Automatic tumor identification and segmentation using medical images

• Main Authors: BENGTSSON NIELS GUSTAV TORD, CHAMPION DE CRESPINI ANA J. S, FREDERIKSEN JEFFREY O, CARANO RICARDO ALBERTO DOMENICO, GEMMA MARK S
• Format: Patent
• Language: Chinese; English
• Published: 29.07.2022
• Subjects: CALCULATING; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

A medical image is input into a detection network to generate a mask identifying a set of regions within the medical image, wherein the detection network predicts that each region identified in the mask includes a depiction of a tumor of one or more tumors within a subject. For each region, the regions of the medical image are processed using a tumor segmentation network to generate one or more tumor segmentation boundaries for the tumor present within the subject. For each tumor and by using a plurality of organ-specific segmentation networks, an organ within which at least a portion of the tumor is located is determined. An output is generated based on the one or more tumor segmentation boundaries and a location of the organ within which at least a portion of the one or more tumors are located. 将医学图像输入到检测网络中以生成识别医学图像内的一组区域的掩膜，其中所述检测网络预测在所述掩膜中识别的每个区域包括对受试者内的一个或多个肿瘤中的肿瘤的描绘。针对每个区域，使用肿瘤分割网络处理所述医学图像的所述区域，以针对存在于受试者内的所述肿瘤生成一个或多个肿瘤分割边界。针对每个肿瘤并且通过使用多个器官特定分割网络，确定所述肿瘤的至少一部分位于其内的器官。基于所述一个或多个肿瘤分割边界和所述一个或多个肿瘤的至少一部分位于其内的所