Abdominal multi-organ segmentation

DiNTS architectures for volumetric (3D) segmentation of the abdominal from CT image This model uses the DiNTS model architecture searched on Medical Segmentation Decathlon Pancreas [1] and re-trained for multi-organ segmentation from abdominal CT images [2,3]. Data This model was trained on an abdominal CT dataset in portal venous phase collected from Aichi Cancer Center in Japan. Since this is a private dataset, similar models can be trained using other public multi-organ datasets like BTCV. For this bundle, we split the 420 cases into training, validation and testing with 300, 60 and 60 cases, respectively. Output