{"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/patch-based-convolutional-neural-network-for","title":"Patch-based Convolutional Neural Network for Whole Slide Tissue Image Classification","arxiv_id":"1504.07947","date":"2015-04-29","proceeding":"CVPR 2016 6","authors":["Le Hou","Dimitris Samaras","Tahsin M. Kurc","Yi Gao","James E. Davis","Joel H. Saltz"],"abstract":"Convolutional Neural Networks (CNN) are state-of-the-art models for many\nimage classification tasks. However, to recognize cancer subtypes\nautomatically, training a CNN on gigapixel resolution Whole Slide Tissue Images\n(WSI) is currently computationally impossible. The differentiation of cancer\nsubtypes is based on cellular-level visual features observed on image patch\nscale. Therefore, we argue that in this situation, training a patch-level\nclassifier on image patches will perform better than or similar to an\nimage-level classifier. The challenge becomes how to intelligently combine\npatch-level classification results and model the fact that not all patches will\nbe discriminative. We propose to train a decision fusion model to aggregate\npatch-level predictions given by patch-level CNNs, which to the best of our\nknowledge has not been shown before. Furthermore, we formulate a novel\nExpectation-Maximization (EM) based method that automatically locates\ndiscriminative patches robustly by utilizing the spatial relationships of\npatches. We apply our method to the classification of glioma and non-small-cell\nlung carcinoma cases into subtypes. The classification accuracy of our method\nis similar to the inter-observer agreement between pathologists. Although it is\nimpossible to train CNNs on WSIs, we experimentally demonstrate using a\ncomparable non-cancer dataset of smaller images that a patch-based CNN can\noutperform an image-based CNN.","url_abs":"http://arxiv.org/abs/1504.07947v5","url_pdf":"http://arxiv.org/pdf/1504.07947v5.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":"patch-based-convolutional-neural-network-for","repo_url":"https://github.com/MichaelIcaza/localization-from-image-labels","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1504.07947","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}