Abstract
This paper analyses the performance of four different feature-selection approaches of the Karhunen-Loève expansion (KLE) method to select the most discriminant set of features for computer-assisted classification of bioprosthetic heart-valve status. First, an evaluation test reducing the number of initial features while maintaining the performance of the original classifier is developed. Secondly, the effectiveness of the classification in a simulated practical situation where a new sample has to be classified is estimated with a validation test. Results from both tests applied to a reference database show that the most efficient feature selection and classification (≥97% of correct classifications (CCs)) are performed by the Kittler and Young approach. For the clinical databases, this approach provides poor classification results for simulated ‘new samples’ (between 50 and 69% of CCs). For both the evaluation and the validation tests, only the Heydorn and Tou approach provides classification results comparable with those of the original classifier (a difference always≤7%). However, the degree of feature reduction is particularly variable. The study demonstrates that the KLE feature-selection approaches are highly population-dependent. It also shows that the validation method proposed is advantageous in clinical applications where the data collection is difficult to perform.
Similar content being viewed by others
References
Andrews, D. F., andHerzberg, A. M. (1985): ‘Data: a collection of problems from many fields for the student and research worker’, (Springer Series in Statistics), pp. 5–9
Bailey, T., andJain, A. K. (1978): ‘A note on distance-weightedK nearest neighbor rules’,IEEE Trans. Syst. Man. Cybern.,SMC-8, pp. 311–313
Chien, Y. T., andFu, K. S. (1967): ‘On the generalized Karhunen-Loève expansion’,IEEE Trans. Inf. Theory,IT-13, pp. 518–520.
Ciaccio, E. J., Dunn, S. M., andAkay, M. (1993): ‘Biosignal pattern recognition and interpretation systems: Part I’,IEEE Eng. Med. Biol., pp. 89–95
Devijver, P. A., andKittler, J. (1982): ‘Pattern recognition: a statistical approach’ (Prentice Hall, Englewood Cliffs, New Jersey), chap. 9, pp. 301–341
Durand, L. G., Blanchard, M., Sabbah, H. N., Cloutier, G., Kemp, S. R., andStein, P. D. (1990): ‘Comparison of pattern recognition methods for computer-assisted classification of spectra of heart sounds in patients with a porcine bioprosthetic valve implanted in the mitral position’,IEEE Trans.,BME-37, pp. 1121–1129
Durand, L. G., De Guise, J., Cloutier G., Guardo, R., andBrais, M. (1986): ‘Evaluation of FFT-based and Modern parametric methods for the spectral analysis of bioprosthetic valve sounds’,IEEE Trans.,BME-33, pp. 572–578
Durand, L. G., Guo, Z., Sabbah, H. N., andStein, P. D. (1993): ‘Comparison of spectral techniques for computer assisted classification of spectra of heart sounds in patients with a porcine bioprosthetic valve’,Med. Biol. Eng. Comput.,31, pp. 229–236
Foale, R. A., Joo, T. H., McClellan, J. H., Metzinger, R. W., Grant, G. L., Meyers, G. S., andLees, R. S. (1983): ‘Detection of aortic porcine valve dysfunction by maximum entropy spectral analysis’,Circulation,68, pp. 42–49
Fukunaga, K. (1990): ‘Introduction to statistical pattern recognition’ (Academic Press, New York)
Geva, A. B., andPratt, H. (1994): ‘Unsupervised clustering of evoked potentials by waveform’,Med. Biol. Eng. Comput.,32, pp. 543–550
Guo, Z., Durand, L.-G., Lee, H. C., Allard, L., Grenier, M.-C., andStein, P. D. (1994): ‘Artificial neural networks in computer assisted classification of heart sounds in patients with porcine bioprosthetic valves’,Med. Biol. Eng. Comput.32, (3), pp. 311–316
Jain, A. K. (1987): ‘Advances in statistical pattern recognition’,NATO ASI Series, vol. F-30, Pattern Recognition Theory and applications. pp. 1–19
Joo, T. H., McClellan, J. H., Foale, R. A., Meyers, G. S., andLees, R. S. (1983): ‘Pole-zero modeling and classification of phonocardiograms’,IEEE Trans.,BME-30, pp. 110–118
Kittler, J. (1977): ‘Feature selection methods based on the KarhunenLoève expansion’,inFu andWhinston (Eds.): ‘Pattern recognition theory and application’, (Noordhoff, Leyden) pp. 61–74
Kittler, J. (1986): ‘Handbook of pattern recognition and image processing’, (Academic Press), chap. 3, pp. 59–83
Kittler, J., andYoung, P. C. (1973): ‘A new approach to feature selection based on the Karhunen-Loève expansion’,Pattern Recognition,5, pp. 335–352
McNitt-Gray, M. F., Huang, H. K., andSayre, J. W. (1995): ‘Feature selection and pattern classification problem of digital chest radiograph segmentation’,IEEE Trans. Med. Imaging,14, pp. 537–547
Potocki, J. K., andTharp, H. S. (1992): ‘Reduced-order modelling for hyperthermia control’,IEEE Trans.,BME-39, pp. 1265–1273
Sheikholeslami, N., andStashuk, D. (1994) ‘Information based feature selection for supervised motor unit action potential classification’. 16th Annual Int. Conf. of IEEE Eng. in Med. Biology, pp. 1350–1351
Shiavi, R. (1990): ‘Quantitative representation of electromyographic patterns generated during human locomotion,IEEE Eng. Med. Biol.,9, pp. 58–60
Stein, P. D., Kemp, S. R., Riddle, J. M., Lee, M. W., Lewis, J. W. Jr.,Magilligan, D. J. Jr., (1985): ‘Relation of calcification to torn leaflets of spontaneously degenerated procine bioprosthetic valves’,Ann. Thorac. Surg.,40, pp. 175–180
Stein, P. D., Sabbah, H. N., Lakier, J. B., Kemp, S. R., andMagilligan, D. J., Jr., (1984): ‘Frequency spectra of the first heart sound of the aortic component of the second heart sound in patients with degenerated porcine bioprosthetic valves’,Am. J. Cardiol.,53, pp. 557–561
Stein, P. D., Sabbah, H. N., Lakier, J. B., Magilligan, D. J. Jr., andGoldstein, S. (1981): ‘Frequency of the first heart sound in the assessment of stiffening of mitral bioprosthetic valves’,Circulation,63, pp. 200–204
Thiene, G., Arbustini, E., Bortolotti, U., Talenti, E., Milano, A., Valente, M., Molin, G., andGallucci, V. (1982): ‘Pathologic substrates of porcine valve dysfunction’,in ‘Cardiac Bioprostheses’, (Yorke Med. Books, New York) chap. 34, pp. 378–400
Tou, J. T., andGonzalez, R. C. (1974): ‘Pattern recognition principles’ (Addison-Wesley Publishing Company, Advanced Book Program, Reading, Massachusetts) chap. 7, pp. 269–283
Tou, J. T., andHeydorn, R. P. (1967): ‘Some approaches to optimum feature extraction’,in ‘Computer and information sciences II’ (Academic Press, New York) pp. 61–89
Watanabe, S. (1965): ‘Karhunen-Loève expansion and factor analysis theoretical remarks and applications’. Proc. 4th. Prague Conf. Information Theory
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Yazdanpanah, M., Allard, L., Durand, LG. et al. Evaluation of Karhunen-Loeve expansion for feature selection in computer-assisted classification of bioprosthetic heart-valve status. Med. Biol. Eng. Comput. 37, 504–510 (1999). https://doi.org/10.1007/BF02513337
Received:
Accepted:
Issue Date:
DOI: https://doi.org/10.1007/BF02513337