Pediatr Radiol (2000) 30: 856±860 Ó Springer-Verlag 2000

Stefano Binaghi François Gudinchet Benedict Rilliet

Received: 20 March 2000 Accepted: 24 April 2000

S. Binaghi ´ F. Gudinchet Department of Radiology, University Hospital of Lausanne, CHUV, 1011 Lausanne, Switzerland

)

S. Binaghi ( ) Service de Radiodiagnostic et Radiologie Interventionnelle, Rue du Bugnon, 1011 Lausanne, Switzerland e-mail: [email protected] Tel.: + 41-21-3 14-44-78 Fax: + 41-21-3 14-44-43 B. Rilliet Department of Neurosurgery, University Hospital of Lausanne, 1011 Lausanne, Switzerland

Three-dimensional spiral CT of craniofacial malformations in children

Abstract Objective. To assess the value of three-dimensional CT (3D CT) in the diagnosis and management of suspected paediatric craniofacial malformations. Materials and methods. Twentyeight children (12 girls, 16 boys) with a mean age of 4 years, suffering from craniofacial or cervical malformations, underwent craniofacial spiral CT. 3D reformatting was performed using an independent workstation. Results. 3D CT allowed the preoperative evaluation of 16 patients with craniosynostosis and the post-surgical management of 2 patients. 3D CT clearly depicted malformations of the skull base involving the petrous bone in seven patients (four

Introduction CT has evolved continuously since its introduction into medical imaging in the early 1970 s. The newest generation of CT, spiral CT, provides a technology which allows for continuous data acquisition and improved 3D reconstruction. Although there are increasing numbers of publications evaluating this technology in the adult for vascular, neck, chest and abdominal imaging, there are few reports, to date, which address its role in paediatrics [l]. The main indications for 3D CT in children focus on imaging the skull, orbit, temporal bone and neck structures [1]. Demand from neurosurgeons and paediatric surgeons is for the preoperative 3D imaging of malformations such as craniostenosis, craniofacial dysraphism, bone dysostoses, craniofacial fractures and craniofacial infections [2].

cases of Goldenhar-Gorlin syndrome, one case of Treacher-Collins syndrome and two cases of Crouzon's disease). Four patients with craniofacial clefts were also evaluated. Radiological findings were confirmed by the clinical and intraoperative findings in all patients that underwent surgical treatment. Movement artefacts and ªLego effectº related to abrupt change of cranial vault border were encountered and are discussed. Conclusions. 3D CT of the skull can safely and reliably identify paediatric craniofacial malformations involving bone, and it should be used as morphological mapping to help the surgeon in planning surgical treatment.

The aim of the present study is to evaluate the role of 3D CT in the preoperative evaluation of paediatric craniofacial malformations diagnosed by plain radiographs.

Materials and methods Patients Twenty-eight paediatric patients with craniofacial malformations were evaluated with spiral 3D CT over a 5-year-period. The patient group consisted of 16 boys and 12 girls, with an age range from 8 days to 15 years (mean 4 years). Patients were referred for evaluation of nonsyndromic and syndromic craniosynostosis, craniofacial clefts, Treacher-Collins syndrome and Goldenhar-Gorlin syndrome.

857

CT data acquisition Spiral 3D CT was performed using a General Electric High Speed Advantage CT scanner. The craniofacial scan protocol consisted of three contiguous imaging volumes. The inferior volume covered from the inferior edge of the mandible or maxilla to the inferior orbital edge using 125 mAs and 3- or 5-mm slice thickness. The middle volume covered from the inferior orbital edge to the superior border of the frontal sinus, using 85 mAs and 1-mm or 3-mm slice thickness. The superior volume covered the remaining bone structures to the skull vault using 125 mAs and 3-mm slice thickness. Fixed parameters were set as follows: 512 ” 512 matrix, 120 kV and pitch of 1.0. The processor required approximately 3 min to generate between 90 and 110 axial sections with 1-mm nominal thickness, 3-mm nominal thickness at 1-mm intervals (2-mm overlap), and 5-mm nominal thickness at 3-mm intervals (2-mm overlap), respectively. Each spiral acquisition took 25±30 s to perform. Table feed (1±5 mm/s), slice thickness (1±5 mm) and increment (1±2 mm) were adapted in selected cases. Thermoluminescent dosimetry results on our scanner when used at 120 kV showed a mean dose of 0.1 mGy/mAs; 3D craniofacial spiral CT using 160 mAs achieved a dose of 16 mSv. Children under 6 years of age required sedation; general anaesthesia was used in a few patients, with examination times varying between 30 min and 1 h, including anaesthetic induction and recovery time.

Fig. 1 Spiral CT shaded surface display (SSD) showing synostosis of the sagittal suture (arrow) in a 6-month-old boy

CT post-processing All 3D CT studies were reconstructed by an operator (S. B.) on an off-line workstation (Advantage Windows, General Electric, Milwaukee, Wis., USA). The bony structures of the skull and face were segmented by thresholding and visualized by a shaded surface display (SSD) technique [3]. In order to display only the bony structures in the model, the lower threshold of attenuation value was set at 160 HU. The 3D segmented skull and face were displayed using a volume rendering projection [4]. The anatomical features of skull and face were evaluated by rotating the 3D model. The three-dimensional processing of the CT model with the volume rendering technique mentioned above required less than 5 min.

a

Results Twenty-eight children were evaluated. In 15 patients, simple synostosis was identified, divided into 6 cases of unilateral lambdoid or coronal synostosis (plagiocephaly), 4 cases of sagittal synostosis (scaphocephaly or dolichocephaly; Fig. 1), 3 cases of bilateral coronal synostosis (brachycephaly) and two cases of metopic synostosis (trigonocephaly). Nine children had compound craniosynostosis in association with other craniofacial malformations: four cases of Goldenhar-Gorlin syndrome (a combination of facial, ear, eye and vertebral malformations; Fig. 2), 3 cases of Crouzon's disease (craniosynostosis, maxillary hypoplasia, shallow orbits and proptosis; Fig. 3), 1 case of Treacher Collins syndrome (mandibulofacial dysostosis characterized by craniosynostosis, maxillary and mandibular hypoplasia and drooping of the outer inferior orbital rim; Fig. 4) and 1 case of acroceph-

b Fig. 2 a, b 3D SSD of the facial bones in a 6-year-old child with Goldenhar-Gorlin syndrome. a There is absence of the right mandibular ramus (large arrowhead) and external auditory canal (small arrowheads). b The zygomatic arch is absent (arrowhead)

aly (all sutures or coronal suture plus one other). Craniofacial clefts (according to the Tessier classification) were also identified in 4 patients [13, 14]. 3D CT confirmed the clinical diagnosis in all of the above patients and provided morphological topography of the abnormal skull or face. This technique provided adequate images of brain and cerebrospinal fluid spa-

858

a a

b

b Fig. 3 a, b 3D craniofacial SSD in a 2-day-old baby with compound craniostenosis secondary to Crouzon's syndrome. a There is bilateral incomplete coronal synostosis associated with compensatory widening of the metopic (arrow) and temporo-parietal (small arrows) sutures. b This appearances in this case are atypical of Crouzon's syndrome because the patient also shows incomplete sagittal synostosis and compensatory widening of the lambdoid suture

ces, allowing the diagnosis of agenesis of the corpus callosum in one patient. In patients with Goldenhar-Gorlin syndrome a precise evaluation of the associated middle ear abnormalities was obtained. The diagnostic confidence for each abnormality, that is the degree of certainty assigned to the diagnosis [5], was subjectively determined as definite, probable or possible. The level of confidence was highly dependent on CT image quality [5]. In three patients the ªLego effectº artefact (characterised by square edges at slice interfaces at the top of the skull; Fig. 5) [6] was present. With the use of 1±2-mm overlap between the imaging volumes using 1-mm and 3-mm slice thickness, empty space artefact was not produced on 3D reconstructions.

Fig. 4 SSD of the facial bones in a 4-year-old child with TreacherCollins syndrome showing a a hypoplastic zygomatic arch (arrow) and b a hypoplastic mandible with retrognathism and absent external auditory canal (arrow)

With zero-degree gantry tilt there were no tilt-related artefacts. A low-dose technique using 85 mAs was utilised for the middle imaging volume centred on the orbital region to protect the lens. This did not result in loss of information regarding the bony skull structure.

Discussion Craniosynostoses were first described by Virchow in 1851 as a variety of abnormal growth patterns of the skull following premature fusion of cranial sutures. Craniosynostoses are currently classified as either simple synostoses, involving fusion of a single suture, or compound synostosis, involving fusion of two or more sutures [7]. In primary craniosynostosis the underlying brain is generally normal, whereas in secondary craniosynostosis abnormal brain development is present, most commonly microcephaly [8]. Surgical treatment of primary craniosynostosis achieves better cosmetic results during the first 2 years of life, because most of the brain growth occurs during

859

Fig. 5 The ªLego effectº manifested as squaring of the interfaces between each slice at the vertex where the contour of the skull is rapidly changing. This 11-year-old child had previously undergone surgical treatment of Crouzon's disease, which had manifested as bilateral coronal synostoses, brachycephalic anterior cranial vault with shallow hyperteloric orbits

that time, allowing reconfiguration of the misshaped skull after the operation opening the fused sutures. In a child older than 2 years of age, the lack of brain growth cannot induce such reconfiguration [8]. It is therefore believed that in children with craniosynostosis, radiological investigation should be undertaken early and actively [15, 16, 17]. Several complex craniofacial syndromes were also described consisting of various combinations of maxillary or mandibular hypoplasia and ear dysplasia with deafness (i. e. Crouzon's syndrome, Apert's syndrome, Treacher-Collins syndrome, Goldenhar-Gorlin syndrome). As emphasized by McAlister [9], we were aware in our study of the two main pitfalls in making the diagnosis of simple synostosis, particularly the posterior deformational plagiocephaly and sticky or lazy sutures [10, 11, 12]. In our series, 3D CT allowed precise description of the suture edges and all simple synostoses in our study were confirmed at surgery. We encountered several technical problems in performing 3D CT of craniofacial disorders in children [18]. Patient movement during data acquisition produced unsharpness of the two-dimensional image and lack of continuity in this region on the 3D model. When more than one acquisition was performed in obtaining 3D CT craniofacial model, the same X and Y centring coordinates and the same zoom factor were used throughout.

The setting of Hounsfield Units threshold value in segmentation of the 3D bone model had a dramatic effect on depicting anatomical details of bone and on identifying bone disease correctly. A high threshold value induced misdiagnosis by producing false widening of sutures and erosion of thin plates of bone. A low threshold value produced artefactual closure of sutures and a false-positive diagnosis. Therefore, our standard threshold value for 3D segmentation of bone was set at 180 HU. Because gantry tilt may affect the 3D reconstruction causing distortion or elongation artefacts [6], we worked without gantry tilt in performing CT acquisitions. Our observers achieved best diagnostic performance with 3D volumetric images; axial CT images were less useful for synostosis evaluation. We considered, however, standard CT for the evaluation of cerebral abnormalities or other special morphologic features involving the cerebral parenchyma, the orbits and paranasal sinuses. Thermoluminescent dosimetry results on our Hispeed Advantage CT scanner used at 120 kV showed a mean dose of 0.1 mGy/mAs. A 3D spiral using 160 mAs achieved a dose of 16 mSv; further reduction to 85 mAs can reach a mean dose of 8.5 mSv allowing effective protection to the lens. These mean doses can be better appreciated if compared to the dose received from plain skull radiography of around 6.7 mSv [8]. This suggests that avoiding plain films in the radiological assessment of these patients could dramatically reduce the total dose. These results also suggest that in the radiological evaluation of craniofacial malformations, 3D CT of skull at 85 mAs can complement PA and lateral skull plain films, giving better diagnostic performance without excessively increased radiation exposure. In conclusion, our study showed that 3D reconstructed volumetric CT gives precise spatial depiction of paediatric craniofacial malformations involving bone without delivering excessive radiation dose, particularly to the lens. Plain radiography and CT complemented each other in patient management. Plain radiography acted as a screening tool to make the diagnosis of a craniofacial malformation; axial CT assessed all sites of bony bridging in craniosynostosis and its direct consequences or pathologic associations on underlying brain parenchyma and CSF spaces. Finally, 3D CT has been shown to be an important tool to assist spatial orientation in planning the surgical strategy, and for this reason it should be used to evaluate patients requiring reconstructive surgery.

860

References 1. Coleman LT, Zimmerman RA (1994) Pediatric craniospinal CT: current applications and future potential. Semin Ultrasound CT MR 15: 148±155 2. Plumley DA, Grosfeld JL, Kopecky KK, et al (1995) The role of spiral computerized tomography with three-dimensional reconstruction in pediatric solid tumors. J Pediatr Surg 30: 317±321 3. Napel S, Marks MP, Rubin GD, et al (1992) CT angiography with spiral CT and maximum intensity projection. Radiology 185: 607±610 4. Rigauts H, Marchal G, Baert AL, et al (1990) Initial experience with volume CT scanning. J Comput Assist Tomogr 14: 675±682 5. Pilgram TK, Vannier MW, Hildebolt CF, et al (1989) Craniosynostosis: image quality, confidence, and correctness in diagnosis. Radiology 173: 675±679

6. Craven CM, Naik KS, Blanshard KS, et al. (1995) Multispiral three-dimensional computed tomography in the investigation of craniosynostosis: technique optimization. Br J Radiol 68: 724±730 7. Benson ML, Oliverio PJ, Yue NC, et al (1996) Primary craniosynostosis: imaging features. AJR 166: 697±703 8. Fernbach SK (1998) Craniosynostosis 1998: concepts and controversies Pediatr Radiol 28: 722±727 9. McAlister WH (1998) Invited commentary: posterior deformational plagiocephaly. Pediatr Radiol 28: 727±728 10. Kane AA, Mitchell LE, Craven KP, et al (1996) Observations on a recent increase in plagiocephaly without synostosis. Pediatrics 97: 877±885 11. (1992) American Academy of Pediatrics Task Force on Infant Positioning and SIDS. Pediatrics 89: 1120±1126 12. Cohen MM Jr (1995) Lambdoid synostosis is an overdiagnosed condition. Am J Med Genet 61: 98±99

13. Tessier P (1976) Anatomical classification of facial, cranio-facial and laterofacial clefts. J Maxillofac Surg 4: 69±92 14. David JD, Moore MH, Cooter RD (1989) Tessier clefts revisited with a third dimension. Cleft Palate J 26: 163±185 15. Leboucq N, Montoya Y, Martinez P, et al (1992) Pathology of the base of the skull in children-3D imaging. Ann Radiol (Paris) 35: 424±429 16. Klein HM, Bertalanffy H, Mayfrank L, et al (1994) Three-dimensional spiral CT for neurosurgical planning. Neuroradiology 36: 435±439 17. Vannier MW, Hildebolt CF, Marsh JL, et al (1989) Craniosynostosis: diagnostic value of three-dimensional CT reconstruction. Radiology 173: 669±673 18. Fishman EK, Magid D, Ney DR, et al (1991) State of the art: three dimensional imaging. Radiology 181: 321±337