X-ray diffraction measurements for breast tissue analysis
Biological tissues exhibit elastic x-ray scatter properties unique to individual tissue types. It has been suggested that these properties could be exploited in the discrimination of tissues in diagnostic radiology, particularly in mammographic imaging. The work in this thesis investigates the elastic scatter from human breast tissues with a view to moving the Diffraction Enhanced Breast Imaging (DEBI) technique towards the clinical domain. Instrumentation was considered in terms of a suitable scatter detector and its associated collimation. A phosphor coated low-light-level CCD in conjunction with a parallel hole array collimator was introduced as a possible angular dispersive diffraction detector, and conically collimated HPGe was found to be a suitable detector for energy dispersive diffraction. Diffraction imaging was achieved using both planar and tomographic imaging techniques. The planar imaging considered the spatial resolution needed for clinically useful results and showed that depth information of the order of 1 cm can easily be obtained using diffraction imaging techniques. A micro-tomographic diffraction imaging technique was used to further present knowledge of the 'scatter signatures' of human breast tissue. The tissue is naturally inhomogeneous, so precise knowledge of a sample's composition is needed in order to accurately classify its scatter signature. A total of twenty seven breast tissue samples were examined, covering a range of tissue types, with nineteen of these samples being extensively studied. The microCT produced diffraction images of the inhomogeneous tissue samples and their diffraction signatures were related to the results of conventional histopathological analysis. A correlation was observed between the predominant tissue type in a sample and the ratio of the intensity of the sample's scatter at momentum transfers of 1.1 and 1.5 nm-1. A Student's t-test performed on these results gave the probability of a null hypothesis for correlation of less than 1%.