Applicability of Ultrasound Volumetric Approach in the Prenatal Detection of Fetal CNS Anomalies

1. Introduction

Prenatal screening for fetal anomalies is routinely performed in obstetrical practice to detect life-threatening conditions. The development of in-vivo fetal imaging techniques, such as ultrasound and magnetic resonance imaging (MRI), marked a cornerstone in prenatal detection and assessment of central nervous system (CNS) malformations, an essential aspect of paramount importance in fetal wellbeing.

There is a reported high incidence of CNS malformations in as many as 1–2/1.000 fetuses, consequently accounting for the most common congenital abnormalities [1, 2, 3]. Neural tube defects are the most frequent CNS malformations, with a prevalence in pregnancy of 52/100.000, but evidence suggests that cerebral structural anomalies with an intact neural tube might reach an incidence of approximately 1/100 births [3].

Basic mid-trimester transabdominal ultrasound examination of the fetal brain requires the acquisition of three views based on two axial planes: transventricular and transthalamic views required for the evaluation of brain hemispheres and an additional axial transcerebellar view necessary for the evaluation of the posterior fossa [4]. The following structures should be analyzed: cavum septum pellucidi, midline falx, thalami, lateral ventricles with choroid plexus, cerebellum and cisterna magna [4]. Whenever a fetal brain anomaly is suspected or the pregnancy is associated with high-risk for the development of the aforementioned conditions, experts rely on extended ultrasound examination. This comprehensive targeted evaluation is entitled “neurosonography”. It is an advanced ultrasound assessment of the fetal CNS that implies up-to-date knowledge of fetal neurology and brain anatomy. Specific indications for targeted fetal neurosonography can be reviewed in the updated International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) guideline dedicated to fetal CNS examination [2].

As mentioned before, suspicion of fetal CNS abnormalities leads to the necessity for further paraclinical investigations to disentangle doubtful diagnoses. In addition to mid-trimester transabdominal ultrasound, fetal neurosonography encompasses the transvaginal multiplanar approach and fetal MRI. Both ultrasound techniques offer the possibility to study brain growth and development during the fetal period using the volumetric approach. Fetal MRI is a non-invasive diagnostic tool that can characterize the development of the CNS due to increased image resolution and tissue contrast and therefore contributes to pregnancy management [5].

The basis for targeted ultrasound examination of fetal CNS is the multiplanar approach [3]. Compared to routine screening evaluation which focuses on three axial views obtained by transabdominal evaluation, neurosonography prioritizes the transvaginal examination when the fetus is in vertex presentation, due to the higher resolution of images allowing for appropriate display of additional coronal and sagittal planes. For other fetal presentations, an external version might be considered [3]. Four coronal views (transfrontal, transcaudate, transthalamic and transcerebellar) and three sagittal views (mid-sagittal anterior, mid-sagittal posterior, parasagittal) are recommended key planes, but additional views can be necessary for diagnosis, depending on the region of interest; combined approaches – transabdominal and transvaginal are sometimes required as well [3, 6].

In this context, the role of three-dimensional (3D) ultrasound is unquestionable, since it allows the operator to display and analyze simultaneously perfectly aligned views in all orthogonal planes, which also benefit from enhanced image quality [3]. Several international guidelines and protocols provide detailed instructions on the specific methodology required for the relevant acquisition of these views [2, 3, 4].

Depending on the resources each department benefits from, various imaging options have been proposed and studied in order to maximize the cost-benefit balance in the practice of prenatal fetal anomaly screening and diagnosis.

Conventional two-dimensional (2D) ultrasound is a powerful, real-time non-invasive and non-ionizing diagnostic tool in the hands of an experienced operator [7, 8]. It has broad clinical applicability, and the fact that it is a safe imaging technique, allows it to be the preferred diagnostic approach in fetal imaging.

In everyday obstetrical activity access to an ultrasound machine has become indispensable. Depending on the aim of the obstetrical evaluation, 2D ultrasound mode is an adequate resource for the assessment of fetal wellbeing and growth as well as for standard screening for structural anomalies. However, over the last couple of years, an increased requirement for additional ultrasound tools has emerged, especially to improve the detection rates during the performance of targeted sonography. This is mainly related to the operator-dependent imaging acquisition mode, which precisely influences the interpretation of ultrasound images and data. This is a consistent issue especially if the processes of acquisition and interpretation of images are performed by different people: a sonographer and/or a clinician/obstetrician, depending on the specific national health care system requirements.

Multiple clinical applications of 2D planar ultrasound imaging are relevant to the day-to-day obstetrical workflow. Several criteria have promoted this imaging system as a clinically accepted screening and diagnostic tool, mainly during prenatal follow-up: a fast, safe, economical and portable ultrasound examination usually provides a rapid and accurate assessment of both the fetus and the mother, reducing the length of stay in the clinic [9]. Compared to the MRI – the other medical imaging system available for use during pregnancy, which implies a significantly higher financial burden for any healthcare system as well as a time-consuming clinical activity, the 2D ultrasound evaluation offers an acceptably accurate examination in the hands of a skilled user, enough to justify its operation and maintenance costs [9]. The set of images provided and stored by MRI examination are undoubtedly superior but still sensitive to fetal motion artifacts.

On the other hand, the 2D scan allows the user to identify images of different sections of 3D fetal anatomical structures, but in the absence of an automatic frame of reference, the exact orientation of the image is not precisely determined since it is strongly dependent on the hand positioning of the ultrasound probe. Therefore, there is an increased variability in the image acquisition process, although in clinical practice obstetrical guidelines offer strong recommendations concerning the standard views to be used in fetal anomaly screening and diagnosis as well as the anatomical landmarks to aid measurements and localize probe position and orientation [9].

Volumetric 3D ultrasound aims to surpass the operator-dependent characteristic of standard 2D scans, by using specialized probes. By performing a volumetric 3D scan of different sections of the body using a standard, anatomically registered frame of reference, this examination offers the possibility to swipe through a set of individual images belonging to a block of dataset, allowing the user to access sagittal, transverse and/or coronal planes.

Potential limitations that encourage continuous research for more complex and advanced image acquisition strategies refer to the small field of view, limited penetration, diffraction-limited resolution and the fact that a 2D image yields only a selective cross-sectional sampling of a complete 3D anatomical volume; consequently, a significant level of skill and experience are essential for the user to obtain a recognizable, clinically useful, high-quality ultrasound image [7, 10].

Performing a fetal scan either for screening of fetal anomalies or for evaluation of fetal growth is routine practice during prenatal visits. As mentioned before, depending on the aim of the evaluation, capturing in detail a region of interest has become a necessity. Volumetric 3D ultrasound imaging has emerged as a solution for this, based on the motor-controller “wobbler probe” with the internal motorized translation of a one-dimensional array and 2D “matrix array probes”, which allow for acoustic beam steering in the elevation dimension in addition to the azimuth dimension [7, 9]. Nevertheless, image processing and interpretation are still dependent on the operator’s skills, requiring advanced knowledge and experience to obtain maximum information related to the region of interest.

Ultrasound probes using sensing technologies to achieve volumetric image reconstruction represent an alternative to volumetric 3D ultrasound and have been investigated since the beginning of the 1990s. The main difference between them and the classic 2D probe technology resides in the integration of real-time tracking of the position and orientation of the probe and its acquired image data in 3D space [9]. To achieve this, a reference frame meaning a 3D Cartesian coordinate system must be established, and the transformation between each object specified; spatial location tracking is a functional requirement and is obtained using an electromagnetic field sensor or pre-calibrated optical tracking setup to retund both spatial coordinates and quaternions – the accepted standard parameterization of rotation [7].

Researchers at Stanford and Duke University have addressed the increasing demand for financially accessible alternatives and have developed setup devices that do not involve the high costs of the 3D ultrasound equipment mentioned above, but further improvements are needed [7, 9]. Current studies are focused on increasing the accuracy and intuitiveness of ultrasound image acquisition, with less reliance on operator skills; this way, several emerging tracking technologies – although still in the research stage, have proved significant potential to overcome current difficulties regarding motion artifacts and imaging resolution [11, 12]. In the era of the artificial intelligence breakthrough, it has become increasingly necessary to reach high rates of reproducibility in the field of diagnostic imaging by promoting automatization as part of medical progress.

Furthermore, the fetal automatic segmentation is of particular interest at present, since this would represent a pragmatic solution to the abovementioned downfalls of both 2D and 3D volumetric ultrasound approaches. Particularly referred to the challenges associated with the volumetric ultrasound of the whole fetal head; automatic solutions are currently sought to overcome:

• the poor image quality resulting from the speckle noise and low resolution

• low tissue contrast and long-span shadow occlusion caused by the significant acoustic attenuation on the skull

• the large variety of fetal heads, mainly inner structures’ morphology during different gestational periods and fetal poses [13]

Multiple attempts have been dedicated to meet these challenges and even to simultaneously segment the fetus, gestational sac and placenta in ultrasound volumes [14].

However, neuroimaging studies during the fetal period began to advance almost two decades ago. Brain structure segmentation is an important prerequisite for identifying and labelling brain regions and deriving more accurate quantitative measurements [5, 15]. These measurements of the fetal brain and subcortical volumes can facilitate the early identification of predictors for brain dysmaturation [1]. Early detection of fetal brain abnormalities is of key clinical importance due to its involvement in obstetrical management of the affected pregnancies and the potential prediction of neurodevelopmental outcomes.

Several studies have focused on the potential of artificial intelligence to automatically segment brain tissue compartments using MRI, to address the issue of manual segmentation, which requires time and expertise; results showed significant efficacy and reproducibility of the conceived mechanisms, especially for the segmentation of the cerebellum and thalami – key cerebral structures involved in neurocognition and motor behavior [1, 5, 15]. Still, compared to MRI imaging, fetal sonography is strictly dependent on tissue properties and the positioning of the area of interest relative to the ultrasound transducer: anatomical boundaries are typically incomplete, and strong acoustic shadowing can interfere with image acquisition [16]. Initial studies focused on automatic ultrasound detection of specific fetal brain structures but currently, there is growing interest in automatic segmentation techniques, which allow for additional depiction of the structure’s shape and appearance [16]. Huang et al. studied two brain structures that have different ultrasound appearances: the choroid plexus and the corpus callosum and proposed a method that allowed for automatic biometry measurements with acceptable deviations compared to within human inter−/intraobserver deviations and a high region segmentation accuracy [16].

2. Fetal CNS examination in early gestation

Early detection of fetal structural abnormalities at 11 + 0 to 14 + 0 weeks of gestation is the primary objective of the first-trimester morphology scan. ISUOG recommends that this examination should not be limited only to the assessment of fetal crown-rump length (CRL) and nuchal translucency (NT) [17] but should also encomprise a throughout evaluation of fetal anatomy, suggesting that adequate time allocated for a detailed structural survey can increase the detection rate.

Nevertheless, several synergistic factors influence the possibility of ultrasound detection of fetal structural anomalies. The type and quality of ultrasound equipment available for routine screening is an essential element to be considered. This is even more relevant if taking into consideration second-opinion scans or targeted fetal sonographies. Additionally, in these particular cases, the experience and skills of the examiner play a significant role in reaching a successful, complete and correct examination. In the process of ultrasound screening for anomalies, the sonographer needs to consider the epidemiological aspects related to potential structural defects that can be identified in the specific population and at a specific time in pregnancy.

In the first trimester of pregnancy, the main advantage of ultrasound screening resides in the fact that transvaginal probes can be used for adequate image acquisition, allowing excellent representation of fetal brain structures. However, at this early stage in pregnancy, manipulation of the fetus for optimal depiction of morphology corresponding to specific anatomical structures is extremely limited. This is where the value of 3D ultrasound specifically emerges. Experts in the field suggest that the fetal brain can be accurately assessed as early as 7 weeks of pregnancy, stating that embryological development can be followed using ultrasound, allowing for holoprosencephaly, anencephaly and spina bifida to be discovered early in the first trimester [18, 19].

Development of the fetal brain is a dynamic process and ultrasound evaluation of cerebral structures should take into account the rapid changes that the fetal CNS undergoes from the early to the mid and second half of gestation. Advances in imaging technology have known a breakthrough over the past three decades, allowing for studies to address the issues of early detection of fetal CNS anomalies. Early fetal neurosonography at 12–15 weeks of gestation allows for a throughout visualization of brain structures using a high-frequency transvaginal transducer [2, 3, 20], especially since the ossification process of the calvarium is not complete.

At 13–14 weeks of gestation, the focus should be on appraising the normal structure of the diencephalon and the posterior fossa, taking into consideration the fact that complete development of the cerebellum takes place later, in the second trimester, and therefore, suspected isolated vermian abnormalities at this stage in pregnancy associate high-risk of false-positive results [3, 19]. In this latter situation, mid-sagittal views can distinguish between normal development of posterior fossa structures and potential structural anomalies. New studies and protocols address the potential of early spina bifida diagnosis; specifically, the mid-sagittal plane allows the examiner to identify indirect signs and to perform custom measurements depicting the ratio between the brain stem and the distance between the brain stem to the occipital bone which in open spina bifida is >1 [3, 21]. Furthermore, Ushakov underlined the importance of 3D ultrasound volume acquisition at the first-trimester scan when he described “the crash sign” detectable on axial views in fetuses with spina bifida, marking the posterior displacement of the mesencephalon and deformation against the occipital bone [22]. However, as Paladini and Volpe anticipated more than 10 years ago, open spina bifida, cystic major anomalies of the posterior fossa and cerebral ventriculomegaly should remain “potentially detectable abnormalities” during early scans in pregnancy [19].

Specialized literature and internationally up-to-date protocols and guidelines recommend that despite the increasing performance in early neurosonography, a follow-up scan at 20 weeks of gestation should be performed in all pregnancies [3, 19]. Only particular cases of holoprosencephaly, anencephaly and gross cephalocele should disregard this recommendation since these entities should be clear-cut pathological entities to be recognized during first-trimester scan [3, 19].

3. Fetal CNS examination in later gestation

Ultrasound screening for fetal anomalies has been a routine assessment at mid-gestation for more than four decades. Specifically related to the evaluation of the fetal CNS, axial planes have been used for ensuring biometrical measurements and anatomical depiction of the main visible cephalic structures.

While for low-risk populations this protocol for basic examination covers sufficiently the need for prenatal morphological assessment of the fetus, in high-risk populations when suspicion of abnormal development occurs, additional investigations are required. Initially, MRI offered a feasible alternative for the detailed characterization of specific regions of interest and the surrounding areas of the fetal brain, mainly due to its quality as a non-ionizing radiation exposure imaging technique. Nevertheless, associated increased costs and timing for image acquisition, combined with restricted accessibility, emphasized the requirement for ultrasound technologists to provide solutions that would overcome the abovementioned downfalls.

Early studies on retrieving 3D ultrasound volumes began 30 years ago when researchers focused on precise quantitative volume measurement of fetal organs [23]. Rapid evolution in the field started with the construction of reference centiles based on the assessment of fetal brain volume and led in over a decade to a persistent search for improved solutions for more complex clinical issues related to the practice of prenatal anomaly screening: for example, refining the diagnostic ultrasound accuracy for fetal craniofacial dysmorphisms using shape optimization processes after fetal head segmentation [23, 24]. In parallel, comprehensive studies were conducted to identify normal global growth trajectory of fetal cerebral structures such as the cortical plate, deep gray nuclei and ventricles, using MRI [25]. It is already known that brain maturation takes place in the latter half of the pregnancy and that its structure has specific features during cerebral expansion. Disruptions of the expansion processes affecting the growth and regression of certain brain structures are suspected to determine the cognitive outcome of the fetus [25]. Therefore, it is of paramount importance to recognize the contribution MRI and histopathological studies had to the understanding of brain development; improving the slice thickness as part of the advanced image processing techniques to adjust the inter-slice motion and obtain a super-resolution reconstruction of volumetric images maintained fetal brain volumetry examinations essential in the evaluation of fetal development [26].

However, by using only the transverse plane of the fetal head, limited information can be obtained; consequently, the need for further evaluation of the remaining orthogonal planes has become more valuable. This was in accordance with postnatal ultrasound examinations of the brain, carried out through the large fontanel, providing coronal and sagittal planes that did not serve neonatologists for comparison with prenatal images. It was in 2007 that ISUOG issued the first guideline addressing the need for extended CNS sonography, known as “fetal neurosonography” [2, 3]. This way, suspicious findings in the fetal brain would require additional coronal and sagittal planes as part of the extended ultrasound evaluation. As expected, depending on multiple maternal (increased body-mass index) and fetal factors (presentation, anterior placental insertion, multiple pregnancies, etc.), these additional cutting planes are sometimes difficult to obtain.

With the introduction of 3D and 4D ultrasound, volume data sets are generated, and these allow for any cutting planes to be subsequently reconstructed. In 2012, Chaoui et al. were among the first experts to present the clinical applications of multiplanar reconstruction using 3D sonography in fetal CNS evaluation [18]. The digital information stocked in one or multiple volumes can be examined either retrospectively by 3D scanning mode or live by 4D examination. The reconstruction of the plane of interest is performed after the acquisition of the volume, which is further displayed using volume rendering – a similar technique to the one used for MRI and CT scanning.

Different representations of the cutting plane are possible: as a single cutting plane, as three orthogonal planes or as tomography – multiple very thin layer planes using volume contrast mode imaging (VCI) as the default setting to minimize artifacts; additionally, the examiner can draw a line or a curve by himself which can serve as the basis for the creation of the image [18]. This virtual reconstruction of images facilitates access to structures that are not usually sonographically clearly visible during a standard transabdominal ultrasound, such as the corpus callosum and the cerebellar vermis – for these structures, if the fetus lies in cephalic presentation, a transvaginal scan is an alternative for assessing their morphology.

For volume recording, Chaoui recommends setting a cross-sectional plane of the fetal head, parallel to the skull base, containing the cavum septi pellucidi, using a sweep angle >50°, to cover the entire brain [18]. This way, multiplanar rendering offers the possibility to display the architecture of the falx cerebri with the cavum septi pellucidi, anterior and dorsal horns of the lateral ventricles, cortex, cerebellum and cisterna magna. Moreover, the Omniview application allows the sonographer to recreate both sagittal and coronal planes based on the same initial acquisition transverse plane, to further assess the structure of the corpus callosum, cerebellar vermis, anterior complex and longitudinal ventricular diameter (Figure 1) [18].

In specific, targeted cases, multiplanar rendering can be performed after recording a data volume based on a coronal plane, using the cavum as an anatomical landmark and a sweep angle >70° [18]. In this case, in addition to the previously mentioned structures, an image of the interhemispheric gap, the thalami, the lateral insulae as well as the gyri and sulci in the 3rd trimester can be analyzed [18]. In other words, suspected anomalies related to agenesis of the corpus callosum, to the anterior complex (anterior horn and cavum septi pellucidi) or the cortex, can be more precisely identified.

Gray matter migration process occurs in the 12th week of fetal development, which overlaps with the time of corpus callosum development; consequently, dysplasias of the corpus callosum can be accompanied by gray matter heterotopia [27]. Other CNS anomalies such as Dandy-Walker syndrome, absence of cavum septum pellucidi, Chiari malformation, septo-optic dysplasia, etc. can be identified concomitantly to corpus callosum dysplasia secondary to a neural tube disorder caused by specific adverse factors during the dorsal induction process of nerve end plate in the early embryonic development period [27].

A more detailed description of the corpus callosum morphology can be found in the reconstruction of the section based on the volume obtained using an initial sagittal plane of the fetal head and an acquisition angle >50°; in addition to the midline structures, the neighboring brain areas are also depicted and above all, an optimal display of genu, corpus, isthmus and splenium of the corpus callosum can be obtained, facilitating diagnosis of hypoplasia or other dysplasias of the corpus callosum [18]. Also, this is the preferred scan for cerebellar vermis assessment, including fissura prima, fissura secunda and brainstem to vermis angle [18], practical along with the transvaginal ultrasound, whenever anomalies of the fetal posterior fossa are suspected.

Reconstruction of the corpus callosum can be achieved after recording a volume data set based on a cross-sectional image of the fetal head, taking into account the following recommendations: to obtain the sagittal plane, the examiner must adjust the levels so that the A-axis lies in the same plane as the falx cerebri and the B plane perpendicular to the falx and the cavum septi pellucidi, thus resulting in a C-plane corresponding to the thalamus, cavum septi pellucidi and corpus callosum [18]. In a 4D scan, a line can be drawn by the sonographer directly on the region of interest.

Concomitantly, a similar reconstruction process can be performed to examine the cerebellar vermis further. If the basic ultrasound identifies suspicious findings in the area of the fetal posterior cranial fossa, additional sagittal exploration of the vermis and the surrounding structures needs to be accomplished. Reconstruction of this sagittal plane can be performed directly via the transcerebellar plane after volume acquisition, allowing the A-axis to pass through the vermis and the falx, and B plane perpendicular to the centre line; the C level will therefore visualize the vermis and the vermis-brain stem relation [18].

A more recent paper cites the applicability of the 3D volumetric transvaginal approach in the early characterization of posterior fossa structures, underlying the fact that the anterior membranous area and cerebellar vermis can be subject to in detail evaluation of their morphological changes even at first-trimester scan, as a basis for better understanding of pathological development of posterior fossa [28]. However, further research is needed before considering this approach in populational practice.

Experts mention that 3D ultrasound technologies such as volume contrast imaging in combination with C-plane (VCI-C) and tomographic ultrasound imaging (TUI) although used for the assessment of corpus callosum and vermis cerebelli, still have some limitations since they cannot clearly distinguish the corpus callosum from septum pellucidi because the acoustic beam is parallel to fetal head [27]. This is where the implementation of the 3D Omniview technique can provide the corresponding image of the region of interest by manually cutting using either a straight line (Figures 2 and 3) or a curvilinear approach to the 3D data in an arbitrary direction [27].

4. Other implications

The concept of “reorganization of the developing human brain” in the context of pathological conditions or lesions became of particular importance in the early 2000s when a study group reunited Nicolaides, Gratacos, Figueras and co. sought to evaluate the feasibility and reproducibility of volume segmentation of fetal intracranial structures using 3D ultrasound imaging [29]. Their results were in line with previous postnatal research, underlining the existence of selective growth restriction in certain brain regions which could disclose diminished cognitive function and delayed neurodevelopmental processes later in life [30]. The efficacy of a semiautomatic segmentation in 3D ultrasound volumes: Virtual Organ Computer-Aided Analysis (VOCAL) managed to show a reduction in the frontal and an increment in the thalamic volumes in fetuses with intrauterine growth restriction, compared with those appropriate for gestational age, matched by gestational age [29]. Later, the development of automatic segmentation of specific brain structures on 3D ultrasound volumes, addressed the need for manual delineation by an expert operator, centreing the cerebellum as a suitable anatomical landmark for intrauterine growth and health assessment [31, 32]. Calibration over movement, posture and balance is a hallmark of cerebellar function. There are some critical points to be addressed in the matter of cerebellar ultrasound volumetric study, which justify the peculiar interest of sonographers in this brain structure. First, the cerebellar volume is highly correlated with the gestational age and the transverse cerebellar diameter, and the specific organ segmentation provides the proper features for Dandy-Walker syndrome diagnosis; second, 2D ultrasound has several limitations regarding the measurements of lengths, contours or volumes of structures with an irregular shape as is the case of the cerebellum, and the manual segmentation is both time-consuming and inconsistent [33].

Benacerraf, a pioneer of ultrasound applications in prenatal diagnosis, presented an insightful use of volumetric ultrasound approaches in the detection and diagnosis of fetal brain anomalies. She emphasized the role of 3D scan used for the differential diagnosis of encephalocele, as well as for the reconstruction of the three orthogonal planes or for the tomographic cut involved in the diagnosis of septo-optic dysplasia, her work being in agreement with other ultrasound experts [19, 34].

Pooh et al. described the wide spectrum of volumetric ultrasound neuroimaging modes, highlighting the importance of unlimited offline analysis of the brain morphology, mentioning volume contrast imaging techniques (VCI) and HDlive silhouette imaging, at the same time questioning the role of fetal MRI in the diagnostic process [19, 20]. While at the beginning of 3D volumetric ultrasound use neurosonography was considered the centerpiece for fetal anomaly screening before 24 weeks of gestation and MRI after this gestational age, expert panels currently agree that these imagistic tools are complementary, MRI as a second-line investigation being relevant for diagnostic in only 7–15% of cases [3, 20].

Specifically for cases where CNS congenital defects are suspected, different ultrasound modes are more relevant than others, depending on the complexity of diagnostic criteria. For example, while ventriculomegaly can be easily depicted using axial views in 2D ultrasound which allows for accurate measurement of the enlarged atrial width in the lateral ventricles >10 mm, any structural impairment of corpus callosum, cerebellum or other posterior fossa structures still pose diagnostic challenges and require supplementary sagittal views, preferentially in multiplanar modes [20].

Corpus callosum and the cerebellum have been widely studied cerebral structures during intrauterine life, due to their relevance in normal development of motor, sensory and cognitive functions. These fundamental structures begin to develop early in the life of the embryo and their definitive functions are well established either at the end of the first year of postnatal life for the cerebellum or at the end of the second year of life for the corpus callosum [20, 35]. Although there have been almost 15 years since the fundamental principles of fetal neurosonography have been addressed by experts, there are still multiple limitations that encompass this practice: fetal brain development is a dynamic entity and can be susceptible to lesions later during pregnancy and in the postnatal life; parent counseling in cases of CNS anomalies involve diverse scenarios, from “nearly” normal neurological outcomes to severely impaired functions, depending on the complexity of the suspected condition [6, 20, 34]. The intricacy of prenatal CNS anomaly diagnosis goes even further, adding more knowledge to the core of the embryological development of all cerebral anatomical structures: ultrasound evaluation should be performed as a complete examination emerging from the latest international guidelines in the field and special consideration should be given to populational variability too. The most relevant example is that of corpus callosum evaluation, where its progressive morphological development – first of the anterior part and after of the posterior part, should be taken into account when suspecting potential hypoplasia after 20 weeks of gestation. Nevertheless, this adds to the fact that guidelines recommend rather a qualitative assessment than a quantitative one since a short, thin or thick corpus callosum is not necessarily synonymous with an abnormality of this cerebral structure [3, 19].

The rapid spread of fast volumetric ultrasound (4D) along with artificial intelligence technology in different medical fields can only represent the inception of a new era concerning innovations in imagistic diagnosis and therapy [12]. Precise motion tracking with 4D ultrasound is currently intensely studied and holds promising potential in prenatal screening and diagnosis. To fully understand the complexity of the brain, functional neuroimaging techniques could be of use in intrauterine life as well. Solutions could emerge from several investigations of functional ultrasound performed on rodents aiming to develop a method to image dynamic deep brain activation by directly measuring subtle cerebral blood volume changes [36]. Also, ongoing developments in the field of virtual reality enable experts to apply new research opportunities in the field of prenatal imaging as well: novel volumetric measurements based on segmentation of various parts of the fetal body at the end of the first trimester [37] possess the ability to add to an immersive body of knowledge with respect to adaptation mechanisms in early pregnancy which imply contextual adverse outcomes later in life.

To conclude, there is a remarkable multidirectional learning trajectory regarding ultrasound opportunities in prenatal diagnosis. The reliability of 2D ultrasound in obstetrics remains dependent on the skill and experience of the operator but there is evidence that both novice and expert interpretations of key biometric measurements of 3D volumetric datasets are highly reliable [38]. At the same time, the continuous search for domain shifting towards automatization in prenatal screening provides new evidence that with the use of system coordinates, it is now possible to automatically locate and segment the fetal brain and eye sockets in 2D and 3D images [39]. In contrast, based on economic and financial disparities resembling different healthcare systems, other researchers are focusing on reconstructing ultrasound volumes from 2D scans, without requiring extra equipment, since this method is widely used at the bedside; the results proved significant potential, at the same time promoting access for vulnerable populations of society to advanced monitoring in pregnancy [40, 41].

Acknowledgments

Acknowledgements to all pregnant patients admitted in the Obstetrics and Gynecology Department of “Sf. Pantelimon” Clinical Emergency Hospital for sharing their trust in their physicians.

No funding was received to assist with the preparation of this manuscript.

Conflict of interest

The author declares no conflict of interest.