{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/spatial-aggregation-of-holistically-nested-1","title":"Spatial Aggregation of Holistically-Nested Convolutional Neural Networks for Automated Pancreas Localization and Segmentation","arxiv_id":"1702.00045","date":"2017-01-31","proceeding":null,"authors":["Holger R. Roth","Le Lu","Nathan Lay","Adam P. Harrison","Amal Farag","Andrew Sohn","Ronald M. Summers"],"abstract":"Accurate and automatic organ segmentation from 3D radiological scans is an\nimportant yet challenging problem for medical image analysis. Specifically, the\npancreas demonstrates very high inter-patient anatomical variability in both\nits shape and volume. In this paper, we present an automated system using 3D\ncomputed tomography (CT) volumes via a two-stage cascaded approach: pancreas\nlocalization and segmentation. For the first step, we localize the pancreas\nfrom the entire 3D CT scan, providing a reliable bounding box for the more\nrefined segmentation step. We introduce a fully deep-learning approach, based\non an efficient application of holistically-nested convolutional networks\n(HNNs) on the three orthogonal axial, sagittal, and coronal views. The\nresulting HNN per-pixel probability maps are then fused using pooling to\nreliably produce a 3D bounding box of the pancreas that maximizes the recall.\nWe show that our introduced localizer compares favorably to both a conventional\nnon-deep-learning method and a recent hybrid approach based on spatial\naggregation of superpixels using random forest classification. The second,\nsegmentation, phase operates within the computed bounding box and integrates\nsemantic mid-level cues of deeply-learned organ interior and boundary maps,\nobtained by two additional and separate realizations of HNNs. By integrating\nthese two mid-level cues, our method is capable of generating\nboundary-preserving pixel-wise class label maps that result in the final\npancreas segmentation. Quantitative evaluation is performed on a publicly\navailable dataset of 82 patient CT scans using 4-fold cross-validation (CV). We\nachieve a Dice similarity coefficient (DSC) of 81.27+/-6.27% in validation,\nwhich significantly outperforms previous state-of-the art methods that report\nDSCs of 71.80+/-10.70% and 78.01+/-8.20%, respectively, using the same dataset.","url_abs":"http://arxiv.org/abs/1702.00045v1","url_pdf":"http://arxiv.org/pdf/1702.00045v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"3d-medical-imaging-segmentation","task_name":"3D Medical Imaging Segmentation"},{"task_slug":"computed-tomography-ct","task_name":"Computed Tomography (CT)"},{"task_slug":"medical-image-analysis","task_name":"Medical Image Analysis"},{"task_slug":"organ-segmentation","task_name":"Organ Segmentation"},{"task_slug":"pancreas-segmentation","task_name":"Pancreas Segmentation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"superpixels","task_name":"Superpixels"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-medical-imaging-segmentation-on-tcia","task":"3D Medical Imaging Segmentation","dataset":"TCIA Pancreas-CT","model":"Holistic-nested CNN","rank_in_archive_order":1,"of":2,"metrics":{"Dice Score":"81.3"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}