{"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/development-and-validation-of-deep-learning","title":"Development and Validation of Deep Learning Algorithms for Detection of Critical Findings in Head CT Scans","arxiv_id":"1803.05854","date":"2018-03-13","proceeding":null,"authors":["Sasank Chilamkurthy","Rohit Ghosh","Swetha Tanamala","Mustafa Biviji","Norbert G. Campeau","Vasantha Kumar Venugopal","Vidur Mahajan","Pooja Rao","Prashant Warier"],"abstract":"Importance: Non-contrast head CT scan is the current standard for initial\nimaging of patients with head trauma or stroke symptoms.\n  Objective: To develop and validate a set of deep learning algorithms for\nautomated detection of following key findings from non-contrast head CT scans:\nintracranial hemorrhage (ICH) and its types, intraparenchymal (IPH),\nintraventricular (IVH), subdural (SDH), extradural (EDH) and subarachnoid (SAH)\nhemorrhages, calvarial fractures, midline shift and mass effect.\n  Design and Settings: We retrospectively collected a dataset containing\n313,318 head CT scans along with their clinical reports from various centers. A\npart of this dataset (Qure25k dataset) was used to validate and the rest to\ndevelop algorithms. Additionally, a dataset (CQ500 dataset) was collected from\ndifferent centers in two batches B1 & B2 to clinically validate the algorithms.\n  Main Outcomes and Measures: Original clinical radiology report and consensus\nof three independent radiologists were considered as gold standard for Qure25k\nand CQ500 datasets respectively. Area under receiver operating characteristics\ncurve (AUC) for each finding was primarily used to evaluate the algorithms.\n  Results: Qure25k dataset contained 21,095 scans (mean age 43.31; 42.87%\nfemale) while batches B1 and B2 of CQ500 dataset consisted of 214 (mean age\n43.40; 43.92% female) and 277 (mean age 51.70; 30.31% female) scans\nrespectively. On Qure25k dataset, the algorithms achieved AUCs of 0.9194,\n0.8977, 0.9559, 0.9161, 0.9288 and 0.9044 for detecting ICH, IPH, IVH, SDH, EDH\nand SAH respectively. AUCs for the same on CQ500 dataset were 0.9419, 0.9544,\n0.9310, 0.9521, 0.9731 and 0.9574 respectively. For detecting calvarial\nfractures, midline shift and mass effect, AUCs on Qure25k dataset were 0.9244,\n0.9276 and 0.8583 respectively, while AUCs on CQ500 dataset were 0.9624, 0.9697\nand 0.9216 respectively.","url_abs":"http://arxiv.org/abs/1803.05854v2","url_pdf":"http://arxiv.org/pdf/1803.05854v2.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":[{"paper_slug":"development-and-validation-of-deep-learning","repo_url":"https://github.com/ArseniusNott/60DaysofUdacity","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"development-and-validation-of-deep-learning","repo_url":"https://github.com/jarodroland/ConvOuch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1803.05854","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}