Fundamental study on preliminary image processing at time development of CNN using chest radiography

Cancers were discovered in bronchi and lungs, chest radiography is getting important. Some lung diseases affect only specific individuals such as lung cancer, but some diseases are related to infectious diseases such as pulmonary tuberculosis, threatening human health. Chest radiography is most convenient method to protect human health from such threats, such as a censer. In this study, we investigated whether preliminary image processing at the time development of a Convolution Neural Network (CNN) for judging presence or absence of a nodule by chest radiography contributes to improvement in accuracy.