RESUMO
Scintigraphic images provide morphologic and functional information referring to pointed human body tissues and organs. Depending on the ability and experience of the nuclear medicine physician and the degree of scintigram complexity, the diagnosis process becomes very difficult sometimes. Image processing might decrease subjectivity and help physicians in decision making in complex cases as bone metastasis diagnosis based on scintigraphic explorations. The aim of this paper is to describe the compulsory steps of a pre-processing method in order to build a database for an automatic final appreciation of pathologic bone scan areas as a percentage of the total bone scintigraphic surface. This may include the scintigraphic result in some metastasis probability category with more accuracy than a simple, subjective appreciation of the scintigram, especially in doubtful cases. This paper points to the steps of the processing method of the database used in the rule-based nuclear medicine aide-decision expert system (NMADES). The objective evaluation of the pathological sites requires image preprocessing operations in a number of steps: histogram transforms, correlated superposition of direct and reversed incidences to reinforce the uptake sites, smoothing by pseudo-cepstrum methods, symmetry axes extraction by robust linear regression and symmetric areas search with fuzzy methods. Some for and against's are underlined in the last section, devoted to conclusions and future work.
