Fracture Detection in an MRF-Based Hierarchical Bayesian Framework

  • Ananda S. ChowdhuryEmail author
  • Suchendra M. Bhandarkar
Part of the Advances in Computer Vision and Pattern Recognition book series (ACVPR)


The previous chapter discussed the extraction of fracture points and fracture surfaces in the event of a major fracture where the bone fragments are well separated. In contrast, the focus of this chapter is on the detection of hairline fractures or minor fractures. Mandibular fractures are observed to possess certain distinct patterns in X-ray or CT images. In some cases, the fractures are observed to be hairline or minor in nature. By the terms hairline fracture or minor fracture we refer to those situations where the broken bone fragments are not visibly out of alignment or have incurred very little relative displacement, respectively. The presence of noise makes the detection and subsequent visualization of such types of fractures in X-ray or CT images a very challenging task. In case of a major fracture, i.e., fractures where the broken fragments are clearly displaced relative to each other, surgical intervention is almost mandatory. However, in the case of a hairline fracture or minor fracture, the decision regarding surgical intervention is less clear since the surgeon can rely on natural bone healing for fracture reduction without having to perform reconstructive surgery. In this chapter, we propose a Markov Random Field (MRF)-based hierarchical Bayesian paradigm for detection of hairline or minor fractures and generation of the reconstructed jaw (i.e., target pattern) in such cases. Here, we model the fracture as a local stochastic degradation of a hypothetical intact mandible. In the presence of noise, the detection and subsequent visualization of hairline fractures becomes a clinically challenging task. Furthermore, the decision regarding surgical intervention for this type of fracture is often unclear as a surgeon can choose to rely solely on natural bone healing without any surgical intervention. In addition to aiding in the detection and visualization of the hairline fracture, the generated target pattern depicts how a jaw with a hairline fracture would appear if allowed to heal naturally without explicit surgical intervention. The Bayesian estimation of the mode of the posterior distribution corresponds to the target pattern (i.e., reconstructed jaw), and the differences in intensity between the input data and the MAP estimate at specific pixel locations denote the occurrence and location of a hairline fracture.


Minor Fracture Markov Random Field Outer Contour Target Pattern Block Pair 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 2.
    Ogundare BO, Bonnick A, Bayley N (2003) Pattern of mandibular fractures in an urban major trauma center. J Oral Maxillofac Surg 61(6):713–718 CrossRefGoogle Scholar
  2. 39.
    Gelman A, Carlin JB, Stern HS, Rubin DB (2000) Bayesian data analysis. Chapman & Hall/CRC, London Google Scholar
  3. 40.
    Li SZ (1995) Markov random field modeling in computer vision. Springer, London Google Scholar
  4. 41.
    Geman S, Geman D (1984) Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images. IEEE Trans Pattern Anal Mach Intell 6(6):721–741 CrossRefzbMATHGoogle Scholar
  5. 54.
    Bhandarkar SM, Chowdhury AS, Tang Y, Yu JC, Tollner EW (2007) Computer vision guided virtual craniofacial reconstruction. Comput Med Imaging Graph 31(6):418–427 CrossRefGoogle Scholar
  6. 81.
    Prima S, Ourselin S, Ayache N (2002) Computation of the mid-sagittal plane in 3D brain images. IEEE Trans Med Imaging 21(2):122–138 CrossRefGoogle Scholar
  7. 130.
    Gelman A, Carlin JB, Stern HS, Rubin DB (2004) Bayesian data analysis. Chapman & Hall/CRC, Boca Raton zbMATHGoogle Scholar
  8. 133.
    Berger JO (1985) Statistical decision theory and Bayesian analysis. Springer, New York zbMATHGoogle Scholar
  9. 134.
    Harris CJ, Stephens M (1988) A combined corner and edge detector. In: Proc 4th Alvey vision conference, Manchester, UK, pp 147–151 Google Scholar
  10. 135.
    Besag J (1986) On the statistical analysis of dirty pictures. J R Stat Soc Ser B 48(3):259–302 zbMATHMathSciNetGoogle Scholar
  11. 136.
    Lee SJ, Rangarajan A, Gindi G (1995) Bayesian image reconstruction in SPECT using higher order mechanical models as priors. IEEE Trans Med Imaging 14(4):669–680 CrossRefGoogle Scholar
  12. 137.
    Choi SM, Lee JE, Kim J, Kim MH (1997) Volumetric object reconstruction using the 3-D MRF model-based segmentation. IEEE Trans Med Imaging 16(6):887–892 CrossRefGoogle Scholar
  13. 138.
    Chan AK, Zheng L (2001) An artificial intelligent algorithm for tumor detection in screening mammogram. IEEE Trans Med Imaging 20(7):559–567 CrossRefGoogle Scholar
  14. 139.
    Salli E, Aronen HJ, Savolainen S, Korvenoja A, Visa A (2001) Contextual clustering for analysis of functional MRI data. IEEE Trans Med Imaging 20(5):403–414 CrossRefGoogle Scholar
  15. 140.
    Chen T, Metaxas DN (2004) Markov random field models. In: Yoo TS (ed) Insight into images. AK Peters, Wellesley Google Scholar
  16. 141.
    Richard FJP (2005) A comparative study of Markovian and variational image-matching techniques in application to mammograms. Pattern Recognit Lett 26(12):1819–1829 CrossRefGoogle Scholar
  17. 142.
    Popa C, Zdunek R (2005) Penalized least-squares image reconstruction for borehole tomography. In: Proc algorithmy, Podbanske, Slovakia, pp 260–269 Google Scholar
  18. 143.
    Martin-Fernandez M, Alberola-Lopez C (2005) An approach for contour detection of human kidneys from ultrasound images using Markov random fields and active contours. Med Image Anal 9(1):1–23 CrossRefGoogle Scholar
  19. 144.
    Raj A, Singh G, Zabih R (2006) MRF’s for MRI’s: Bayesian reconstruction of MR images via graph cuts. In: Proc IEEE int cof on computer vision pattern recognition (CVPR), New York, NY, USA, pp 1061–1068 Google Scholar
  20. 145.
    Chowdhury AS, Bhattacharya A, Bhandarkar SM, Datta GS, Yu JC, Figueroa R (2007) Hairline fracture detection using MRF and Gibbs sampling. In: Proc eighth IEEE int workshop on applications of computer vision (WACV), Austin, TX, USA, p 56 CrossRefGoogle Scholar
  21. 146.
    Giannoudis PV, Dinopoulos H (2005) Current concepts of the inflammatory response after major trauma: an update. Injury 36(1):229–230 CrossRefGoogle Scholar
  22. 147.
    Vodovotz Y et al. (2006) In silico models of acute inflammation in animals. Shock 26(3):235–244 CrossRefGoogle Scholar
  23. 148.
    Molina R (1994) On the hierarchical Bayesian approach to image restoration: Applications to astronomical images. IEEE Trans Pattern Anal Mach Intell 16(11):1122–1128 CrossRefGoogle Scholar

Copyright information

© Springer-Verlag London Limited 2011

Authors and Affiliations

  • Ananda S. Chowdhury
    • 1
    Email author
  • Suchendra M. Bhandarkar
    • 2
  1. 1.Department of Electronics & Telecommunication EngineeringJadavpur UniversityKolkataIndia
  2. 2.Department of Computer ScienceThe University of GeorgiaAthensUSA

Personalised recommendations