Enhancing Ultrasound Interrogation of the Fetal Brain with AI Innovations
Elena Sinkovskaya, MD, PhD
Professor of Obstetrics and Gynecology
Director of Ultrasound Research & Education
Program Director of the Fetal Cardiology Fellowship
Macon & Joan Brock Virginia Health Sciences at Old Dominion University, Norfolk, VA
Introduction
As clinicians deeply involved in prenatal care, we recognize the pivotal role prenatal detection of fetal brain pathologies plays in improving maternal and fetal outcomes. A detailed understanding of normal fetal brain anatomy, accurate delineation of diagnostic planes, and the ability to quickly identify Central Nervous System (CNS) abnormalities during conventional 2D ultrasound exams are essential for accurate fetal CNS evaluation. These tasks, which are highly operator-dependent, can hinder the early detection of subtle anatomical changes, impacting decision-making on further genetic testing, fetal surgery, or pregnancy termination. Although many congenital CNS anomalies can be identified during routine anatomical ultrasound surveys, detection rates still vary considerably.
Despite significant advances in ultrasound technology over the past few decades, prenatal detection of fetal cerebral anomalies remains a challenge. These include difficulties in acquiring the appropriate diagnostic imaging planes, operator-dependent variability, and the inherent limitations in visualizing complex anatomical structures such as the fetal brain. The human eye can miss subtle abnormalities, particularly in the intricate fetal brain during routine screening examinations, highlighting the need for improved imaging techniques. In this article, we will discuss how a suite of artificial intelligence (AI) technologies are enhancing the visualization and confidence in detailed assessment of the fetal brain. By improving the interrogation of the fetal brain, these technologies not only advance diagnostic accuracy but also pave the way for better maternal and fetal health outcomes.
The Role of AI in Fetal Neurosonography
The incremental integration of artificial intelligence (AI) into medical imaging is set to revolutionize the field, with significant advancements seen in areas such as prenatal ultrasound. Over recent years, automated software utilizing machine learning (ML) algorithms has increasingly been adopted into clinical practice. In particular, AI-driven technologies for image recognition and reconstruction are playing an increasingly pivotal role in transforming clinical workflows, training methods, and diagnostic accuracy, especially in prenatal ultrasound. Automated tools assist in analyzing highly complex three dimensional (3D) anatomical structures, such as the fetal heart and central nervous system (CNS). The application of these tools has shown considerable potential in improving the accuracy and efficiency of prenatal examinations by standardizing imaging across different practitioners, reducing interoperator variability that can otherwise affect diagnostic quality1-3.
AI-powered software solutions can be beneficial to both novice and experienced examiners. Inexperienced operators can rely on these tools for more accurate basic examinations and biometric measurements, while experienced sonographers can use them to efficiently perform comprehensive fetal neurosonograms, conserving time and resources. However, it is important to note that advanced software solutions are not yet widely accessible in routine clinical practice, and traditional ultrasound neuro-imaging techniques still need to be taught and mastered. Ongoing developments in AI and ultrasound software will continue to enhance the clinical utility of these technologies, and further studies are needed to evaluate their effectiveness in clinical settings4-6.
AI Technologies for Neurosonography
A comprehensive evaluation of the fetal CNS Enhancing Ultrasound Interrogation of the Fetal Brain with AI Innovations Elena Sinkovskaya, MD, Professor of Obstetrics and Gynecology Director of Ultrasound Research & Education Program Director of the Fetal Cardiology Fellowship Macon & Joan Brock Virginia Health Sciences at Old Dominion University, Norfolk, VA A closer look at AI-driven advancements in fetal neurosonography including advanced volume techniques to improve maternal and fetal outcomes.
The incremental integration of artificial intelligence (AI) into medical imaging is set to revolutionize the field, with significant advancements seen in areas such as prenatal ultrasound includes labeling and measuring key anatomical structures in the brain. Neurosonography requires visualizing very small and complex three-dimensional structures which can be difficult to delineate. When assessing the fetal brain in 3D, the exam complexity and manipulation can be even more challenging and require significant expertise which includes advanced manipulation of the 3D volume.
Live ViewAssist™ (Fig. 2) and associated licensed technologies such as ViewAssist™ (Fig. 1 and Biometry Assist™ Fig. 1 (Samsung, Seoul, South Korea) automatically recognizes the standard imaging planes, labels anatomical landmarks, and automatically provides biometric measurements while live scanning and extracts targeted images and measurements with no user interaction required Fig. 2. In the advanced 3D assessment of the fetal brain, 5D CNS+ (Samsung, Seoul, South Korea) is an automatic 3D feature for fetal brain assessment & biometric measurements. Provides 9 fetal brain views for diagnosis following the ISUOG guidelines. Displays axial, sagittal, and coronal views and 6 measurements from 3 transverse views generated from a single volume which reduces the complexity related to 3D Volume manipulation.
AI technologies utilizing three-dimensional (3D) ultrasound - 5D CNS+
The integration of high-resolution ultrasound probes and the expansion of 3D ultrasound into routine prenatal practice have improved the detection and diagnosis of CNS anomalies Fig. 3. The advantages of three-dimensional (3D) ultrasound in examining the fetal brain have been well-documented, but challenges remain. These include issues with fetal orientation within the 3D volume, difficulty in manipulating images across the x, y, and z planes, and a lack of standardization in the technique.
The ability to correctly identify diagnostic planes for detailed neurosonography from 3D volumes is highly operator-dependent and requires significant expertise7, with notable intra- and interobserver variability and low reproducibility. Samsung’s 5D CNS+™ is a notable advancement in addressing these challenges. By offering automated imaging planes, labeling, and measurements, this technology significantly reduces the reliance on manual input. This reduces variability across different practitioners and helps ensure that each patient receives consistent, high-quality care, regardless of the facility or operator.
Key Features of 5D CNS+ include:
Enhanced Visualization: The 5D CNS+ Fig. 4 offers multi-dimensional imaging, providing a detailed, comprehensive view of the fetal brain. It automatically produces all the required diagnostic imaging planes, facilitating a thorough neurosonogram that complies with clinical guidelines.
Automated Measurements: AI-driven algorithms automatically generate the necessary measurements of fetal brain structures Fig. 4, eliminating human error and the time spent on manual calculations. This feature ensures greater consistency and improves diagnostic accuracy.
Improved Workflow: With AI automating image acquisition, labeling, and measurement, clinicians can focus more on interpretation and decision-making rather than adjusting imaging planes or taking repeated measurements. This leads to faster diagnoses, better patient care, and more efficient use of resources. The Impact of AI on Clinical Practices In clinical practice, the impact of AI on fetal assessments is profound.
These technologies streamline the ultrasound process, ensuring that we capture all relevant anatomical landmarks with greater precision. In my experience, this means that clinicians can focus on the interpretation of the images, rather than on the technicalities of image acquisition. This shift ultimately benefits patient care, as it allows us to provide faster, more accurate diagnoses. Furthermore, AI technologies can serve as a valuable educational tool for medical professionals.
As we continually refine our skills in interpreting complex fetal images, AI can help us learn more effectively. Through training modules and simulations, AI can assist in honing ultrasound interpretation skills, especially in areas like fetal brain pathology where expertise is critical. While transitioning to new technologies may pose challenges, such as resistance from practitioners accustomed to traditional methods, I believe that ongoing education and training can ease this process. Ensuring that clinicians are well-supported and understand the potential benefits of AI will be key to the successful integration of these technologies in prenatal care.
Conclusion
The integration of AI into ultrasound technology represents a transformative shift in how we evaluate the fetal brain. By overcoming the limitations of traditional 2D imaging, these technologies not only enhance diagnostic accuracy but also enable more efficient workflows, helping us provide better care for expectant mothers and their babies. As AI continues to evolve, the future of prenatal care looks even brighter, with the potential for more accurate, timely, and personalized interventions. As clinicians, it’s essential that we stay abreast of these technological advancements and continue to incorporate them into our practices. With AI, we have the opportunity to enhance both our diagnostic capabilities and patient care, ultimately improving outcomes for both mother and child.