公开(公告)号：US10667778B2

公开(公告)日：2020-06-02

申请号：US15704719

申请日：2017-09-14

申请人： University of Louisville Research Foundation, Inc.

发明人： Ayman S. El-Baz ,  Ahmed Soliman ,  Fahmi Khalifa ,  Ahmed Shaffie ,  Neal Dunlap ,  Brian Wang

IPC分类号： A61B6/00 ,  G06T7/33 ,  G06T7/149 ,  G06T7/143 ,  G06T7/174 ,  G06T7/38 ,  G06T7/246 ,  A61B6/03 ,  G06T7/00 ,  A61N5/10 ,  A61B5/08 ,  A61B5/00

摘要： A system and computation method is disclosed that identifies radiation-induced lung injury after radiation therapy using 4D computed tomography (CT) scans. After deformable image registration, the method segments lung fields, extracts functional and textural features, and classifies lung tissues. The deformable registration locally aligns consecutive phases of the respiratory cycle using gradient descent minimization of the conventional dissimilarity metric. Then an adaptive shape prior, a first-order intensity model, and a second-order lung tissues homogeneity descriptor are integrated to segment the lung fields. In addition to common lung functionality features, such as ventilation and elasticity, specific regional textural features are estimated by modeling the segmented images as samples of a novel 7th-order contrast-offset-invariant Markov-Gibbs random field (MGRF). Finally, a tissue classifier is applied to distinguish between the injured and normal lung tissues.

公开(公告)号：US20180070905A1

公开(公告)日：2018-03-15

申请号：US15704719

申请日：2017-09-14

申请人： University of Louisville Research Foundation, Inc.

发明人： Ayman S. El-Baz ,  Ahmed Soliman ,  Fahmi Khalifa ,  Ahmed Shaffie ,  Neal Dunlap ,  Brian Wang

IPC分类号： A61B6/00 ,  A61B6/03 ,  A61N5/10 ,  G06T7/33 ,  G06T7/149 ,  G06T7/143 ,  G06T7/174 ,  G06T7/38 ,  G06T7/246

摘要： A system and computation method is disclosed that identifies radiation-induced lung injury after radiation therapy using 4D computed tomography (CT) scans. After deformable image registration, the method segments lung fields, extracts functional and textural features, and classifies lung tissues. The deformable registration locally aligns consecutive phases of the respiratory cycle using gradient descent minimization of the conventional dissimilarity metric. Then an adaptive shape prior, a first-order intensity model, and a second-order lung tissues homogeneity descriptor are integrated to segment the lung fields. In addition to common lung functionality features, such as ventilation and elasticity, specific regional textural features are estimated by modeling the segmented images as samples of a novel 7th-order contrast-offset-invariant Markov-Gibbs random field (MGRF). Finally, a tissue classifier is applied to distinguish between the injured and normal lung tissues.