Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
TECHNICAL REPORT
An approach for three-dimensional visualization using
high-resolution MRI of the temporomandibular joint
Y Hayakawa*,1, C Kober2, M Otonari-Yamamoto1, T Otonari1, M Wakoh1 and T Sano1
1
Department of Oral and Maxillofacial Radiology, Tokyo Dental College, Chiba, Japan; 2Faculty of Engineering and Computer Science,
University of Applied Science, Osnabrück, Germany
Objectives: To visualize the temporomandibular joint (TMJ) and the surrounding tissues in detail
utilizing high-resolution MR images for the diagnosis of soft- and hard-tissue abnormalities.
Clinically routine MR slices are processed by tissue segmentation and three-dimensional (3D)
reconstruction and viewed with visualization software.
Materials and methods: A 1.5 T MRI system was used. The double-echo procedure for taking
oblique sagittal images was applied to obtain both proton density-weighted (PDW) and T2 weighted
(T2W) images simultaneously, with separate examinations in both open and closed mouth positions.
Diagnosis of the abnormality in the placement and morphology of articular discs and the joint
effusion status is usually performed using multiple MRI slices. Clinically routine continuous MR
slices were processed by segmentation, reconstruction and visualization algorithms, and the
mandibular condyle, fossa, articular disc and other intra-articular tissues were visualized on the 3D
and two-dimensional (2D) – 3D fusion images.
Results: In a clinical case, the anterior disc displacement without reduction, with mouth open
and closed, was clearly depicted in the 3D images. Also 2D – 3D superposed images with
changeable tissue transparency successfully depicted the stereoscopic TMJ morphology in three
dimensions.
Discussion and conclusion: High-resolution PDW- and T2W MR images could be processed by
tissue segmentation and 3D-reconstruction procedures, and the resultant images showed the
anatomical details in an easily recognizable way. By the simultaneous visualization of both bony
surfaces and soft tissues, disc displacement and deformity can be recognized in a 3D context. The
additional superposition of the 3D visualization with the original 2D MR slices allows for a
combination with conventional diagnostics.
Dentomaxillofacial Radiology (2007) 36, 341–347. doi: 10.1259/dmfr/12894471
Introduction
The temporomandibular joint (TMJ) is one of the most tool for soft tissue abnormalities of the TMJ.1 – 8 In addition,
complicated joints in the human body. It consists of the MRI is equal to or possibly better than standard tomography
mandibular fossa above, the mandibular condyle below and for the detection of cortical bone components despite the
the TMJ capsule with articular disc and several tissues in limited visualization of bone.9 Therefore, MRI has replaced
between them. Pathological changes of the TMJ often cause CT and arthrographic examinations of the TMJ.1 – 9
painful disorders. For that reason, TMJ imaging is subject to For medical three-dimensional (3D) image rendering, the
intensive research. MRI has evolved as the prime diagnostic shaded surface reconstruction is applied based on manual or
semi-automatic segmentation. Referring to this kind of
research, several attempts at 3D representation of the TMJ
*Correspondence to: Yoshihiko Hayakawa, Department of Computer Sciences,
Faculty of Engineering, Kitami Institute of Technology, 165 Koen-cho, Kitami,
based on MR images have been carried out to assess the 3D
Hokkaido 090-8507, Japan; E-mail: hayakawa@cs.kitami-it.ac.jp morphology and dynamics.10 – 13 Segmentation consists of
Received 23 May 2006; revised 13 August 2006; accepted 31 August 2006 reduction of the considered images to binary data separated
3D visualization of TMJ
342 Y Hayakawa et al
into black and white according to a user defined strategy. If software (Ver. 3.1.1; Mercury Computer Systems, Berlin,
tissue differentiation is sufficiently clear, as for bony organs in Germany, http://amira.zib.de/, http://www.amiravis.com/).15
CT data for instance, the shaded surface reconstruction is of
generally accepted benefit. However, as regards TMJ-MRI 3D surface reconstruction of mandibular condyle and
data of 3 mm slice thickness, the inherent spatial information fossa
about the articular disc is very scarce. Furthermore, the exact After preparatory image processing of the MRI slices
tissue classification is unclear. This possibly makes the through interactive editing, especially contour filtering, the
segmentation and rendering processes tedious, error-prone mandibular condyle and fossa were semi-automatically
and highly user dependent. segmented in the 2D MR slices. In order to overcome the
The purpose of this article is to describe an alternative slice thickness of 3 mm and the additional gap of 20%
approach to 3D rendering of the mandibular condyle and (0.6 mm) the segmented slices were subjected to cubic
fossa, articular disc and several tissues that compose the interpolation to nearly isotropic voxel size. After gaussian
TMJ. For the sake of future application in the medical flattening of the segmented labels, the built-in Amira
routine of diagnosis and simulation, the method has been surface generation algorithm was applied. Finally, the
designed to be appropriate for processing TMJ-MRI resulting surface model was improved by stepwise
data stemming from clinical routine practice. The scarcity simplification and smoothing.16
of spatial anatomical information caused by the high slice
thickness is overcome by a combined application of Rendering of intraarticular condition of the TMJ including
different visualization techniques. The method is demon- the articular disc
strated for high-resolution MRI. For rendering the articular disc in 3D context, the voxels
comprising the lateral pterygoid muscle and the space
between the head of the condyle and the glenoid fossa were
Materials and methods segmented in the 2D MRI slices. Thereafter, they were
isolated from the rest of the data set and resampled to
MR image acquisition nearly isotropic voxel size. Only these voxels containing
MRI was carried out using a 1.5 T MRI system (Magnetom the information about the intra-articular condition of the
Symphony Maestro Class; Siemens, Erlangen, Germany), TMJ, especially the articular disc and the retrodiscal tissue,
with the use of bilateral surface coils, “double loop array,” were subjected to direct volume rendering, which is a very
as receivers. MR image acquisition by double loop array, intuitive method for visualizing 3D scalar fields. Each
with a size of 70 mm in diameter, is optimized for the TMJ point in a data volume is assumed to emit and absorb light.
region. As the clinical routine MRI-TMJ examination, The amount and colour of emitted light and the amount of
oblique sagittal cross-sectional plane images, which are absorption is determined from the scalar data by using a
oriented according to the individual angle of the mandibular colour map which includes alpha values. With regard to the
condyle, are obtained as described by Musgrave et al.14 The applied colour map, there is a wide variety of possibilities
examination to obtain oblique sagittal images is repeated at of monochromatic, dichromatic, or polychromatic colour
both open and closed mouth positions. The slice thickness in maps. The articular disc is depicted by low signal in both
each image was 3 mm, which is standard in clinical routine. the PDW images and the T2W images. For the sake of
To reduce the cross-talk artefact, caused by the interference emphasis on the disc, the voxels comprising the lateral
between slices, there are 20% gaps between slices. There- pterygoid muscle, and the space between the head of the
fore either seven or eight cross-sectional images are taken condyle and the glenoid fossa, were inverted. The range of
for the sagittal examination of each TMJ region. rendered grey values can be manipulated by the lower and
At those kinds of imaging slices, proton density-weighted upper bound of the colour scale. By this, it is possible to put
images (PDWI) and T2 weighted images (T2WI) are the focus on tissue characterized by low signal such as the
simultaneously obtained by the double-echo sequence.6 articular disc, for instance, or on tissue characterized by
Settings of repetition time (TR)/echo time (TE) were high signal such as joint effusion.
3300 ms/14 ms (PDWI) and 3300 ms/85 ms (T2WI). The Finally, the intraarticular tissues, namely the lateral
matrix was 265 £ 265 and the field-of-view (FOV) was pterygoid muscle and the space between the head of
150 mm. The total examination time for the double-echo condyle, the fossa and the lateral pterygoid muscle’s
sequences was less than 6 min. Two-dimensional (2D) MR tendon (direct volume rendering) the mandibular condyle
slices were provided for the diagnosis of positional and (shaded surface rendering) and the glenoid fossa (trans-
morphological abnormalities of articular discs and mandib- parent surface rendering) were simultaneously visualized.
ular condyle, the absence or presence of the joint effusion etc.
Dentomaxillofacial Radiology
3D visualization of TMJ
Y Hayakawa et al 343
Dentomaxillofacial Radiology
3D visualization of TMJ
344 Y Hayakawa et al
Dentomaxillofacial Radiology
3D visualization of TMJ
Y Hayakawa et al 345
Dentomaxillofacial Radiology
3D visualization of TMJ
346 Y Hayakawa et al
Figure 6 Upper row: three-dimensional (3D) reconstruction of the left temporomandibular joint (TMJ) based on T2W images. The volume rendering of
TMJ with superposed 3D-reconstructed condyle (left). The 3D reconstruction was superposed with a T2WI slice (right). Lower row: 3D reconstruction of
the left TMJ based on proton density-weighted (PDW) images The volume rendering of TMJ with superposed 3D-reconstructed condyle (left). The 3D
reconstruction was superposed with a PDWI slice (right)
Figure 7 Three-dimensional (3D) reconstruction of the left temporomandibular joint (TMJ) based on T2W images at the open mouth examination. The
volume rendering of TMJ with superposed 3D-reconstructed condyle (left). The 3D reconstruction was superposed with a T2WI slice (right). Arrows:
reversed marked high signal of the articular disc
Dentomaxillofacial Radiology
3D visualization of TMJ
Y Hayakawa et al 347
the original slices, there is no user-dependent step in the In conclusion, by the simultaneous visualization of
method. Though still a prototype, the application of the both bony surfaces and soft tissues, disc displacement
method is not very time consuming. Furthermore, it can and deformity can be recognized in a 3D context. The
be easily automated. The clinical diagnosis in the case additional superposition of the 3D visualization with the
presented showed no changes after these 3D tools were original 2D MR slices allows for a combination with
applied, but its impact for visualization of the articular conventional diagnostics.
disc displacement, disc reduction and disc deformity in
the clinical practice can be emphasised. We also plan
3D imaging to contribute to a biomechanical/biomedical
simulation study of the jaw bone/mandible.19 – 21 The Acknowledgments
research will be oriented to the optimization and The second author would like to thank the International Bureau of
automation of the tissue segmentation for clinical the German Federal Ministry of Education and Research for a
relevance and validation as diagnostic and biodynamic grant for a short-term visit to Tokyo Dental College in the first
simulation tools for jaw function. days of February 2006.
References
1. Katzberg RW, Westesson P-L, Tallents RH, Anderson R, Kurita K, 13. Goessi D. On the relationship between functional properties and
Manzione JV Jr, et al. Temporomandibular joint: MR assessment of anatomical structure of the temporomandibular joint. PhD Thesis,
rotational and sideways disk displacements. Radiology 1988; 169: Swiss Fed Inst Tech Zurich, Switzerland, 2004.
741 –748. 14. Musgrave MT, Westesson P-L, Tallents RH, Manzione JV, Katzberg
2. Sano T. Recent developments in understanding temporomandibular RW. Improved magnetic resonance imaging of the temporomandib-
joint disorders. Part 1: bone marrow abnormalities of the mandibular ular joint by oblique scanning planes. Oral Surg Oral Med Oral
condyle. Dentomaxillofac Radiol 2000; 29: 7 –10. Pathol 1991; 71: 525 –528.
3. Sano T. Recent developments in understanding temporomandibular 15. Stalling D, Westerhoff M, Hege H-C. Amira: A highly interactive
joint disorders. Part 2: changes in the retrodiscal tissue. Dentomax- system for visual data analysis. In: Hansen CD, Johnson CR (eds).
illofac Radiol 2000; 29: 260 –263. The visualization handbook. Amsterdam, The Netherlands: Elsevier,
4. Sano T, Yamamoto M, Okano T. Temporomandibular joint: MR 2005, 38: pp 749 –767.
imaging. Neuroimag Clin North Am 2003; 12: 583 – 595. 16. Kober C, Sader R, Zeilhofer H-F. 3D-reconstruction and visualization
5. Yamamoto M, Sano T, Okano T. Magnetic resonance evidence of of bone mineral density of the ethmoid bone. In: Buzug TM, Lueth
joint fluid with temporomandibular joint disorders. J Comput Assist TC (eds). Perspective in image-guided surgery. Singapore: World
Tomogr 2003; 27: 694 – 698. Scientific, 2004, pp 490 – 496.
6. Sano T, Widmalm SE, Yamamoto M, Sakuma K, Araki K, Matsuda Y, 17. Erickson SJ, Cox IH, Hyde JS, Carrera GF, Strandt JA, Estkowski
et al. Usefulness of proton density and T2-weighted vs. T1-weighted LD. Effect of tendon orientation on MR imaging signal intensity: a
MRI in diagnoses of TMJ disk status. Cranio 2003; 21: 253 – 258. manifestation of the “Magic Angle” phenomenon. Radiology 1991;
7. Sano T, Yamamoto M, Okano T, Gokan T, Westesson P-L. Common 181: 389 – 392.
abnormalities in temporomandibular joint imaging. Curr Probl Diagn 18. Sano T, Westesson P-L, Yamamoto M, Okano T. Differences in
Radiol 2004; 33: 16 –24. temporomandibular joint pain and age distribution between marrow
8. Larheim TA, Westesson P-L. Maxillofacial imaging. Berlin, edema and osteonecrosis in the mandibular condyle. Cranio 2004; 22:
Germany: Springer, 2006. 283 – 288.
9. Laskin DM. Diagnosis of pathology of the temporomandibular joint. 19. Kober C, Erdmann B, Lang J, Sader R, Zeilhofer H-F. Sensitivity of
Clinical and imaging perspectives. Radiol Clin North Am 1993; 31: the temporomandibular joint capsule for the structural behaviour of
135 –147. the human mandible. Biomed Eng 2004; 49: 372 – 373.
10. Price C, Connell DG, MacKay A, Tobias DL. Three-dimensional 20. Kober C, Erdmann B, Hellmich C, Sader R, Zeilhofer H-F.
reconstruction of magnetic resonance images of the temporomandib- Consideration of anisotropic elasticity minimizes volumetric rather
ular joint by I-DEAS. Dentomaxillofac Radiol 1992; 21: 148 –153. than shear deformation in human mandible. Computer Methods
11. Motoyoshi M, Sadowsky PL, Bernreuter W, Fukui M, Namura S. Biomechanics Biomed Eng 2006; 9: 91 –101.
Three-dimensional reconstruction system for imaging of the tempor- 21. Kober C, Erdmann B, Lang J, Sader R, Zeilhofer H-F. Adaptive finite
omandibular joint using magnetic resonance imaging. J Oral Sci element simulation of the human mandible using a new physiological
1999; 41: 5– 8. model of the masticatory muscles. PAMM 2004; 4: 332 –333.
12. Chirani RA, Jacq JJ, Meriot P, Roux C. Temporomandibular joint: a
methodology of magnetic resonance imaging 3-D reconstruction.
Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2004; 97:
756 – 761.
Dentomaxillofacial Radiology