Development of on-line learning algorithm of organ recognition for on-line CAD

我们研究了一种用于医学图像自动器官识别和疾病检测的在线学习算法。我们的主要目标是支气管和血管。研究了它们的特征描述和组合，使识别精度更高。在我们的实验中，识别精度约为。支气管70%，血管80%。此外，支气管和血管的解剖结构的识别可以实现大约。90%的准确率。

We studied about a on-line learning algorithm for automatic organ recognition and disease detection from medical images. Our main targets were the bronchus and blood vessels. We research their feature description and their combinations to make recognition accuracy higher. In our experiments, the recognition accuracies were approx. 70% for the bronchus and 80% for the blood vessels. Also, recognition of anatomical structure of the bronchus and blood vessels could achieve approx. 90% accuracy.