人工免疫システムを用いたX線胸部単純画像の自動診断

概要

論文の詳細を見る
In this paper, we have proposed artificial immune system for the purpose of aiding automatic diagnosis of lung cancer focus candidate density by Chest Roentgenology. Previously many methods such as discriminant analysis, rule-based judge system, and artificial neural networks have been available, but they have now serious problems as Saturation of acknowledgement, necessity of many calculations. Then artificial immune system has been considered based on human immune network that could recognize external object using internal memory, and that if objects are not yet memorized, then They should be added into internal memory. This function could be used for pattern recognition, optimization, and adaptation mechanism in engineering field. So using eight kinds of image characteristic parameters such as average CT-value, ration of the maximum CT/average CT, and statistical features from Region of interest(ROI) of a chest Radiographs, our artificial immune system could judge automatically the cancer candidate. In our experiment of 31 images, which first checked clearly by doctors, immune system has been confirmed that it could have nearly same ability as previous method.
前橋工科大学の論文
2003-03-27