| 摘要 |
The invention refers to a method for computer-aided analysis of medical images, where a pair of medical images is analysed, the pair including a first image (I1) which is a contrasted scan of a part in a human or animal body and a second image (12) which is a native scan of the same part of the human or animal body. In the method of the invention, anatomic structures (AS1, AS2) are identified within both the first image (I1) and the second image (I2). By using those anatomic structures (AS1, AS2), centerlines of vessels in the first image (I1) are mapped to the second image (I2). Candidate calcified plaques (CCP) are extracted in the second image (12) and calcified plaques (CCP) out of the candidate calcified plaques (CCP) are identified by a machine learning classifier. The positional information of the centerlines (CL2) in the second image (12) is used for extracting the candidate calcified plaques (CCP) in the second image (12) and/or for identifying the calcified plaques (CCP) out of the candidate calcified plaques (CCP) by the machine learning classifier. |
