Application of a digital pattern recognition system to

An image discrimination technique was developed to improve specific identification of some toxic Dinophysis cells (marine dinoflagellates involved in diarrhoeic shellfish poisoning), especially the "acuminata" and "sacculus" groups, which can be present at different ratios in natural sea-water samples collected during toxic episodes. This work was performed with image analysis software SAMBA (TITN Alcatel) using preserved cells directly observed through an inverted microscope and recorded with a video camera before further processing. All morphometric parameters used for discrimination of 7 different species or morphotypes of Dinophysis were tested with discriminant analysis. This study indicates that Dinophysis sp. and D. pavillardi seem well classified at the species level, whereas D. cf. acuminata and D. sacculus appear to be morphotypes of D. acuminata.