Objective To compare sonographic visualization of midline cerebral structures obtained by two-dimensional (2D) imaging and three-dimensional (3D) multiplanar and volume contrast imaging in the coronal plane (VCI-C), with transfrontal 3D acquisition.
Methods Sixty consecutive healthy fetuses in vertex presentation at a mean gestational age of 24 (range, 20–33) weeks underwent 2D and 3D ultrasound examination. Sagittal cerebral planes were reconstructed using 3D acquisition from axial planes by multiplanar analysis and by VCI-C. The reconstructed midline images of both these techniques were compared with the midline structures visualized directly in the A-plane by transfrontal 3D acquisition using a sweep angle of 30°. Measurement of the corpus callosum and cerebellar vermis and visualization of the fourth ventricle and the main vermian fissures were compared. The sharpness of the images was also assessed qualitatively. Mid-sagittal tomographic ultrasound imaging (TUI) was also performed. 3D planes were compared with 2D transfontanelle median planes obtained by transabdominal or, when required, transvaginal sonography.
ResultsThe midline plane could be obtained in 88% of multiplanar, 82% of VCI-C and 87% of transfrontal analyses. Measurements of the corpus callosum and cerebellar vermis obtained by 3D median planes were highly correlated. The clearest and sharpest definition of midline structures was obtained with transfrontal acquisition. Primary and secondary fissures of the cerebellar vermis could be detected in 13–26% of multiplanar, 18–35% of VCI-C and 52–79% of transfrontal analyses. 2D visualization was superior or equal to the 3D transfrontal approach in all the parameters compared.
Congenital brain anomalies involving the median structures are difficult to diagnose during fetal life1, 2. This is mainly due to the fact that the standard axial sonographic views used routinely to assess the fetal brain do not display some cerebral structures, most notably the corpus callosum and the cerebellar vermis1–3. The most widely accepted imaging modality for assessing the brain in fetuses in vertex presentation is transvaginal neurosonography4. However, this technique requires some experience and is dependent on the skill of the operator5. Recently, the use of three-dimensional (3D) reconstruction of sagittal views from volumes acquired from transabdominal axial planes, by either multiplanar analysis of static volumes or volume contrast imaging in the C-plane (VCI-C), have been proposed6–8. Although these approaches may be valuable for rapid and simple assessment of fetal brain anatomy, physical limitations suggest that further studies are needed before clinical decisions can be based on such 3D reconstructions9.
In 2001, Visentin et al.10 reported the use of the transfrontal view as a new two-dimensional (2D) approach for the visualization of the fetal midline cerebral structures. Using the frontal suture as an acoustic window, it is possible to visualize simultaneously the facial profile and the midline structures of the brain. We decided, based on the original description of the transfrontal view, to evaluate transfrontal 3D acquisition using the frontal suture for volumetric analysis of the midline fetal brain structures.
The purpose of this study was to compare visualization of midline cerebral structures obtained by 2D imaging and by the 3D multiplanar and VCI-C modalities, with transfrontal 3D acquisition.
A total of 60 consecutive healthy second-trimester fetuses in vertex presentation were evaluated prospectively. All examinations were performed by the same operator (F.V.) using a Voluson 730 Expert ultrasound machine (GE Medical Systems, Kretz Ultrasound, Zipf, Austria) equipped with a 4–8-MHz transducer. All women underwent transabdominal multiplanar 3D acquisition from an axial plane, VCI-C from the same axial plane, and transfrontal acquisition from a view through the metopic suture. Measurements were taken using each of the three techniques. They were compared with the standard 2D transfontanelle image obtained transabdominally (n = 50) or transvaginally (n = 10).
Sagittal cerebral planes were reconstructed from axial planes by multiplanar analysis and by VCI-C as described previously6, 8. For transfrontal 3D acquisition, in order to use the metopic suture as an acoustic window, the ultrasound beams were aligned with the frontal suture and a mid-sagittal view of the fetal facial profile was obtained. When the fetus was in an unfavorable position, gentle manipulation of the fetal head by the free hand of the examiner was required to access the midline intracranial structures and to avoid acoustic shadows from the frontal bones. As is standard procedure in volume acquisition, the image was enlarged to at least a third of the screen; the transmitter focus was positioned inferior to the occipital bone and the render box was placed such that the whole fetal head was included within it. The sweep acquisition angle was set at 30°, the quality at high 2 and the frequency of the probe at mid/high harmonics. The acquisition began when the corpus callosum and the cerebellar vermis were clearly visible. If the free hand of the examiner was needed to control the fetal head position, the acquisition was activated using a foot switch. These settings provided a rapid 3D sweep, minimizing artifacts from fetal or maternal movement.
No more than two volume datasets per 3D technique were acquired. Patients gave oral informed consent for storage of volumes for offline evaluation. Offline analysis of the volume datasets was performed using specialized 3D software (4D View, GE Medical Systems). One examiner reviewed the volume datasets. For transfrontal volume display, the reference dot was first positioned on the cavum septi pellucidi in the acquisition plane (A-plane). In order to obtain an orthogonal midline section in the acquisition plane, the interhemispheric fissure on the B-plane was aligned with the y-axis, with slight z-rotation. Mid-sagittal tomographic ultrasound imaging (TUI), with a 1.5-mm interslice distance, was then displayed for the A-plane. A static VCI filter was used in order to evaluate whether it improved the quality of the image.
The sagittal planes were compared by measurement of the corpus callosum and the craniocaudal and anteroposterior diameters of the cerebellar vermis, and by visualization of the fourth ventricle and primary and secondary vermian fissures. Measurements were compared with published nomograms6, 8, 11. Pearson's correlation coefficients were calculated. The sharpness of the images was also assessed qualitatively. All cases were followed up postnatally and found to be normal.
The mean gestational age of the fetuses was 24 (range, 20–33) weeks. The 2D median plane could be obtained transabdominally or transvaginally in 59 of the 60 cases. Complete information, including measurement of the corpus callosum and cerebellar vermis, was obtained in 53 (88%) of the multiplanar reconstructions, 49 (82%) of the VCI-C reconstructions and 52 (87%) of the transfrontal 3D acquisitions.
The length of the corpus callosum estimated by measurements from the transfrontal approach correlated significantly with results obtained from multiplanar measurements (r = 0.88; P < 0.0001) and VCI-C measurements (r = 0.90; P < 0.0001). Craniocaudal cerebellar vermian diameter estimated by measurements from the transfrontal approach correlated significantly with results obtained from multiplanar measurements (r = 0.73; P < 0.0001) and VCI-C measurements (r = 0.79; P < 0.0001). Anteroposterior cerebellar vermian diameter estimated by measurements from the transfrontal approach correlated significantly with results obtained from multiplanar measurements (r = 0.73; P < 0.0001) but not with those from VCI-C measurements.
The primary vermian cerebellar fissure was visualized in 14/53 (26%) of the multiplanar reconstructions, 17/49 (35%) of the VCI-C reconstructions and 41/52 (79%) of the transfrontal 3D acquisitions (P < 0.0001); 2D median planes allowed visualization of this fissure in 49/59 cases (83%). The secondary vermian cerebellar fissure was visualized in 7/53 cases (13%) of the multiplanar reconstructions, 9/49 (18%) of the VCI-C reconstructions and 27/52 (52%) of the transfrontal 3D acquisitions (P < 0.0001); 2D median planes allowed visualization of these fissures in 46/59 cases (78%). On comparing visualization of the vermian fissures, there was no statistical difference between transfrontal 3D acquisition and the 2D median plane for the primary fissure (P = 0.572) but there was a difference for the secondary fissure (P = 0.003). The median views obtained by transfrontal 3D acquisition were superior in quality to those of the other 3D techniques and they were comparable to the 2D median images. The transfrontal technique allowed the corpus callosum to be identified as a thin sonolucent arched shape with defined contours overlying the cavum septi pellucidi. During multiplanar and VCI-C reconstruction, we regularly visualized acoustic shadowing by the petrous portion of the temporal bone and this frequently obscured the midbrain, pons and medulla, impeding adequate visualization of the cerebral aqueduct, the fourth ventricle and the normal relation between the brain stem and the cerebellar vermis. The volume analysis from the transfrontal 3D approach provided detailed visualization of these structures, of a quality comparable to that of the 2D median plane (Figures 1 and 2). VCI in the static plane was a useful filter that increased the quality of the image by enhancing the contrast and improving the depiction of margins (Figure 3).
3D volume sonography offers several advantages over 2D fetal imaging. An infinite number of 2D planes of a target anatomical region are acquired and displayed in three orthogonal planes. The operator has the possibility of offline multiplanar navigation through the volume, which allows reconstructed 2D planes to be created and displayed with fine calibration of the view to be used for reference and measurements5, 12, 13. Manipulation of the image and the use of some rendering modes enable anatomical views to be investigated that were not demonstrated in the original acquisition plane14. The use of post-processing filters, such as in VCI in the static plane, can provide images with no speckled pattern and with improved tissue contrast15, 16. In addition, the use of TUI allows a variable number of reconstructed 2D parallel sections to be displayed on a single panel, as in computed tomography or magnetic resonance imaging17. Efforts to achieve standardized techniques for acquisition and volume display should result in uniformity when retrieving 2D planes from volume datasets and in a simplified approach to training18. A volume can be stored and reviewed offline and even sent for examination by experts at a different site by volumetric telemedicine19.
The median plane of the fetal brain can be used to depict structures such as the corpus callosum, the cavum septi pellucidi, the head of the caudate nucleus, its thin covering tela choroidea, the thalamus, parts of the midbrain, the cerebellar vermis and the cisterna magna. Some elements of the ventricular system can also be visualized, such as the third and fourth ventricles and, depending on the image resolution, the cerebral aqueduct4, 20. However, this scanning plane is particularly difficult to obtain, requiring skill, time and often transvaginal examination8. Hence, some recent studies have assessed the feasibility of 3D sonography for obtaining median views from volumes acquired from transabdominal axial planes, by either multiplanar analysis of static volumes or VCI-C6–8. These studies demonstrated that 3D median planes were more easily obtained compared with 2D ones, providing a simple, rapid and effective approach for detailed evaluation of normal and abnormal cerebral anatomy. However, as a rule, in 3D sonography, the resolution of reconstructed planes is inferior to that of the acquisition plane. In the median planes obtained by multiplanar or VCI-C methodology, the corpus callosum could not be differentiated clearly from the inferior cavum septi pellucidi, generating a hyperechogenic complex from which the presence of the corpus callosum could be inferred8. However, hyperechogenicity at this level may signify potential pathology, mainly callosal lipoma9, 21. Acoustic shadowing from the petrous portion of the temporal bone obscures the midbrain and pons, also impeding precise assessment of their relationship with the cerebellar vermis. Furthermore, the principal features of the cerebellar vermis are often not clearly demonstrated8. It has recently been reported that the presence of an apparently normal vermis on a sagittal sonographic plane is far from sufficient to assess the normality of the posterior fossa or to rule out pathology with a poor prognosis3. The high proportion of disagreement between prenatal diagnosis of posterior fossa anomalies and autopsy findings emphasizes the need for accurate sonographic demonstration of the cerebellar vermis and its relationships22.
In our center, visualization of the fetal facial profile is an essential part of the assessment of the mid-trimester fetus. The inclusion of this view, we feel, should be standard in routine fetal sonographic evaluation used for the diagnosis of a wide range of craniofacial anomalies and phenotypic expression of different chromosomal and genetic abnormalities. Based on the original description of the transfrontal view10, we decided to evaluate transfrontal 3D acquisition using the frontal suture for volumetric analysis of the midline fetal brain structures. A recent study illustrates that the onset of fusion of the frontal bones, starting from the glabella, is evident from the 32nd week of intrauterine life23. Thus, the frontal suture should provide a clear window into the intracranial anatomical structures at least until this stage of pregnancy.
The brief acquisition time associated with transfrontal 3D imaging allowed us to reduce acquisition artifacts and provided clear and direct visualization of the median structures, with anatomical details comparable to those of the median sagittal plane obtained through the anterior fontanelle. The simple volume display proposed enabled rapid assessment of the midline structures, easily evaluated using TUI. By displaying multiple sagittal ‘slices’ simultaneously, TUI allows an immediate visual correlation of the midline cerebral structures, improving understanding of the anatomy24. Moreover, offline analysis of the volume allowed us to improve tissue contrast using static VCI. We have noticed that the application of this filter frequently leads to greater precision in the definition of the landmarks of structures such as the cerebellar vermis and the fourth ventricle15, 16.
While often the secondary fissure of the cerebellar vermis was not clearly demonstrated, perhaps due to the parallel orientation of the ultrasound beam accessing from the frontal sutures, the relationship of the cerebellar vermis to the brain stem was consistently well defined (Figure 4). 3D sagittal access cannot assess the integrity of the cavum septi pellucidi, but, from the sagittal plane displayed by TUI, the selection of ‘C’ as a reference image displays a coronal section. To maintain uniform orientation of the fetal head, the plane should be rotated along the z-axis. This tomographic coronal section allows visualization of the leaves that demarcate the borders of the cavum septi pellucidi, thus displaying the integrity of the structure (Figure 5). Sagittal slices obtained using TUI are similar to those created by computed tomography or magnetic resonance imaging, and can be understood easily by a pediatric neurologist or neuroradiologist, facilitating a second opinion and counseling.
Although the real potential of transfrontal 3D acquisition cannot be determined at this stage, the quality of the images obtained and its capacity to assess the integrity of the median fetal cerebral structures are all promising advantages. We believe that, in the near future, visualization of the sagittal plane will become an integral part of screening for fetal anomalies, and that this volumetric methodology is one step towards incorporating a comprehensive fetal neuroscan into routine targeted organ evaluation.