Automatic 3D segmentation of individual facial muscles using unlabeled prior information
- 202 Downloads
Segmentation of facial soft tissues is required for surgical planning and evaluation, but this is laborious using manual methods and has been difficult to achieve with digital segmentation methods. A new automatic 3D segmentation method for facial soft tissues in magnetic resonance imaging (MRI) images was designed, implemented, and tested.
A region growing algorithm based on local energy functions, using intensity similarities among neighboring regions as criteria, was developed. This local energy function includes the neighborhood relationships not only in the same dataset but from other training datasets. This approach differs from previous studies where the prior information was represented as manual segmented atlases. In this study, a consensus of many datasets, none of which was labeled a priori, is used to guide the segmentation. The method was tested in MRI scans for adult facial structures. MRI scans were obtained from the Alzheimer’s disease neuroimaging initiative database. Comparison was made to results from expert manual segmentation and region growing techniques.
The volumetric overlap between automated 3D segmentation results and the ground truth was 82.6% for masseter and 78.8% for temporalis tissues.
A new automated method to segment various facial soft tissues was implemented and the results compared with standard region growing results. The proposed method shows 3.9% improvement in accuracy over the standard method. Reduction in segmentation errors was consistently achieved in MRI scans.
KeywordsFacial soft tissue Magnetic resonance images Markov random field Prior information Segmentation Region growing
Unable to display preview. Download preview PDF.
- 2.Xia JJ, Gateno J, Teichgraebe J, Christensen A, Lasky R, Lemoine A, Liebschner M (2007) Accuracy of the computer-aided surgical simulation (CASS) system in the treatment of patients with complex craniomaxillofacial deformity: a pilot study. J Oral Maxillofac Surg 65: 248–254PubMedCrossRefGoogle Scholar
- 5.Rohlfing T, Brandt R, Menzel R, Russakoff DB, and Maurer CR (2005) Quo vadis, atlas-based segmentation, chap 11. In: Suri J, Wilson DL, Laxminarayan S (eds) The handbook of medical image analysis: registration models, vol 3. Kluwer Academic/Plenum Publishers, New York, pp. 435–486Google Scholar
- 9.Noble JH, Warrenb FM, Labadie RF, Dawant BM (2008) Automatic segmentation of the facial nerve and chorda tympani using image registration and statistical priors. In: Proceedings of the SPIE, The International Society for Optical Engineering, pp 69140P-1–69140P-10Google Scholar
- 13.Li SZ (2009) Markov random field modeling in image analysis (Advances in pattern recognition)Google Scholar
- 15.Bouman CA, Thompson AM, Brown JC, Kay JW (1995) Markov random fields and stochastic image models. IEEE international conference on image processing tutorial. http://dynamo.ecn.purdue.edu/~bouman/publications/Index-Tutorials.html
- 16.Alzheimer’s disease neuroimaging initiative, http://www.loni.ucla.edu/ADNI
- 17.Amira 3-D scientific visualization and data analysis package (ZIB, Berlin, Germany; Indeed–visual concepts GmbH, Berlin, Germany; TGS Inc., San Diego, CA)Google Scholar