Editorial review: pediatric 3D ultrasound

Abstract

Three-dimensional ultrasound is an established diagnostic imaging technique in many specialties. However, in neonates, infants and children three-dimensional ultrasound still is underutilized, partially due to time constraints for post-processing and restricted availability, of devices as well as dedicated pediatric transducers. Also reimbursement issues still need to be addressed. This editorial review presents more or less established pediatric three-dimensional ultrasound applications with proven diagnostic benefit as well as potential future applications of three-dimensional/four-dimensional ultrasound in infants and children, aiming at enhancing research and promoting practical use of three-dimensional ultrasound in relevant pediatric conditions. Particularly, applications in neonatal neurosonography, ultrasound of the urogenital tract as well as some other small part and miscellaneous queries are highlighted. Additional other potential and future indications are discussed briefly, also mentioning restrictions and potential future developments. In summary, three-dimensional ultrasound holds some potential to widen sonographic diagnostic capabilities throughout childhood and hopefully will be increasingly investigated and introduced into clinical practice provided respective equipment and pediatric three-dimensional/four-dimensional ultrasound transducers become available.

Introduction

Since three-dimensional ultrasound (3DUS) has started to enter clinical use decades ago efforts have been made to evaluate its usability and potential for pediatric imaging⁽¹⁾. In many fields, particularly in obstetric and prostate imaging but also for assessment of the breast and many other organs mostly in adults, 3DUS has become part of the routine with proven benefits for diagnosis and management^(2–19). However, partially due to restricted availability of dedicated pediatric probes and the somewhat more difficult handling in pediatrics 3DUS has yet not gained a comparable role in neonates, infants and children. This editorial review tries to highlight potential useful applications in children, also mentioning potential future promising applications.

Technique and benefits of 3DUS

Since its beginning, the basic approach of 3DUS has not changed that much for most of its clinically available devices. It is based on a series of 2D images combined with some sort of position information retrieved either from the transducerinherent geometry or by external positioning devices (mostly using electromagnetic or infrared based tracking systems). These data sets are then reconstructed and thereafter all kind of viewing and manipulation tools can be applied to the data set: multi-planar reconstructions in any desirable plane, surface or volume rendering, or data segmentation. Furthermore, recently promoted approaches offer fusion techniques that allow to combine real-time ultrasound (US) images with CT- or MR-data sets for navigating in the respective volume, particularly useful for biopsies. A relatively new technique is the use matrix transducers that allow to simultaneously scan an entire volume, however, only a few vendors offer this technique and there still are restriction with the range of available transducers and conspicuous viewing options – therefore this approach is still mostly used for echocardiography. Fast frame rates achievable with this technique have significantly improved the option to study moving objects such as the heart valves; this four-dimensional US (4DUS) – with time being the fourth dimension – holds promising future potential. Additionally to these improved new viewing and display options other benefits 3D/4DUS include improved volume calculation for irregularly shaped objects, better documentation with great potential for comparison to other imaging techniques, counselling and second opinion, and medico-legal aspects^(20–28). However, a number of 3DUS artefacts that are imported into the 3DUS data set and may cause confusing images as well as 3DUS specific artefacts have to be respected, recognized and considered⁽²⁹⁾.

Pediatric applications

Neurosonography

The use of 3DUS for assessing the fetal brain has been reported in several papers^(30–32). The same principles may be applied to neonates, too; using the open fontanel as access for the data acquisition even more details can be depicted with dedicated pediatric transducers^(33–36). Neonatal brain 3DUS offers a reproducible data set, ideal for comparing to other sectional imaging such as CT or MR. Furthermore, by clearly delineating the borders of some more complex shaped structures such as the ventricular system, 3DUS has been shown beneficial for assessment of the respective (ventricular) volume^(37–42). The initially long time for manual segmentation has been overcome by introducing intuitive and reliable segmentation options (fig. 1). The option to reconstruct any desirable plane, particularly the diagnostically often crucial axial plane, from the data set implies great potential for improving detailed anatomic assessment – in some cases not only improving, but enabling a definitive diagnosis⁽⁴²⁾. Furthermore, rendering of complex and tortuous structures such as twisting vessels can be helpful for detailed analyses of their course and shape – again improving diagnostic potential (fig. 2).

Fig. 1.

3DUS of the neonatal brain: viewing options and benefits. Multi-planar view of a “cerebral cyst.” The three orthogonal axes allow for superior localisation and definition of this tiny cerebral cyst defined by the tiny dot: the axial reconstructed view (left lower box) superiorly confirms that its position thus enabling differentiation of a para-ventricular leucomalacic cyst from an intraventricular plexus cyst in this preterm infant – with significant prognostic implications

Fig. 2.

3DUS of cerebral vessels using aCDS data. 3DUS including aCDS data: the rendered view conspicuously allows for an angiography-like overview the large vein of Galen vascular malformation with its major feeding and draining vessels

The main setback in pediatric 3DUS is the restricted availability of suitable transducers that offer sufficient resolution in the near and far field at reasonable frame rates with acceptable range of frequencies, as basic physics of US cannot be overcome by 3DUS. This means: only the image quality you can achieve with the two-dimensional ultrasound (2DUS) acquisition can eventually be transferred into the 3DUS data set, where inhomogeneity throughout the image, poor lateral resolution, and high or inconsistent image noise significantly deteriorate the quality and usability. Furthermore, the free-hand 3DUS option (using an estimate of the transducer translation for reconstructing the 3DUS data set from a 2DUS image series) as offered in some less expensive equipment is very cumbersome, particularly when trying to acquire a useful 3DUS volume through a small fontanelle. Finally, data manipulation and viewing is somewhat time-consuming and not well accepted by clinicians as mostly pediatricians and neonatologists are performing these scans in some countries. Furthermore, neonatal brain US is often performed at the bedside – often mobile small devices are preferred, which often do not offer 3DUS options. And in infants and older children only a transcranial/transtemporal approach is visible with inherently restricted image quality; however, reports exist on its usefulness particularly when looking at cerebral vessels using a power Doppler acquisition^(42–44). But the lack of adequate transducers and high quality Doppler offered by the devices usually in use for these applications have hindered an increased clinical use, particularly as color Doppler 3DUS (due to inherent restrictions such as blurry images with restricted special resolution, incomplete accessibility of the entire course of the imaged vessels) cannot compete with the diagnostically still necessary CTA, MRA or catheter angiography procedures eventually becoming necessary. Future options in pediatric neurosonography include new creative approaches that may combine various imaging and viewing techniques and inter-operative image fusion approaches. One interesting application would be assessment of the brain surface using surface rendering techniques (fig. 3). This might improve US potential to assess gyration alterations and could prove helpful for early bedside diagnosis of a number of conditions presently underestimated by US. Another option would be combining (amplitude coded) color Doppler or even contrast-enhanced (Doppler) sonography with fast repetitive 3DUS acquisitions, allowing for dynamic perfusion imaging. This could improve US potential to diagnose stroke in the early face or for improving assessment of tumor entity and treatment response. Furthermore dynamic 4DUS of the spinal cord might be useful for assessing and documenting spinal cord motion disorders such as early detection of tethering (only applicable in the first month of live as long as US can access the spinal canal and the spinal cord is visible). 3DUS of the orbit may also offer some options; however, a higher resolution of respective data sets will be necessary for achieving clinical importance. Then surface viewing of the inner eye surface and the retina might be valuable for diagnosis and follow-up of small retinoblastoma under treatment, and the improved volume calculation of 3DUS may be beneficial for assessing, e.g., a tumor volume. It furthermore may offer an option for intermittent follow-up due to its improved comparability to MR by providing the ability to reconstruct analogous sections for direct comparison.

Fig. 3.

Brains 3DUS – rendering options in 3DUS: A, B. thick slab rendering for viewing the para-ventricular area demonstrating periventricular echogenicities in a preterm infant (A) and mild cystic PVL in an axial thick slab rendered reconstruction (B); C. surface rendering for viewing gyration; D. rendering of dilated ventricles – inverted view

Pediatric urogenital 3DUS

The urogenital tract is one of the most common targets of a pediatric US investigation. Many of the very first pediatric 3DUS reports have looked into its potential for assessing the pediatric urinary bladder and kidney. These observation initially focused on the high accuracy of 3DUS in volume calculations; by being able to segment and outline even complex shaped structures such as the collecting system in a hydronephrotic kidney or the inner counter of an irregularly shaped urinary bladder, even these volumes can be calculated with high accuracy, whereas the conventional 2DUS measurement based estimates (using geometrical equations of the resembling diametrical form) suffer from a high inaccuracy to up to ±70% error in irregularly shaped organs. Therefore 3DUS have been proposed for assessment of a bladder volume⁽⁴⁵⁾. And for assessment of real renal parenchymal volume in hydronephrotic kidneys – becoming feasible by deducting the segmented collecting system from the overall renal volume thus giving a reliable number for the size of the renal parenchyma. Comparing both kidneys with each other, split renal size can then be calculated – these 3DUS based results give comparable results to split renal size in scintigraphy or MR-urography that well compare to the respective function, provided there are no acute perfusion or function restriction (fig. 4)^(46–50). Using the 3DUS data set and its multi-axial reconstruction options differential diagnosis of complex pathology can be improved, such as detecting the tortures neck of a caliceal diverticula difficult follow on conventional 2DUS⁽⁵⁰⁾. Applying the various rendering options the conspicuous display with a more intuitive comprehension of shape and spatial relation, e.g. of the calices, may improve the diagnostic yield: differentiation of a dysplastic from a dilated collecting system is easier, the pelvi-ureteric junction – particularly if the ureter also has some dilatation – is nicely demonstrated, and the relation of vessels with the draining urinary system can be superiorly demonstrated (fig. 5). Applying surface rendering to the 3DUS data set the inner bladder surface can be displayed allowing for a virtual cystography^(51–54). This may improve depiction and demonstration of surface irregularities such as trabeculation or small ureteroceles and diverticula; furthermore a convincing depiction of the bladder trigone with the ostia and the bladder neck, demonstration of polyps and tumors is feasible (fig. 6). Restrictions of urogenital and any other abdominal 3DUS in infants and small children is their reduced cooperability: breath hold assessments are difficult to achieve, thus fast acquisitions in times of little motion (particularly no relevant motion form breathing) are necessary to get good image quality and reliable calculation results. This necessitates some experience of the investigator – using fast acquisitions in moments with little respiratory movement (e.g. a sleeping baby) and targeting the investigations to that part of the respiratory cycle with little motion (i.e. second phase of expiration) are tricks that will enable good 3DUS results even in the pediatric abdomen. Of course all other rules, tips and tricks for performing pediatric abdominal US apply as well; these are an indispensable prerequisite for reliable and reproducible 3DUS results.

Fig. 4.

Renal 3DUS in hydronephrosis: segmentation of collecting system and renal parenchymal volume calculation. 3DUS for renal parenchymal volume assessment in different 3DUS systems: different segmentation options for delineating the outer renal contours and segmenting the collecting system for deducting the segmented dilated pelvo-caliceal system from the entire renal volume allowing for reliable calculation of renal parenchymal volume and split renal size estimation. A. Serial steps with showing the segmented data using a multi-planar view with rendered (inverted) image of the segmented data in the right lower box of two different infants with hydronephrosis; the first one demonstrating the entire kidney volume, the second image the delineated dilated collecting system. B. Single reconstructed coronal slice, serial images the kidney in an infant with hydronephrosis: initial image, segmented renal parenchyma, and inverted view of the segmented dilated collecting system fused with segmented parenchyma in the right box

Fig. 5.

3DUS visualization of dysplastic dilated calices by inverted rendering. 3DUS of a kidney with dilated pelvo-caliceal system: the rendered gray-scale inverted view (right lower box) enables convincing perception of the dyplastic configuration rendering this system more a dysplastic than an obstructed collecting system

Fig. 6.

Bladder 3DUS: virtual cystoscopy. Virtual 3DUS based cystoscopy based on surface rendering (right lower box) superiorly demonstrates the bladder calculus in a child with cystinuria

At present 3DUS of the female inner genitalia has only been found useful for assessment for uterine anomalies as in adults (fig. 7)^{(9, 16, 17)}. Maybe a combination of 3D/4DUS with sonogenitography after saline filling of various structures in genital malformations may improve the conspicuity and diagnosis in these queries. This, however, will rely – just as perineal 3DUS – on adequate high resolution 3DUS transducers that offer sufficient detail information for pediatric needs. Although 3DUS can be easily applied to the scrotum, there is little additional information to 2DUS and thus it is not used routinely.

Fig. 7.

3DUS for uterine anomalies – crucial coronal plane. Coronal reconstruction of a 3DUS of an infants’ uterus: the crucial reconstructed coronal plane enables a definite definition and diagnosis this uterine malformation/duplication – this is much more difficult or even impossible to classify on conventional 2DUS sections

Potential future options include combination of (amplitude coded) color Doppler data or contrast-enhanced US with 3DUS or even 4DUS, thus potentially improving detection and documentation of scares or allowing for assessment of perfusion disturbances. 4DUS of the urinary bladder will allow a real time virtual cystoscopy – with great potential for assessment of functional changes. It may further improve the detection of the ostia by visualizing opening and closing of the ostia or the ureteric inflow jet, observation of bladder neck activities may help detecting bladder dysfunction. This may imply great clinical usability for example by helping to properly select those who need to undergo reflux assessment thus reducing the number of unnecessary exams and associated burden to the affected children. Finally, 4DUS – potentially combined with image fusion techniques – may improve detection and characterization of urogenital tumors and improve accuracy and handling for sonographically guided biopsies.

3DUS of the child's small part and musculoskeletal system

There are a number of incidental reports on potential applications in the pediatric musculoskeletal system and small parts^(55–57). However, no consistent and systematic research has been performed to evaluate 3DUS potential in these areas. All reports focus on the known 3DUS abilities, such as the improved volume calculation accuracy of particularly irregularly shaped structures. This can be performed for various tumors and lesions, and can help measuring volumes of organs with a somewhat complex shape such as the thyroid gland^{(58, 59)}. Here 3DUS has not only a higher accuracy than 2DUS, but also allows to propose new correction factors for the 2DUS based calculations. This on the other hand would need adaptation of conventionally used normal volume charts for US size assessment. The conspicuous visualization of tortuous and complex structures by 3DUS let to applications for viewing complex vascular lesions such as vascular malformations, the complex course of winding vessels such as in carotid kinking, or depicting fistula tracts in subcutaneous and other soft tissue abscesses and collections (fig. 8)^(60–63).

Fig. 8.

3DUS with CDS: superficial vessels in an arteriovenous malformation. 3DUS of a gluteal vascular malformation: the rendered view (right lower box) superiorly shows the dimension and vascularity of this vascular malformation with multiple vessels shunts, far less conspicuously demonstrated by the 2DUS source images

Rendering of the bony surface enables to conspicuously view fractures, particularly useful in situations with equivocal findings on plane film (e.g. patella fracture, skull fractures, sternal and rip fractures; fig. 9)^{(61, 62)}. By being able to reconstruct crucial planes not assessable by conventional 2DUS, such as the coronal plane in breasts and other tumors, spiculated traction of tissues around a tumor has become depictable sonographically, and thus differentiation of tumor entities may become more accurate. Furthermore, 3DUS of the pelvic floor, the anal canal and the particularly female urethra has improved sonographic diagnostic potential for assessment of ruptures and tears, as mostly reported for adults⁽⁶³⁾ (fig. 10). Many other soft tissue and musculoskeletal applications are feasible, for example the assessment of rip anomalies using reconstruction tools to conspicuously view the shape of also the cartilaginous parts which difficult to assess by plain film (fig. 11). Again, the restrictions are not only the immanent limitations of any sonographic diagnosis but also the difficult handling with sometimes bulky and heavy transducers, and the restricted availability of these transducers and technique in places where many of these routine or emergency US examinations are performed in neonates, infants and children.

Fig. 9.

3DUS rendering – skull fracture and sutures. The surface rendered view of the skull improves perception and differentiation of skull sutures (A) or a skull fracture (B) by conspicuously demonstrating shape and course

Fig. 10.

Pelvic floor 3DUS – tomographic serial coronal reconstructions. 3DUS of the pelvic floor from a perineal access: the coronal reconstructions perfectly demonstrate the anatomy of the perineum/pelvic floor in a girl – all relevant anatomic structures such as the labia, the clitoris or the urethral opening can be conspicuously displayed in a girl wit urogenital sinus. Note the absence of a clear separate vaginal opening

Fig. 11.

3DUS rending for conspicuous visualization of rib anomaly. The thick slap coronal rendered 3DUS view (right lower box) superiorly demonstrates the branching of the ventral cartilaginous aspect of the ribs in a rib malformation non visible by plain film.

Future options include not only similar approaches as described with the other reported applications – based on surface rendering, dynamic 4DUS, improved transducer technology with higher resolution, more conspicuous viewing and fast calculations, or optimized handling and rendering options – but also including data from contrast-enhanced US, or expanding US elastography into the plane perpendicular to the US beam thus being able to assess the shear wave velocity and behavior in all directions, potentially improving depiction of lesion and characterization of respective findings.

Pediatric 3D-echocardiography

Some reports exist on the benefits of 3D/4DUS in assessing the neonatal and pediatric heart^(64–67). Again no systematic research has been undertaken to evaluate the possible benefits, which are improved calculation of volumes (similar to MR volume calculation of ventricular muscular or luminal volume, e.g. for improved systolic and diastolic volumes, with a better assessment of the ejection fraction) and the more conspicuous viewing of complex anatomy particularly when assessing valves (fig. 12). It has been shown that 3D/4DUS of the pediatric heart is feasible and that challenges from the high heart rate in neonates can be successfully met. One of the problems still is the inability to include directional Doppler data into the 3DUS data set and applying adequate angle correction for individual areas of the data set.

Fig. 12.

Pediatric 4D/3DUS of the heart (Courtesy of Dr. B. Nagel, Dept. of Pediatrics, Division of Pediatric Cardiology, University Hospital Graz): A. umbrella for closing an atrioseptal defect – 3DUS rendered surface view from inside the atrium towards the septum with the device in place; B, C. heart valve – two images from a dynamic 4DUS clip: the leaflets of the mitral valve can be seen open (B) and then closed (C) using a surface rendered view

Tips and tricks

As mentioned above most systems require a defined minimum time without motion of the targeted volume for a successful 3DUS acquisition. Only the modern two-dimensional matrix volume transducer are fast enough to overcome this problem. Post-processing options for motion correction are presently only available for cardiac imaging (“stick technology”). They could be expanded to other areas where regular motion occurs and can be tracked for retrospective gated reconstruction. Till than it depends on the sonographers skill to acquire the volume in a situation where there is little motion, which first needs to create a cooperative patient or stabilizing the area scanned. Simple tricks, such as stabilizing by positioning with supporting pillows or using restrainment devices (as used for CT, MR or trauma handling) may be an option. This, however, will not overcome motion from respiration, heart beat and respective vascular pulsations – it will only be overcome using very fast acquisitions or motion correction techniques. The respiratory movement can be minimized by trying to acquire the 3DUS data set in that phase of the respiratory cycle which has the littlest movement, usually the second part of the expiration. Furthermore, a sonographic window has to be found which enables a sufficient acquisition of the entire targeted region, as reconstructions and particularly volume calculations only can live up to the expectations when all the borders of the targeted organ are properly included. The device must be set properly to 3DUS-adequate 2DUS presets; usually high frame rates with a rather crisp image, a relatively low dynamic range and hard post-processing will deliver the best images. Sometimes it is better to use a slightly lower frequency with thus less image noise in spite of the reduced resolution, as presently voxel size in the 3DUS data set is large and very high resolution cannot be supported. However, harmonic imaging and speckle reduction filters usually are very helpful, whereas image compounding only reduces frame rate without significant improvement of the 3DUS data. Of course, all other tips and tricks as applied to scanning pediatric patients are helpful, such as a quiet room, distracting toys and pets, films and books, music, pacifier, a bottle with tea, warm US gel, heater, blankets and cushions as well as helping personnel. It furthermore is helpful if an external workstation will allow for data processing and reconstruction as well as rendering and viewing, as this speeds up the investigation making it easier to have a quiet and cooperative patient, who usually cannot stand a long investigation without becoming more motorically active. This will take pressure from the examiner as they can fiddle around with the data set once the patient has left. So all they need to do is to perform the requested 2DUS study and then quickly acquire those 3DUS data sets which appear to be interesting or diagnostically useful – leaving all further assessment and handling for the reading at the external workstation. Of course, adequate storage systems must be available which still is a problem with many PACS (picture archiving and communication systems) systems, as there still is a big variation in 3DUS data standards, even if first efforts of standardizing DICOM formats also for 3D/4DUS data sets.

Conclusion

3DUS is a maturing technique with numerous established applications in adults, particularly for obstetric imaging. However, in spite of many potentially useful applications in neonates, infants and children, it has yet not reached clinical routine in pediatric sonography. Further research is necessary to establish its value, and dedicated transducers and equipment must become available at reasonable cost for eventually being able to exploit the vast potential of 3D/4DUS in childhood. Then this technique may probably further contribute to reducing radiation burden by improving US diagnosis rendering other imaging less often necessary.

Conflict of interest

The author is member of the Kretz/GE European advisory for 3DUS, and has ongoing cooperation with Kretz, GE and Siemens for development and clinical adaptations of pediatric US devices.