医用画像からの運動器系軟部組織の抽出支援技術(4部計測)

概要

論文の詳細を見る
An automatic segmentation method of locomotor organs and soft tissues, bones, muscular tissues and joint disc from medical images has been required in the field of plastic surgery and biomechanics. For this purpose, a region-growing method was used generally. In the general region-growing method, a core grows to add a homogeneous pixel judged from its features (edge, intensity and so on) based on image information. However, the bones and tissues in the medical images of the joint are not always extracted by a region-growing method based on only image information. A geometric model of the tissues should not be used for this purpose, because a number of locomotor organs exist and we cannot prepare models for all the tissues. Thus, we improve the conventional region-growing method by using not only image information but also anatomical features of the tissues. The bones and other tissues are extracted by the present method without geometric models. The shape of the core is determined from the core of the previous image because of spatial continuity of the tissues. Next, the core position is estimated roughly in the limitation area as follows: A template pattern of the core was obtained from the previous image, and the core position in the present image was determined by template-matching. The template-matching was performed in the limitation area that was determined by structural features of the bone and the tissues. As a result, estimated core position corresponds to the best matching position of the template pattern obtained from the previous image. When there is continuity of the tissue shape between the cross-section images, the pixel adjoining the core is judged to either add to the core or not by using following anatomical features: location of the tissue (range from the core), complexity of the shape (perimeter/area) and size (area). These anatomical features are described in fuzzy rules and the judgment was performed by the anatomical features and image information using fuzzy reasoning method. In the ideal case, manual digitization of the tissues was required only at the first of the cross-section images. In reality, the region segmented by the present method was not always suitable. If the segmented region was unsuitable, redigitization was required. Using the present method, intereosseous membrane was segmented from coronal MR images of forearms, and the movements of the articular disc were obtained from a temporal series of sagittal MR images. The segmented shapes by this method agreed with the shape segmented by a medical doctor. Three-dimensional shapes of the intereosseous membrane in supination and pronation were obtained from the segmented region. As a result, it was found that the present method would be a new method of measuring of joint detail motion for biomechanics.
バイオメカニズム学会の論文
1996-07-25

医用画像からの運動器系軟部組織の抽出支援技術(4部計測)

スポンサーリンク

概要

著者

関連論文

スポンサーリンク

医用画像からの運動器系軟部組織の抽出支援技術(4部 計測)

スポンサーリンク

概要

著者

関連論文

スポンサーリンク

医用画像からの運動器系軟部組織の抽出支援技術(4部計測)