Research Of Compton Scattering Imaging Based On Energy Spectral Data

Compton scattering imaging is a radiation imaging technique that forms the image of an object by measuring Compton scattering emitted from the object while it is irradiated. It presents images of electron density distribution inside the object. It has particular advantage in such cases when the traditional transmitted radiation imaging modality fails. Therefore this technique is very promising in the field of medical diagnosis and industrial non-destructive testing.This thesis first stduies electron density imaging based on Compton scattering energy spectral data. First, the thesis discusses Compton scattering energy spectral data simulation using Monte Carlo Method. The comparision of Monte Carlo simulation result and Macro Statistic simulation result shows the superiority of the former.Then, an improved point-by-point imaging method based on the line-scanning energy spectral data is proposed. In this modality, line-scan is adopted instead of point-by-point-scan, and the count of scattered photons is obtained by parseing line-scanning energy spectral data. Simplifying experimental setting, shorting scan time, making the most of radiation dose and avoiding focus error, this method has significant value to real-time medical diagnosis. Meanwhile, we put forward symmetric correction to resolve inherent error cumulation of point-by-point imaging modality. Some numerical results presented in this paper indicate the effect of symmetric correction is close to transmitted correction when imaging for object surface and non-thick and non-dense objects, however, the former supports back scattering imaging which is not supported by the latter.Finally, an improved successive approximation process algorithm based on histogram constraint and Tikhonov-Miller regularization is proposed. Due to the fact that the problem of reconstructing image from Compton scattering energy spectral data is an ill-posed problem, any error or noise in the measurements will be amplified in the reconstruction results. To obtain a stable solution, in our method, successive approximation process is constrained by prior histogram and Laplacian regularization. Comparing with traditional successive approximation process algorithm, our method can improve quality of reconstruction result while speeding up convergence. Further, proper threshold is adopted to enhance the robustness of the histogram constraint.,...