A Comparison of Data Averaging of Radioisotope Scan Data by Photographic and Dimensional Computer Techniques

Most data recording systems used in radioisotope scanning display each count or group of counts with a mark of very small spatial dimensions placed on the record in direct relationship to the centre of the collimator. Although few collimators used in scanning show an appreciable response to spatial frequencies greater than one cycle per centimetre, the spatial dimensions of the display dot is frequently no larger than one millimetre. For this reason a conventional scan display will exhibit an appearance of high frequency noise which greatly detracts from the ability of the reviewer to recognize small differences in the more gross or low frequency detail. Several techniques have been devised to reduce this noise and improve the visualization of the detail of structures consistent with dimensions resolvable by the collimator. One method has been to combine the counts recorded in adjacent areas to obtain an average count for the entire area segment. Various investigators have used elements ranging from 25 mm to 1 cm, averaged in matrices containing five or nine components. When averaged by computer the program will also adjust values in the matrix that are statistically improbable. A second system has been one of employing a large photographic light dot pattern with dimensions approximating the space and intensity distribution of the point spread function of the collimator. If the increments of x and y axis traversal are small in proportion to the diameter of the light pattern, the intensity at any one spot can be derived from the integration of counts received when the collimator is directed at regions around the centre point as well as when directed to the centre point itself. It is the purpose of the present investigation to evaluate the similarities and differences of these methods by constructing three-dimensional models of intensity versus spatial dimensions using photographic discrete point data recording, discrete point recording followed by matrix averaging, and by large photographic spot averaging. (author)