PROBLEM TO BE SOLVED: To provide a technique for extracting an image of a target cross section at high speed with a small amount of calculation from two-dimensional or three-dimensional image data acquired by a medical imaging device. A cross-sectional image of a target cross-section including a predetermined structure is extracted from image data of a subject. A learning model already learned by learning data including a learning target cross-sectional image including a structure image, and a learning region-of-interest cross-sectional image obtained by cutting out and enlarging a partial region of the learning target cross-sectional image that includes the structure. Is applied to a plurality of cross-sectional images obtained from image data to detect a region of a predetermined structure included in the cross-sectional image, and a cross-sectional image of a target cross-section based on the detected region of the predetermined structure. To extract. [Selection diagram]【課題】医用撮像装置で取得した２次元又は３次元画像データから、目的断面の画像を少ない演算量で高速に抽出する技術を提供する。【解決手段】被検体の画像データから、所定の構造物を含む目的断面の断面画像を抽出する。構造物の像を含む学習用目的断面画像と、学習用目的断面画像のうち構造物が含まれる一部領域を切り出して拡大した学習用関心領域断面画像とを含む学習データによって学習済みの学習モデルを、画像データから得られる複数の断面画像に適用することにより、断面画像に含まれる所定の構造物の領域を検出し、検出された前記所定の構造物の領域に基づいて目的断面の断面画像を抽出する。【選択図】 図６
