METHOD OF CLASSIFYING CANCER CELL, DEVICE FOR CLASSIFYING CANCER CELL, AND PROGRAM FOR CLASSIFYING CANCER CELL

摘要

PROBLEM TO BE SOLVED: To perform further detailed classification based on the feature of the cancer cells even when the cancer cells have the same shape by acquiring further appropriate information based on the feature of the cancer cells.

SOLUTION: Data of protein amount and agglomeration degree of protein is extracted as required data from the data about the cancer cells acquired by a laser scan cytometer, a dispersion map is prepared, the dimension of the data showing the dispersion map is reduced, and input data is prepared. The dispersion map and the input data are input to an input layer of assembly of neurons and an input layer of a network having a map layer, respectively. Learning about a joint load between the input layer and the map layer is performed, topological mapping is performed, and a self-organizing map is formed. The feature of the cancer cells is discriminated based on the distribution form of protein in the formed self-organizing map, and the cancer cells can be classified.

COPYRIGHT: (C)2009,JPO&INPIT