Digital Image Processing Applied to Scintillation Images from Biomedical Systems

Several nonlinear filtering methods have been developed for two-dimensional image signals obtained from an Anger Scintillation camera. The first method termed "pruning" is implemented in the frequency domain and filters out certain components of the two-dimensional amplitude spectrum. The other methods called "E-filtering" and "P-filtering" utilize a nonlinear transformation of the spatial coordinates which is proportional to the two-dimensional signal amplitude and energy spectrum, respectively. These nonlinear filtering techniques have been applied to phantom scintigrams as well as brain and heart scans of normal and abnormal patients. In addition, the detection of boundaries using the hexagonal Golay transform on filtered and unfiltered images is included. Finally, these filtering and edge detection techniques were for the first time applied to a new technique for the detection of acute myocardial infarcts.