[TAKAHASHI, Haruaki ▲高▼橋　玄顕, DOKE, Takehito 道家　健仁, TAKAHASHI, Ryohei ▲高▼橋　遼平, KIN, Taichi 金　太一, SAITO, Toki 斎藤　季, KUNII, Naoto 國井　尚人, OYAMA, Hiroshi 小山　博史, SAITO, Nobuhito 齊藤　延人] 4F, 4-1-3, Hongo, Bunkyo-ku, Tokyo 〒1500002 東京都文京区本郷四丁目１番３号４階 Tokyo1500002;3-1, Hongo 7-chome, Bunkyo-ku, Tokyo 〒1138654 東京都文京区本郷七丁目３番１号 Tokyo1138654 An acquisition unit of an image processing device acquires intended data that is three-dimensional medical image data from which an area of interest is to be extracted. A conversion unit: performs a conversion process so as to ensure that at least one of the size, direction, and position of the intended data coincides with standard data, which is three-dimensional medical data representing a standard, so that the degree of coincidence between the intended data and the standard data increases; and generates pre-converted data, which is three-dimensional medical data that has been subjected to the conversion process on the intended data. An area extraction unit inputs the pre-converted data to a pre-learned model that has been trained in advance in order to extract an area of interest included in three-dimensional medical image data from the three-dimensional medical image data, and extracts the area of interest included in the pre-converted data through computation by the pre-learned model. An acquisition unit of an image processing device acquires intended data that is three-dimensional medical image data from which an area of interest is to be extracted. A conversion unit: performs a conversion process so as to ensure that at least one of the size, direction, and position of the intended data coincides with standard data, which is three-dimensional medical data representing a standard, so that the degree of coincidence between the intended data and the standard data increases; and generates pre-converted data, which is three-dimensional medical data that has been subjected to the conversion process on the intended data. An area extraction unit inputs the pre-converted data to a pre-learned model that has been trained in advance in order to extract an area of interest included in three-dimensional medical image data from the three-dimensional medical image data, and extracts the area of interest included in the pre-converted data through computation by the pre-learned model.