Quantitative assessment of scanning-system parameters.

Abstract

There is a recognized need to investigate radio nuclide scanning on a system basis beginning with the radiopharmaceutical and the clinical problem and encompassing the instrumentation and observer performance. To be effective such investigations must be based upon quantitative measurement of sys tem performance over a range of system-design pa rameters and over a range of medical problems. Although many investigators have measured the performance of discrete components, there has been no systematic investigation of the entire scanning system. This has been due largely to the difficulties associated with varying realistic phantom organs and real detector systems and the time consumed in repeating tests to compensate for statistical flue tuations. Similarly, the limited availability of appro priate data-storage and data-processing equipment has prevented effective exploration of the effects of data processing on performance. Techniques and instrumentation have now been developed to overcome these limitations based on a combination of scanning hardware components and digital simulation to form a system for quantitatively investigating all parameters of the scanning tech nique. The system provides the capability for study ing the effects on image interpretation due to physical characteristics of the radionuclide, charac tenistics of the clinical problem, characteristics of the detector and effects of various data-processing operations. In this system the input data are pro vided by either a digital simulation of a scanning system viewing an organ or alternatively by a scan ner at the Johns Hopkins Medical Institutions scan fling a phantom or a patient. The digital simulation of a particular organ and detector provides the detector-output data in a form representing a normalized scan without statistical fluctuation noise. Stored data from these scans are then operated on by a general-purpose computer which is programmed to introduce the parameters associated with activity levels, lesion size and loca tion, detector efficiency and scanning time, among others. After being transformed to include quantum fluctuation noise, these data represent a complete simulation of a scan. These simulation data are processed in an on-line digital computer and pre sented on a digitally-controlled, TV-like display. The effect of the combined parameters of the system are studied by measuring the performance of ob servers in detecting and characterizing lesions. The value of computed quantities such as area counts in assisting the visual evaluation of a scan can be assessed. Many different parameterscan be rapidly varied, and large numbers of tests can be run so that quantitative results may be obtained consistent with the statistical nature of the problem. The system has been used to study the effects of the density of counts, or counts per unit area, on the detectability of lesions of graded sizes in the scan image of a kidney. This density of counts in the scan image is a direct function of the activity of the radiopharmaceutical or equivalently of the total time devoted to the scan. Hence, the resultant curves of the probability of lesion detection as a function of peak counting rate provide for the first time a quantitative value on the quantity of radioactivity