In the development of devices to represent the three-dimensional structure of radio-activated organs, Nuclear Medicine has been following the progress in Radiology. That parallelism could be observed with all principles used to get three-dimensional data :
- analogical systems working by simultaneous displacement of the detector and the object,
- coded-aperture imaging devices which consist of special collimators designed to obtain a dependance between the object-to-code distance and the detector response,
- multiple-incidence techniques, the 3D reconstruction being extracted from the whole set of projections of the object at different orientations.
That last principle was chosen in the studies that we are working on now. It is close to the principle used in Radio-Tomo-densitometry : a detector gets a set of projections as it turns around the object. From these projections, the classical reconstruction algorithms can be used : ART, SIRT, Convolution algorithms... But we have to take into account here some more specific properties : the statistic noise, the self attenuation of the radiation, the distance-dependant resolution. In this paper some correction process will be considered which can be more or less easily implemented depending of the algorithm used. Different compromises can be proposed : they depend strongly not only of the algorithm and the data-processing but also of the detector performances. In this field, this method will greatly take profit of the powerful calculators designed for the Radiographic tomodensitometry.