画像処理エキスパートシステムによるマイクロ・キャリア上動物細胞の視覚認識・計測法 : 第3報, ニューラルネットワークを用いた不定形重複細胞群の認識

This paper deals with an expert system for the recognition and counting of an animal cell for "interferon β type" by image processing and the neural network. These cells are cultured on microcarrier (micro bead) which is 150-200 micrometers in diameter and is floating in culture solution. The image processing technique provides a method of measuring the location of microcarriers and the shape of a cell. The neural network based on the expert system recognizes the overlapping cells and dust from image data. This method is an improved one as compared with the former one in the point that the former one can't recognize overlapping cells. This system provides correct answers at the rate of 83 percent. According to the experimental results, this system will be able to substitute human experts in the counting of cells on the micro bead.