An information-theoretical model for breast cancer detection.

Authors: Blokh D; Zurgil N; Stambler I; Afrimzon E; Shafran Y; Korech E; Sandbank J; Deutsch M

Abstract: OBJECTIVES: Formal diagnostic modeling is an important line of modern biological and medical research. The construction of a formal diagnostic model consists of two stages: first, the estimation of correlation between model parameters and the disease under consideration; and second, the construction of a diagnostic decision rule using these correlation estimates. A serious drawback of current diagnostic models is the absence of a unified mathematical methodological approach to implementing these two stages. The absence of a unified approach makes the theoretical/biomedical substantiation of diagnostic rules difficult and reduces the efficacy of actual diagnostic model application. METHODS: The present study constructs a formal model for breast cancer detection. The diagnostic model is based on information theory. Normalized mutual information is chosen as the measure of relevance between parameters and the patterns studied. The "nearest neighbor" rule is utilized for diagnosis, while the distance between elements is the weighted Hamming distance. The model concomitantly employs cellular fluorescence polarization as the quantitative input parameter and cell receptor expression as qualitative parameters. RESULTS: Twenty-four healthy individuals and 34 patients (not including the subjects analyzed for the model construction) were tested by the model. Twenty-three healthy subjects and 34 patients were correctly diagnosed. CONCLUSIONS: The proposed diagnostic model is an open one, i.e. it can accommodate new additional parameters, which may increase its effectiveness.

Keywords: Breast Neoplasms/*diagnosis; Female; Fluorescence Polarization; Humans; *Information Theory; *Models, Statistical
Journal: Methods of information in medicine
Volume: 47
Issue: 4
Pages: 322-7
Date: Aug. 12, 2008
PMID: 18690365
Select reference article to upload


Categories: Information Theory
Citation:

Blokh D, Zurgil N, Stambler I, Afrimzon E, Shafran Y, Korech E, Sandbank J, Deutsch M (2008) An information-theoretical model for breast cancer detection. Methods of information in medicine 47: 322-7.



Update (Admin) | Auto-Update

Comment on This Data Unit