.07935v2

Developing innovative informatics approaches to enhance fetal monitoring is a rapidly growing field in reproductive medicine. While several reviews have focused on using artificial intelligence (AI) techniques to improve pregnancy outcomes, they often have limitations by focusing on specific data related to the mother's care during pregnancy. In this systematic survey, the aim was to explore how AI can assist with fetal growth monitoring through ultrasound (US) images. To conduct the survey, eight medical and computer science bibliographic databases were used to retrieve studies published between 2010 and 2021. A total of 1269 studies were initially retrieved, but after excluding duplicates and evaluating titles and abstracts, 107 distinct studies were included in the survey. The findings of the survey revealed that 2D ultrasound images were more popular than 3D and 4D ultrasound images. The most commonly used AI technique was classification, followed by segmentation, classification integrated with segmentation, and other miscellaneous techniques such as object detection, regression, and reinforcement learning. Within the pregnancy domain, the most common areas of focus were the fetus head, body, heart, abdomen, and face. In recent studies, deep learning techniques were primarily utilized followed by machine learning algorithms such as artificial neural networks and reinforcement learning. These AI techniques played a crucial role in predicting fetal diseases and identifying fetus anatomy structures during pregnancy. However further research is needed to validate these technologies from a physician's perspective. This includes conducting pilot studies and randomized controlled trials on AI applications in a hospital setting. Additionally some areas for improvement in existing studies include increasing sample sizes and incorporating additional performance evaluation metrics such as AUC (Area Under Curve), DSC (Dice Similarity Coefficient), and F1 score. Overall this systematic survey provides valuable insights into how AI can assist with fetal growth monitoring using ultrasound images. It highlights the potential of these technologies in improving pregnancy outcomes but also emphasizes the need for further research and validation in order to ensure safety standards are met before implementation into clinical practice settings.

• - Developing innovative informatics approaches to enhance fetal monitoring is a rapidly growing field in reproductive medicine.
• - AI techniques can assist with fetal growth monitoring through ultrasound images.
• - 2D ultrasound images are more popular than 3D and 4D ultrasound images.
• - The most commonly used AI technique is classification, followed by segmentation, classification integrated with segmentation, and other miscellaneous techniques such as object detection, regression, and reinforcement learning.
• - Common areas of focus within the pregnancy domain include the fetus head, body, heart, abdomen, and face.
• - Deep learning techniques are primarily utilized followed by machine learning algorithms such as artificial neural networks and reinforcement learning.
• - AI techniques play a crucial role in predicting fetal diseases and identifying fetus anatomy structures during pregnancy.
• - Further research is needed to validate these technologies from a physician's perspective, including conducting pilot studies and randomized controlled trials on AI applications in a hospital setting.
• - Areas for improvement in existing studies include increasing sample sizes and incorporating additional performance evaluation metrics such as AUC (Area Under Curve), DSC (Dice Similarity Coefficient), and F1 score.

Summary: Scientists are finding new ways to use computers to help doctors take care of babies before they are born. One way is by using special pictures called ultrasound images to check how the baby is growing. Some pictures show the baby in 2D, which means flat like a drawing, and these are more common than 3D or 4D pictures. Computers can use a type of artificial intelligence called classification to help doctors understand the pictures better. They can also predict if the baby might have a problem or find important parts of the baby's body like the head, heart, and abdomen. But more research is needed to make sure these computer techniques work well for doctors. Definitions- Informatics: Using computers and information technology to solve problems. - Fetal monitoring: Checking how a baby is doing before it is born. - Reproductive medicine: The branch of medicine that deals with problems related to having babies. - AI techniques: Using computers to do things that normally require human intelligence. - Ultrasound images: Pictures made using sound waves that show what's inside someone's body. - Classification: Sorting things into different groups based on their similarities. - Segmentation: Separating something into different parts or sections. - Object detection: Finding and identifying specific objects in an image or video. - Regression: Predicting future values based on past data. - Reinforcement learning: Teaching a computer program how to make decisions by giving it rewards or punishments based on its actions. - Deep learning techniques

AI Techniques for Fetal Growth Monitoring: A Systematic Survey

In reproductive medicine, the use of innovative informatics approaches to enhance fetal monitoring is a rapidly growing field. Artificial intelligence (AI) techniques have been used in several reviews to improve pregnancy outcomes, but these studies often have limitations by focusing on specific data related to the mother's care during pregnancy. To explore how AI can help with fetal growth monitoring through ultrasound (US) images, a systematic survey was conducted. This article will discuss the findings of this survey and provide insights into how AI can be used in improving pregnancy outcomes.

Methods

The survey was conducted using eight medical and computer science bibliographic databases to retrieve studies published between 2010 and 2021. After excluding duplicates and evaluating titles and abstracts, 107 distinct studies were included in the survey.

Findings

The findings revealed that 2D ultrasound images were more popular than 3D or 4D ultrasound images when it came to utilizing AI techniques for fetal growth monitoring. The most commonly used technique was classification followed by segmentation, classification integrated with segmentation, object detection, regression, and reinforcement learning. In terms of areas of focus within the pregnancy domain, the fetus head, body, heart, abdomen and face were most common areas studied using AI techniques. Deep learning techniques such as artificial neural networks (ANNs) and reinforcement learning algorithms were primarily utilized in recent studies while machine learning algorithms also played an important role in predicting fetal diseases as well as identifying fetus anatomy structures during pregnancy.

Conclusion

This systematic survey provides valuable insights into how AI can assist with fetal growth monitoring using ultrasound images which could potentially lead to improved pregnancy outcomes if implemented correctly into clinical practice settings after further research has been conducted to validate these technologies from a physician's perspective including conducting pilot studies and randomized controlled trials on AI applications in a hospital setting . Additionally some areas for improvement include increasing sample sizes as well as incorporating additional performance evaluation metrics such as AUC (Area Under Curve), DSC (Dice Similarity Coefficient), F1 score etc.,