We have compared various image-processing and display methods used in nuclear medicine. The technique of superimposing computer-simulated tumours onto normal clinical data was used in an effort to achieve both clinical relevance and knowledge of “truth”.
A series of 100 clinically normal brain scintigrams were selected from departmental records and 50 “mathematical tumours” simulated and added at clinically relevant positions, no more than one per view. The series was distributed to hospitals active in the field.
The survey involved seven hospitals, 25 observers, and 48 methods of generating the display, including analogue display on X-ray film and Polaroid, and digital display on colour and black-and-white television, electrostatic printer-plotter, and storage display oscilloscope. Experimental techniques used included Latin square ordering of data to study the effect of learning.
A simple rating system was used by each observer to classify the degree of abnormality in each image. Receiver operating characteristic curves (involving lesion localissation) were used to analyse the results. A points system, based on a very simple cost-benefit analysis, was also adopted.
The experiment was then repeated using five observers and 12 display methods for a series of 100 normal liver scintigrams including 50 with superimposed “cold spots”.
The major conclusions were that interpolated data and analogue displays performed well, that enhancement and convexity filters tended to improve observer performance, and that the role of learning should always be considered in surveys of this type.
Unable to display preview. Download preview PDF.
- 1.IAEA (1973) IAEA co-ordinated research programme on the intercomparison of computer-assisted scintigraphic techniques. In: Medical radioisotope scintigraphy Proc. Symp. Monte-Carlo, 1972. IAEA, ViennaGoogle Scholar
- 2.IAEA (1977) IAEA co-ordinated research programme on the intercomparison of computer-assisted scintigraphic techniques. In: Medical radionuclide imaging: Proc. Symp. Los Angeles, 1976. IAEA, ViennaGoogle Scholar
- 4.Houston AS (1977) An intercomparison of three quantitative data-processing techniques using mathematical tumours and receiver operating characteristic analysis. In: Medical radionuclide imaging: Proc. Symp. Los Angeles, 1976. IAEA, ViennaGoogle Scholar
- 8.Houston AS (1978) A comparison and evaluation of image processing and display methods in brain scintigraphy. In: Biosigma 78: Proc. Symp. Paris, 1978Google Scholar
- 9.Brown NJG, Budd T, Britton KE (1975) The Quantitation of difference in scintigraphy using an interactive display system. In: Quality factors in nuclear medicine. Proc. 13th Internationale Jahrestagung der Gesellschaft für Nuklearmedizin, Copenhagen, 1975. Munkner T (ed) FADL, CopenhagenGoogle Scholar
- 10.Houston AS (1975) The step filter and its use in brain scintigraphy. In: Munkner T (ed) Quality factors in nuclear medicine, Proc. 13th Internationale Jahres- tagung der Gesellschaft für Nuklearmedizin, Copenhagen, 1975 FADL, CopenhagenGoogle Scholar
- 11.Corfield JR (1975) Development of a fast image enhancement filter for routine use. In: Munkner T (ed) Quality factors in nuclear medicine. Proc. 13th Internationale Jahrestagung der Gesellschaft für Nuklearmedizin, Copenhagen, 1975. FADL, CopenhagenGoogle Scholar
- 14.Nagai T, Fukuda N, Iinuma TA (1969) Computer-focusing using an appropriate Gaussian function. J Nucl Med 10: 209–212Google Scholar
- 15.Pizer SM (1971) Digital spatial filtering and its variations. In: Quantitative organ visualisation in nuclear medicine. University Press, Miami, pp 581–603Google Scholar
- 16.Lorenz WJ, Georgi P, Meder HG, Pistor P, Walch G, Wiebelt H (1973) Interactive processing and displaying of digital scintiscans. In: Medical radioisotope scintigraphy: Proc. Symp. Monte-Carlo, 1972. IAEA, ViennaGoogle Scholar
- 18.Gustafsson T, Todd-Pokropek AE (to be published) Design and application on filters with variable shape. 2nd European Meeting on Data Handling and Image Processing in Scintigraphy, Hannover, 1971Google Scholar
- 20.Metz CE (1969) A mathematical investigation of radioisotope scan image processing Ph D Thesis, University of PennsylvaniaGoogle Scholar