The present invention extracts a graph (G) representing an estimation of the hepatic vein from image data. A plurality of shape models (M1, M2, etc.) are then acquired which represent a plurality of subtree structures that branch off from sections corresponding to regions of origin in a tree structure representing the general shape of the hepatic vein. A prescribed cost function is used to associate the plurality of shape models (M1, M2, etc.) to the graph, and a determination is made as to whether sections of the graph extending from the vertices of graph sections (G1, G2, etc.) associated with the shape models (Mi) are present in the graph (G). (a) If a section of the graph is determined not to be present, the position information of the vertex thereof is acquired as information representing the position of the point of origin, and (b) if a section of the graph is determined to be present, the node that is present in the position closest to an estimated position (P) of the point of origin is defined by tracing, from the vertex of the section of the graph, a node on the section of the graph that extends from the vertex so as to approach the estimated position (P), and the position information of this node is acquired as information representing the position of the point of origin.La présente invention extrait un graphique (G) représentant une estimation de la veine hépatique à partir de données d'image. Plusieurs modèles de forme (M1, M2, etc.) sont ensuite acquis, lesquels représentent une pluralité de structures sous-arborescentes qui s'écartent de sections correspondant à des régions d'origine dans une structure arborescente représentant la forme générale de la veine hépatique. Une fonction de coût prescrite est utilisée pour associer les différents modèles de forme (M1, M2, etc) au graphique, et une détermination est réalisée quant au point de savoir si des sections du graphique s'étendant à partir des sommets de sections de graphique (G1, G2, etc) associées au
