Multi-modal Categorization of Medical Images Using Texture-based Symbolic Representations

Filip Florea, Eugen Barbu, Alexandrina Rogozan, Abdelaziz Bensrhair

2006

Abstract

Our work is focused on the automatic categorization of medical images according to their visual content for indexing and retrieval purposes in the context of the CISMeF health-catalogue. The aim of this study is to assess the performance of our medical image categorization algorithm according to the image’s modality, anatomic region and view angle. For this purpose we represented the medical images using texture and statistical features. The high dimensionality led us to transform this representation into a symbolic description, using block labels obtained after a clustering procedure. A medical image database of 10322 images, representing 33 classes was selected by an experienced radiologist. The classes are defined considering the images medical modality, anatomical region and acquisition view angle. An average precision of approximately 83% was obtained using k-NN classifiers, and a top performance of 91.19% was attained with 1-NN when categorizing the images with respect to the defined 33 classes. The performances raise to 93.62% classification accuracy when only the modality is needed. The experiments we present in this paper show that the considered image representation obtains high recognition rates, despite the difficult context of medical imaging.

References

  1. Darmoni, S., Leroy, J., Thirion, B., Baudic, F., Douyére, M., Piot, J.: Cismef: a structured health resource guide. Meth Inf Med 39 (2000) 30-35
  2. Liu, Y., Teverovskiy, L., Carmichael, O., Kikins, R., Shenton, et al.: Discriminative mr image feature analysis for automatic schizophrenia and alzheimer's disease classification. In: Proc. of MICCAI'04. (2004) 393-401
  3. Clough, P., Muëller, H., Deselaers, T., Grubinger, M., Lehmann, T., Jensen, J., Hersh, W.: The clef 2005 cross-language image retrieval track. In: Working Notes of the CLEF Workshop, Vienna, Austria (2005)
  4. Lehmann, T.M., Güld, M.O., Thies, C., Fischer, B., Keysers, M., Kohnen, D., Schubert, H., Wein, B.B.: Content-based image retrieval in medical applications for picture archiving and communication systems. In Proceedings of Medical Imaging. 5033, San Diego, California (2003) 440-451
  5. Güld, M., Keysers, D., Deselaers, T., Leisten, M., Schubert, H., Ney, N., Lehmann, T.: Comparison of global features for categorization of medical images. In: Proceedings SPIE 2004. Volume 5371. (2004)
  6. Marée, R., Geurts, P., Piater, J., Wehenkel, L.: Biomedical image classification with random subwindows and decision trees. In: Proc. ICCV workshop on Computer Vision for Biomedical Image Applications. Volume 3765. (2005) 220-229
  7. Florea, F., Rogozan, A., Bensrhair, A., Darmoni, S.: Medical image retrieval by content and keyword in an on-line health-catalogue context. In: Computer Vision/Computer Graphics Collaboration Techniques and Applications, INRIA Rocquencourt, France (2005) 229-236
  8. Florea, F., Rogozan, A., Bensrhair, A., Darmoni, S.: Comparison of feature-selection and classification techniques for medical image modality categorization. In: accepted at 10th IEEE OPTIM2006, SS Technical and Medical Applications, Brasov, Romania (2006)
  9. Kaufman, L.: Finding groups in data: an introduction to cluster analysis. In: Finding Groups in Data: An Introduction to Cluster Analysis. Wiley, New York (1990)
  10. Stanfill, C., Waltz, D.: Toward memory based reasoning. Communications of the ACM 29 (1986) 1213-1228
  11. Gersho, A., Gray, M.: Vector quantization and signal compression. Kluwer Academic Publishers, Boston (1992)
  12. Florea, F., Rogozan, A., Bensrhair, A., Dacher, J.N., Darmoni, S.: Modality categorisation by textual annotations interpretation in medical imaging. In et al., R.E., ed.: Connecting Medical Informatics and Bio-Informatics. (2005) 1270-1275
Download


Paper Citation


in Harvard Style

Florea F., Barbu E., Rogozan A. and Bensrhair A. (2006). Multi-modal Categorization of Medical Images Using Texture-based Symbolic Representations . In 6th International Workshop on Pattern Recognition in Information Systems - Volume 1: PRIS, (ICEIS 2006) ISBN 978-972-8865-55-9, pages 48-57. DOI: 10.5220/0002500300480057


in Bibtex Style

@conference{pris06,
author={Filip Florea and Eugen Barbu and Alexandrina Rogozan and Abdelaziz Bensrhair},
title={Multi-modal Categorization of Medical Images Using Texture-based Symbolic Representations},
booktitle={6th International Workshop on Pattern Recognition in Information Systems - Volume 1: PRIS, (ICEIS 2006)},
year={2006},
pages={48-57},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0002500300480057},
isbn={978-972-8865-55-9},
}


in EndNote Style

TY - CONF
JO - 6th International Workshop on Pattern Recognition in Information Systems - Volume 1: PRIS, (ICEIS 2006)
TI - Multi-modal Categorization of Medical Images Using Texture-based Symbolic Representations
SN - 978-972-8865-55-9
AU - Florea F.
AU - Barbu E.
AU - Rogozan A.
AU - Bensrhair A.
PY - 2006
SP - 48
EP - 57
DO - 10.5220/0002500300480057