Journal of Medical Systems

, Volume 34, Issue 3, pp 281–290 | Cite as

Differentiation of Two Subtypes of Adult Hydrocephalus by Mixture of Experts

  • Elif Derya Übeyli
  • Konuralp Ilbay
  • Gul Ilbay
  • Deniz Sahin
  • Gur Akansel
Original Paper

Abstract

This paper illustrates the use of mixture of experts (ME) network structure to guide model selection for diagnosis of two subtypes of adult hydrocephalus (normal-pressure hydrocephalus–NPH and aqueductal stenosis–AS). The ME is a modular neural network architecture for supervised learning. Expectation-Maximization (EM) algorithm was used for training the ME so that the learning process is decoupled in a manner that fits well with the modular structure. To improve classification accuracy, the outputs of expert networks were combined by a gating network simultaneously trained in order to stochastically select the expert that is performing the best at solving the problem. The classifiers were trained on the defining features of NPH and AS (velocity and flux). Three types of records (normal, NPH and AS) were classified with the accuracy of 95.83% by the ME network structure. The ME network structure achieved accuracy rates which were higher than that of the stand-alone neural network models.

Keywords

Mixture of experts Expectation-maximization algorithm Normal-pressure hydrocephalus Aqueductal stenosis Classification accuracy 

References

  1. 1.
    Kestle, J. R. W., In Batjer, H. H., & Loftus, C. M. (Eds.), Textbook of Neurological Surgery. Philadelphia: Lippincott Williams Wilkins, 871–875, 2003.Google Scholar
  2. 2.
    Klinge, P. M., In Ramina, R., Aguiar, P. H. P., & Tatagiba, M. (Eds.), Samii’s Essentials in Neurosurgery. Berlin: Springer-Verlag, 249–258, 2008.Google Scholar
  3. 3.
    Fukuhara, T., and Luciano, M. G., Clinical features of late-onset idiopathic aqueductal stenosis. Surgical neurology. 55:132–137, 2001. doi:10.1016/S0090-3019(01)00359-7.CrossRefGoogle Scholar
  4. 4.
    Little, J. R., Houser, O. W., and MacCarty, C. S., Clinical manifestation of aqueductal stenosis in adults. Journal of neurosurgery. 43:546–552, 1975.CrossRefGoogle Scholar
  5. 5.
    Tissel, M., Tullberg, M., Hellstrom, P. et al., Neurological symptoms and signs in adult aqueductal stenosis. Acta neurologica Scandinavica. 107:311–317, 2003. doi:10.1034/j.1600-0404.2003.00124.x.CrossRefGoogle Scholar
  6. 6.
    Thompson, D., In: Moore, A. J., & Newel, D. W., (Eds.), Neurosurgery. London: Springer-Verlag, 425–442, 2005.Google Scholar
  7. 7.
    Anik, Y., Demirci, A., Anik, I. et al., Apparent diffusion coefficient and cerebrospinal fluid flow measurements in patients with hydrocephalus. Journal of computer assisted tomography. 32:392–396, 2008. doi:10.1097/RCT.0b013e31812f4edc.CrossRefGoogle Scholar
  8. 8.
    Kim, D., Choi, J. U., Huh, R. et al., Quantitative assessment of cerebrospinal fluid hydrodynamics using a phase-contrast cine MR image in hydrocephalus. Childs Nerv Syst. 15:461–467, 1999. doi:10.1007/s003810050440.CrossRefGoogle Scholar
  9. 9.
    Stoquart-El Sankari, S., Lehmann, P., Gondry-Jouet, C. et al., Phase-Contrast MR imaging support for the diagnosis of aqueductal stenosis. AJNR American journal of neuroradiology. Oct.2, 2008 [Epub ahead of print].Google Scholar
  10. 10.
    Nitz, W. R., Bradley, W. G. Jr., Watanabe, A. S. et al., Flow dynamics of cerebrospinal fluid: assessment with phase-contrast velocity MR imaging performed with retrospective cardiac gating. Radiology. 183:395–405, 1992.Google Scholar
  11. 11.
    Enzmann, D. R., Pelc, N. J., 14:1301–1307, 1993; discussion 1309–1310.Google Scholar
  12. 12.
    Bradley, W. G. Jr., Whittemore, A. R., Kortman, K. E. et al., Marked cerebrospinal fluid void: indicator of successful shunt in patients with suspected normal-pressure hydrocephalus. Radiology. 178:459–466, 1991.Google Scholar
  13. 13.
    Takayuki, S., Osamu, Y., Keizo, T. et al., Analysis of cerebrospinal fluid flow in the aqueduct using cine phase-contrast imaging Fourier analysis and anew technique to correct maxwell term phase errors. Prog Computed Imaging. 23:33–41, 2001.Google Scholar
  14. 14.
    de Marco, G., Idy-Peretti, I., Didon-Poncelet, A. et al., Intracranial fluid dynamics in normal and hydrocephalic states: systems analysis with phase-contrast magnetic resonance imaging. Journal of computer assisted tomography. 28:247–254, 2004. doi:10.1097/00004728-200403000-00015.CrossRefGoogle Scholar
  15. 15.
    Schroeder, H. W., Schweim, C., Schweim, K. H. et al., Analysis of aqueductal cerebrospinal fluid flow after endoscopic aqueductoplasty by using cine phase-contrast magnetic resonance imaging. Journal of neurosurgery. 93:237–244, 2000.CrossRefGoogle Scholar
  16. 16.
    Barkhof, F., Kouwenhoven, M., Scheltens, P. et al., Phase-contrast cine MR imaging of normal aqueductal CSF flow. Effect of aging and relation to CSF void on modulus MR. Acta radiologica. 35:123–130, 1994.Google Scholar
  17. 17.
    Parkkola, R. K., Komu, M. E., Aärimaa, T. M. et al., Cerebrospinal fluid in children with normal and dilated ventricles studied by MR imaging. Acta radiologica. 42:33–38, 2001. doi:10.1080/028418501127346431.CrossRefGoogle Scholar
  18. 18.
    Gideon, P., Ståhlberg, F., Thomsen, C. et al., Cerebrospinal fluid flow and production in patients with normal pressure hydrocephalus studied by MRI. Neuroradiology. 36:210–215, 1994. doi:10.1007/BF00588133.CrossRefGoogle Scholar
  19. 19.
    Gideon, P., Sørensen, P. S., Thomsen, C. et al., Assessment of CSF dynamics and venous flow in the superior sagittal sinus by MRI in idiopathic intracranial hypertension: a preliminary study. Neuroradiolgy. 36:350–354, 1994. doi:10.1007/BF00612116.CrossRefGoogle Scholar
  20. 20.
    Parkkola, R. K., Komu, M. E. S., Kotilainen, E. M. et al., Cerebrospinal fluid flow in patients with dilated ventricles studied with MR imaging. European radiology. 10:1442–1446, 2000. doi:10.1007/s003300000376.CrossRefGoogle Scholar
  21. 21.
    Sharma, A. K., Gaikwad, S., Gupta, V. et al., Measurement of peak CSF flow velocity at cerebral aqueduct, before and after lumbar CSF drainage, by use of phase-contrast MRI: utility in the management of idiopathic normal pressure hydrocephalus. Clinical neurology and neurosurgery. 110:363–368, 2008. doi:10.1016/j.clineuro.2007.12.021.CrossRefGoogle Scholar
  22. 22.
    Egeler-Peerdeman, S. M., Barkhof, F., Walchenbach, R. et al., Cine phase-contrast MR imaging in normal pressure hydrocephalus patients: relation to surgical outcome. Acta neurochirurgica Supplementum (Wien). 71:340–342, 1998.Google Scholar
  23. 23.
    Luetmer, P. H., Huston, J., Friedman, J. A. et al., Measurement of cerebrospinal fluid flow at the cerebral aqueduct by use of phase-contrast magnetic resonance imaging: technique validation and utility in diagnosing idiopathic normal pressure hydrocephalus. Neurosurgery. 50:534–543, 2002. doi:10.1097/00006123-200203000-00020.CrossRefGoogle Scholar
  24. 24.
    Gideon, P., Thomsen, C., Ståhlberg, F. et al., Cerebrospinal fluid production and dynamics in normal aging: a MRI phase-mapping study. Acta neurologica Scandinavica. 89:362–366, 1994.CrossRefGoogle Scholar
  25. 25.
    Mase, M., Yamada, K., Banno, T. et al., Quantitative analysis of CSF flow dynamics using MRI in normal pressure hydrocephalus. Acta neurochirurgica. Supplementum (Wien). 71:350–353, 1998.Google Scholar
  26. 26.
    Lee, J. H., Lee, H. K., Kim, J. K. et al., CSF flow quantification of the cerebral aqueduct in normal volunteers using phase contrast cine MR imaging. Korean Journal of Radiology. 5:81–86, 2004.CrossRefGoogle Scholar
  27. 27.
    Übeyli, E. D., Comparison of different classification algorithms in clinical decision-making. Expert systems. 24:117–31, 2007. doi:10.1111/j.1468-0394.2007.00418.x.CrossRefGoogle Scholar
  28. 28.
    Übeyli, E. D., Wavelet/mixture of experts network structure for EEG signals classification. Expert systems with applications. 34:31954–1962, 2008. doi:10.1016/j.eswa.2007.02.006.CrossRefGoogle Scholar
  29. 29.
    Übeyli, E. D., Combining neural network models for automated diagnostic systems. Journal of medical systems. 30:6483–488, 2006. doi:10.1007/s10916-006-9034-z.CrossRefGoogle Scholar
  30. 30.
    Übeyli, E. D., A mixture of experts network structure for breast cancer diagnosis. Journal of medical systems. 29:5569–579, 2005. doi:10.1007/s10916-005-6112-6.CrossRefGoogle Scholar
  31. 31.
    Übeyli, E. D., Implementing wavelet transform/mixture of experts network for analysis of electrocardiogram beats. Expert Systems. 25:2150–162, 2008. doi:10.1111/j.1468-0394.2008.00444.x.CrossRefGoogle Scholar
  32. 32.
    Übeyli, E. D., Modified mixture of experts for diabetes diagnosis. Journal of medical systems, 2008 (in press).Google Scholar
  33. 33.
    Übeyli, E.D., Adaptive neuro-fuzzy inference systems for automatic detection of breast cancer. Journal of medical systems, 2008 (in press).Google Scholar
  34. 34.
    Übeyli, E. D., & Doğdu, E., Automatic detection of erythemato-squamous diseases using k-means clustering. Journal of medical systems, 2008 (in press).Google Scholar
  35. 35.
    Jacobs, R. A., Jordan, M. I., Nowlan, S. J., and Hinton, G. E., Adaptive mixtures of local experts. Neural computation. 3:179–87, 1991. doi:10.1162/neco.1991.3.1.79.CrossRefGoogle Scholar
  36. 36.
    Chen, K., Xu, L., and Chi, H., Improved learning algorithms for mixture of experts in multiclass classification. Neural networks. 12:91229–1252, 1999. doi:10.1016/S0893-6080(99)00043-X.CrossRefGoogle Scholar
  37. 37.
    Hong, X., and Harris, C. J., A mixture of experts network structure construction algorithm for modelling and control. Applied intelligence. 16:159–69, 2002. doi:10.1023/A:1012869427428.MATHCrossRefGoogle Scholar
  38. 38.
    Jordan, M. I., and Jacobs, R. A., Hierarchical mixture of experts and the EM algorithm. Neural computation. 6:2181–214, 1994. doi:10.1162/neco.1994.6.2.181.CrossRefGoogle Scholar
  39. 39.
    Mangiameli, P., and West, D., An improved neural classification network for the two-group problem. Computers & operations research. 26:5443–460, 1999. doi:10.1016/S0305-0548(98)00076-8.MATHCrossRefGoogle Scholar
  40. 40.
    Hu, Y. H., Palreddy, S., and Tompkins, W. J., A patient-adaptable ECG beat classifier using a mixture of experts approach. IEEE transactions on biomedical engineering. 44:9891–900, 1997. doi:10.1109/10.623058.CrossRefGoogle Scholar
  41. 41.
    Haykin, S., Neural networks: A Comprehensive Foundation. Macmillan, New York, 1994.MATHGoogle Scholar
  42. 42.
    Chaudhuri, B. B., and Bhattacharya, U., Efficient training and improved performance of multilayer perceptron in pattern classification. Neurocomputing. 34:11–27, 2000. doi:10.1016/S0925-2312(00)00305-2.MATHCrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2008

Authors and Affiliations

  • Elif Derya Übeyli
    • 1
  • Konuralp Ilbay
    • 2
  • Gul Ilbay
    • 3
  • Deniz Sahin
    • 3
  • Gur Akansel
    • 4
  1. 1.Department of Electrical and Electronics Engineering, Faculty of EngineeringTOBB Ekonomi ve Teknoloji ÜniversitesiAnkaraTurkey
  2. 2.Departmant of Neurosurgery, Faculty of MedicineKocaeli UniversityKocaeliTurkey
  3. 3.Departmant of Physiology, Faculty of MedicineKocaeli UniversityKocaeliTurkey
  4. 4.Departmant of Radiology, Faculty of MedicineKocaeli UniversityKocaeliTurkey

Personalised recommendations