The objective of the present invention is to improve estimation accuracy of a magnetic susceptibility value of tissue by computing a contour image which represents the contour of the tissue on a magnetic susceptibility distribution and to reduce background noise without lowering the magnetic susceptibility value of the tissue. The present invention computes an absolute value image and a phase image from a complex image obtained by MRI, from the phase image, computes a low frequency region magnetic susceptibility image in which background noise is greater than a desired value but which has information on magnetic susceptibility value of a tissue, computes a contour information magnetic susceptibility image in which the magnetic susceptibility value is not correct but which has contour information of the tissue on a magnetic susceptibility distribution and computes a high frequency region magnetic susceptibility image in which the magnetic susceptibility value of the tissue is not correct but which is computed by a regularization method using a regularization parameter in which a contrast-to-noise ratio is at a maximum, computes a contour mask from the contour information magnetic susceptibility image, smooths a magic angle region from the contour mask and the low frequency region magnetic susceptibility image and finally smooths a high frequency region using the high frequency region magnetic susceptibility image.Lobjet de la présente invention est daméliorer la précision de lestimation dune valeur de susceptibilité magnétique de tissus en calculant une image de contour qui représente le contour des tissus sur une distribution de susceptibilité magnétique et de réduire le bruit de fond sans diminuer la valeur de susceptibilité magnétique des tissus. La présente invention calcule une image de valeur absolue et une image de phase à partir dune image complexe obtenue par IRM, à partir de limage de phase, calcule une image de susceptibilité magnétique de région basse fré
