A fusion method based on Deep Learning and Case-Based Reasoning which improves the resulting medical image segmentations

The fusion of multiple segmentations of different biological structures is inevitable in the case where each structure has been segmented individually for performance reasons. However, when aggregating these structures for a final segmentation, conflicting pixels may appear. These conflicts can be solved by artificial intelligence techniques. Our system, integrated into the SAIAD project, carries out the fusion of deformed kidneys and nephroblastoma segmentations using the combination of Deep Learning and Case-Based Reasoning. The performances of our method were evaluated on 9 patients affected by nephroblastoma, and compared with other Al and non-Al methods adapted from the literature. The results demonstrate its effectiveness in resolving the conflicting pixels and its ability to improve the resulting segmentations. (C) 2020 Elsevier Ltd. All rights reserved.