Authors: Blokh D; Stambler I; Afrimzon E; Platkov M; Shafran Y; Korech E; Sandbank J; Zurgil N; Deutsch M
Abstract: We present a method for the comparative analysis of parameter groups according to their correlation to disease. The theoretical basis of the proposed method is Information Theory and Nonparametric Statistics. Normalized mutual information is used as the measure of correlation between parameters, and statistical conclusions are based on ranking. The fluorescence polarization (FP) parameter is considered as the principal diagnostic characteristic. The FP was measured in fluorescein diacetate (FDA)-stained individual peripheral blood mononuclear cells (PBMC), derived from healthy subjects and breast cancer (BC) patients, under different stimulation conditions: by tumor tissue, the mitogen phytohemagglutinin (PHA) or without the stimulants. The FP parameters were grouped according to their correlation with breast cancer. It was established that the greatest difference between cells of BC patients and healthy subjects is found in the PHA test (parameter P1).
Keywords: Adult; Aged; Aged, 80 and over; Breast Neoplasms/*diagnosis/*pathology; Case-Control Studies; Computational Biology/methods; Computer Simulation; Early Detection of Cancer; Female; Fluoresceins/pharmacology; Humans; Image Processing, Computer-Assisted/*methods; Leukocytes, Mononuclear/cytology/*pathology; Microscopy, Fluorescence/methods; Middle Aged; Pattern Recognition, Automated; Phytohemagglutinins/chemistry; Statistics, Nonparametric|
Journal: Computer methods and programs in biomedicine Volume: 94 Issue: 3 Pages: 239-49 Date: Feb. 24, 2009 PMID: 19231022 |
Blokh D, Stambler I, Afrimzon E, Platkov M, Shafran Y, Korech E, Sandbank J, Zurgil N, Deutsch M (2009) Comparative analysis of cell parameter groups for breast cancer detection. Computer methods and programs in biomedicine 94: 239-49.
Comment on This Data Unit