Anais do X SIBGRAPI (1997) #AND16
Abstract. We propose a method to study myocardial fiber nuclei morphology through Laser Scanning Confocal Microscopy and multivariate statistical analysis. Image datasets are preprocessed by the means of a blind deconvolution algorithm and the nuclei are segmented by a simple mathematical morphology technique. Features are extracted in order to describe three-dimensional nuclear morphology. A multivariate statistical analysis is performed to discriminate different pathological nuclei and to clarify the idiopathic cardiomyopathy diseases.
[ full version | contact authors | project home page ]