2020/01/26 by Pierre-Henri Conze, Ali Emre Kavur, Conze, Pierre-Henri +11 · 1 citation
Engineering · Medicine · #COVID-19 diagnosis using AI #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #Machine Learning (cs.LG) #Medical Imaging and Analysis #Radiomics and Machine Learning in Medical Imaging #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2001.09521
openalex publication_date 2020/01/26 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
Objective : Abdominal anatomy segmentation is crucial for numerous\napplications from computer-assisted diagnosis to image-guided surgery. In this\ncontext, we address fully-automated multi-organ segmentation from abdominal CT\nand MR images using deep learning. Methods: The proposed model extends standard\nconditional generative adversarial networks. Additionally to the discriminator\nwhich enforces the model to create realistic organ delineations, it embeds\ncascaded partially pre-trained convolutional encoder-decoders as generator.\nEncoder fine-tuning from a large amount of non-medical images alleviates data\nscarcity limitations. The network is trained end-to-end to benefit from\nsimultaneous multi-level segmentation refinements using auto-context. Results :\nEmployed for healthy liver, kidneys and spleen segmentation, our pipeline\nprovides promising results by outperforming state-of-the-art encoder-decoder\nschemes. Followed for the Combined Healthy Abdominal Organ Segmentation (CHAOS)\nchallenge organized in conjunction with the IEEE International Symposium on\nBiomedical Imaging 2019, it gave us the first rank for three competition\ncategories: liver CT, liver MR and multi-organ MR segmentation. Conclusion :\nCombining cascaded convolutional and adversarial networks strengthens the\nability of deep learning pipelines to automatically delineate multiple\nabdominal organs, with good generalization capability. Significance : The\ncomprehensive evaluation provided suggests that better guidance could be\nachieved to help clinicians in abdominal image interpretation and clinical\ndecision making.\n