Virtual reconstruction of midface defects using statistical shape models
Introduction
Computer-assisted surgery (CAS) is well established, with a wide field of applications. CAS is based on computed tomography (CT)/cone-beam computed tomography (CBCT) imaging data for three-dimensional visualization and virtual planning (Heiland et al., 2004, Aschendorff et al., 2009, Levine et al., 2012, Metzger et al., 2013, Strong et al., 2013, Wilde et al., 2014, Cornelius et al., 2015, Scolozzi, 2015). Intraoperative CT scanning is an auxiliary tool to immediately analyze and refine the outcome during the surgical intervention (Heiland et al., 2003, Heiland et al., 2005). Other improvements in CAS are the introduction of cutting guides and patient specific implants (PSI) which enable surgeons to pre-plan and pre-engineer a surgical procedure (Succo et al., 2015, Wilde et al., 2015).
Advantages and benefits of CAS are well-described in the literature. In trauma or after tumor ablation, a virtual reconstruction is usually performed preoperatively by segmenting an unaffected side and mirroring it to the defective side (Gellrich et al., 2002, Schmelzeisen et al., 2004). This information can be forwarded to the implant engineer. This pathway has been shown to enhance the accuracy of bony reconstructions (Schepers et al., 2015).
The CAS workflow is criticized for having too many logistical steps and manipulations (Cornelius et al., 2015). The high costs, potential high expense and extra time lead to reluctance towards using this technology. It should be made more readily available so that even surgeons untutored in specialized computer programs could integrate it into their clinical practice.
The main element to process patient data in CAS relies on segmentation procedures (Metzger et al., 2013). Since the Hounsfield scale range of an imaging modality is crucial for the segmentation procedure, the quality of a data set may be setting limitations to the processing. With the selection of specific Hounsfield scale subranges, the visualization and segmentation of different tissues becomes possible. This procedure was a manual step only 10 years ago. Recently a semi-automatic method has been developed that uses atlas-segmentation (Metzger et al., 2013). A virtual template model of the entire skull and its anatomical subregions is deformed to fit onto the individual dataset of a patient. The information of the atlas is transformed to meet with the individual's conditions (Metzger et al., 2013).
Mirroring the affected to the unaffected side has been used as a gold standard for the facial reconstruction within the last decades (Gellrich et al., 2002, Schmelzeisen et al., 2004). Image registration by using mirrored and registered templates or skulls is a further step towards automated CAS (Wagner et al., 2015). Based on the idea of a symmetrical skull, the injured area is virtually resected and the uninjured skull is mirrored to the injured site. The virtual reconstructed skull is registered by computer algorithms, such as diffeomorph registration as a part of console application of ANTS onto the original skull (Avants et al.,, Wagner et al., 2015). This procedure is limited by the assumption of a symmetrical skull (Schmelzeisen et al., 2004). As described by different authors, the skull is rarely symmetrical (Letzer and Kronman, 1967, Shah and Joshi, 1978, Farkas and Cheung, 1981; Kim et al., 2003). A statistical model, created by CT scans of uninjured skulls, considers asymmetrical and individual properties of the skull.
In this publication, a statistical model of the midface as a key element for refined and semi-automated CAS procedures in craniomaxillofacial surgery is introduced. This technology has the ability to overcome previous problems and to expand current possibilities into new workflows.
Section snippets
Material and methods
This study was performed after obtaining institutional ethics committee approval (450/15). It consisted of three individual and sequential steps. First, a statistical shape model (SSM) of the midfacial region was created by using CT data. In a second step, a uniform defect was created within the right zygomatico-orbital ensemble in a series of 10 skulls. Subsequently, these zygomas were virtually reconstructed using two methods. As a last step, the accuracy of the different methods of
Results
All defective zygomas in the test skulls could be reconstructed with both methods. The analysis showed significant differences in the accuracy among groups. On the affected side, the mean deviation in group I was 1.10 mm ± 0.23 mm and in group II it was 0.85 mm ± 0.26 mm to the original bone. The differences between each group were significant for the affected side (P = 0.001) (Table 1).
Discussion
Since the early 1990s, computer-assisted surgery (CAS) has been gaining ground, with an ever increasing number of technologies. Furthermore, the logistic effort for usage has increased significantly, with the move from virtual planning to more and more patient-specific hardware such as cutting guides and implants. Many studies have proven versatility, facilitated approaches and shown better clinical outcomes, thus indicating progress in the evolution of new surgical standards (Ellis, 1990,
Conclusion
SSM as an additional element could complement and facilitate the CAS workflow in many respects. By comparing this procedure with the present procedure of mirroring, it has been demonstrated to be more accurate. Therefore, there is a need for further studies to better identify the possibilities and limitations of the SSM technology.
Funding
No support of grants.
Conflict of interest statement
The authors declare that they have no conflict of interest.
References (37)
- et al.
Patient-specific reconstruction plates are the missing link in computer-assisted mandibular reconstruction: a showcase for technical description
J Craniomaxillofac Surg
(2015) 3rd: accuracy of model surgery: evaluation of an old technique and introduction of a new one
J Oral Maxillofac Surg
(1990)- et al.
Intraoperative imaging of zygomaticomaxillary complex fractures using a 3D C-arm system
Int J Oral Maxillofac Surg
(2005) - et al.
Asymmetry of the sphenoid bone and its suitability as a reference for analyzing craniofacial asymmetry
Am J Orthod Dentofacial Orthop
(2003) - et al.
Verification of clinical precision after computer-aided reconstruction in craniomaxillofacial surgery
Oral Surg Oral Med Oral Pathol Oral Radiol Endod
(2007) - et al.
Accuracy of fibula reconstruction using patient-specific CAD/CAM reconstruction plates and dental implants: a new modality for functional reconstruction of mandibular defects
J Craniomaxillofac Surg
(2015) - et al.
Navigation-aided reconstruction of medial orbital wall and floor contour in cranio-maxillofacial reconstruction
Injury
(2004) Computer-aided design and computer-aided modeling (CAD/CAM) generated surgical splints, cutting guides and custom-made implants: which indications in orthognathic surgery?
Rev Stomatol Chir Maxillofac Chir Orale
(2015 Dec)- et al.
Three-dimensional treatment planning of orthognathic surgery in the era of virtual imaging
J Oral Maxillofac Surg
(2009) - et al.
Development and first clinical application of automated virtual reconstruction of unilateral midface defects
J Craniomaxillofac Surg
(2015)
Computer-assisted orthognathic surgery: feasibility study using multiple CAD/CAM surgical splints
Oral Surg Oral Med Oral Pathol Oral Radiol
A paradigm shift in orthognathic surgery? A comparison of navigation, computer-aided designed/computer-aided manufactured splints, and “classic” intermaxillary splints to surgical transfer of virtual orthognathic planning
J Oral Maxillofac Surg
Radiologically assisted navigation in cochlear implantation for X-linked deafness malformation
Cochlear Implant Int
Facial asymmetry in healthy North American Caucasians. An anthropometrical study
Angle Orthod
Computer-assisted secondary reconstruction of unilateral posttraumatic orbital deformity
Plast Reconstr Surg
Intraoperative 3D imaging of the facial skeleton using the SIREMOBIL Iso-C3D
Dentomaxillofac Radiol
3D-imaging of the facial skeleton with an isocentric mobile C-arm system (Siremobil Iso-C3D)
Dentomaxillofac Radiol
Cited by (36)
Development and validation of a novel craniofacial statistical shape model for the virtual reconstruction of bilateral maxillary defects
2024, International Journal of Oral and Maxillofacial SurgeryHyM3D: A hybrid method for the automatic 3D reconstruction of a defective cranial vault
2023, Computer Methods and Programs in BiomedicineCranial reconstruction evaluation - comparison of European statistical shape model performance on Chinese dataset
2022, Bone ReportsCitation Excerpt :All statistical analysis, as well as image registration and shape modeling were performed using the open-source statistical/mathematical platform R – and more specifically the R-packages ANTsR, Morpho, Rvcg, mesheR and RvtkStatismo (Goodall, 1991; Lüthi et al., 2009, 2012; R Development Core Team, 2011; Schlager and Rüdell, 2015). We used the same SSM of the cranial vault of European adults as in (Fuessinger et al., 2018; Semper-Hogg et al., 2017, p.) that has been is generated from 131 craniofacial CT scans of pathologically unaffected patients (n = 61 female, n = 70 male patients, average age = 53.2 years). The average slice thickness of the CT scans was 1.2 mm.
Deep learning for cranioplasty in clinical practice: Going from synthetic to real patient data
2021, Computers in Biology and MedicineMedical image segmentation in oral-maxillofacial surgery
2021, Computer-Aided Oral and Maxillofacial Surgery: Developments, Applications, and Future PerspectivesEvaluation of the Midface by Statistical Shape Modeling
2021, Journal of Oral and Maxillofacial SurgeryCitation Excerpt :Both report on the feasibility but did not present quantitative information on the specific model, gender differences, symmetry, or bone thickness. A second use of SSM has been suggested by Fuessinger et al14 and Semper-Hogg et al15 to use SSM to reconstruct missing parts of the midface in virtual surgical planning. As this was a report on feasibility, too, no data to compare their model with our model were given.