Image Texture, Uniformity, Homogenity and Radiation Dose Properties in CT

Abstract

In medical imaging, especially when potentially harmful X-ray radiation is used to obtain diagnostic data, it is important to reduce the risk the patient is exposed to. For that reason it is expected to obtain maximum diagnostic information with minimum exposure to harmful factors, by accurate selection of scanning parameters. This thesis analyzes the noise properties of two CT scanners, GE Lightspeed HD 750, and Toshiba Aquillion ONE, based on reported differences in appearance of the images scanned. Catphan Phantom 600 uniformity module is scanned with both scanners, using range of configurations and extension rings simulating patients of different sizes. 60 datasets were obtained and analyzed in terms of the uniformity, homogeneity, noise texture and spectral power distribution. The results show that GE scanner provides images less contaminated with noise when larger patient is scanned or the scanned region is enveloped with big volumes of matter or high density regions, that are sources of artifacts in CT. On the other hand, it is Toshiba that performed better when desired region of scanning was easily accessable (small or slim patients). Because human body is a source of many artifacts that increase the difficulties of obtaining desired data, it is understandable, that medical practitioners have more success in retriving diagnostically important information from images from GE scanner, which has better results in similar simulated conditions.