The removal of Compton scattered photons included within the pulse height window is recognized as one of the most difficult noise problems in the restoration of nuclear medicine images. A new approach to Compton scatter correction based on factor analysis of dynamic structures (FADS) is presented in this study. The method requires all of the energy information. Acquisition of data can be performed either by list-mode or frame-mode. While the former presents some theoretical advantages, the latter is actually used in this work. Two factors are extracted by FADS, unfortunately no pure photopeak factor can be found by the algorithm. These rough factors lead to incorrect factor images. The innovation reported here is the use of a constrained photopeak factor. This novel algorithm is evaluated both on planar imaging and SPECT data using Monte Carlo simulations and real phantoms. A comparison with the modified method of Jaszczak is also presented. Different parameters are significantly improved with our recombination method in SPECT studies, particularly after attenuation compensation by the iterative method of Chang. Compared with the subtraction method the contrast is increased by 1.5 for planar Monte Carlo simulations and the scatter fraction is reduced four times with our recombination method.

We present one case of peritoneal splenosis, which was not confirmed by the splenic scintiscan with 99mTc-sulphur colloid, but whose diagnosis, carried out during a second scintiscan with 99mTc-heat-damaged RBC, was confirmed by laparotomy and histology. This case confirms that, for the diagnosis of splenosis, heat-damaged RBC scintigraphy must be used rather than either sulphur colloid scintigraphy or computed-tomography.

This article develops two contrast-modification techniques for the display of scintigraphic images. Based on histogram-modification techniques, histogram equalization, where each level of gray is used to the same extent, gives maximum entropy. The first technique uses the application of histogram equalization in the whole image. To eliminate contrast attenuation small but important portion of the gray scale histogram, local area histogram equalization has been applied to images with differences in intensity. Both techniques were tested using a phantom with known characteristics. The global equalization technique is more suitable to bone scintigraphies, and some well-chosen boundaries improved the difference between two comparable areas. For liver scintigraphies, where intensity is quite equal in every pixel, a local area equalization was chosen that allowed detection of heterogeneous structures. The images resulting from histogram-equalization techniques improve the readability of data, but are often far from usual images and necessitate an apprenticeship for the physician.

No abstract

Contour detection is an important step in information extraction from nuclear medicine images. In order to perform accurate quantitative studies in single photon emission computed tomography (SPECT) a new procedure is described which can rapidly derive the best fit contour of an attenuated medium. Some authors evaluate the influence of the detected contour on the reconstructed images with various attenuation correction techniques. Most of the methods are strongly affected by inaccurately detected contours. This approach uses the Compton window to redetermine the convex contour: it seems to be simpler and more practical in clinical SPECT studies. The main advantages of this procedure are the high speed of computation, the accuracy of the contour found and the programme's automation. Results obtained using computer simulated and real phantoms or clinical studies demonstrate the reliability of the present algorithm.