Digital image processing and its application to compton gamma ray camera

The use of gamma ray imaging devices in Nuclear Medicine has been well i established` in the past two decades. The development of the devices in the past has illustrated that their performance cannot be improved beyond a certain limit without sacrificing efficiency considerably. Recently, a new device, Compton gamma ray camera, has been proposed which would overcome the basic limitation of the existing devices by removing the collimator used for focussing the gamma rays. Preliminary investigations have shown that the Compton camera would have high resolution and efficiency.In this thesis a detailed investigation of the Compton camera images is presented and the application of digital image processing techniques to the camera images is considered.'A priori` image processing techniques have been used to define contributions in the image plane and estimation techniques are used to form images. A detailed comparison of the image formation methods based on the results obtained by a Monte Carlo simulation of -the camera is presented. A mathematical model of the image formation process is derived for the camera and shown: to be similar to that -for -many linear imaging devices. This fact is used' in developing 'a posteriori' image processing techniques which can be applied to any images formed by linear imagingdevices_ The carers images' are used as an example to illustrate theImage restoration filters are derived by using a novel concept of roll-off filters. It is shown that it is only necessary to design a low-pass filter which satisfies certain resolution and noise requirements. Quantitative resolution measures like the second moment of the filter response and integral error.. are used to derive roll-off filters. Various exact and approximate solutions satisfying the constraints are derived and their performance compared with the minimum mean squared filters and other commonly used filters. It is shown that the new filters allow greater flexibility in controlling the resolution and the output noise. The problem of implementing and realizing the filters non-recursively using two-dimensional data windows is considered. New data windows are proposed and their properties compared with those of the commonly used windows. Results of applying the filters before and after, windowing are evaluated in detail.performance of restoration filters.