Computer-aided diagnosis in chest radiography. Study of the effect caused by varied image quality on texture analysis

Computer-aided diagnosis (CAD) system has been developing in the field of medical imaging. As CAD becomes more common in diagnostic radiology, it is expected that diagnostic performance will be improved. In actual use, however, the results of CAD might be affected by the quality of images, which are obtained under different exposure conditions and/or with differing devices. In this study, we investigated the effect of varied image quality on the results of texture analysis, a method developed by Katsuragawa, et al. and used in CAD for the detection of interstitial lung disease. To provide samples of varying image quality, we obtained images using various exposure doses and scatterer thicknesses. It was found that texture analysis was affected by differences in scatterer thickness. (author)