{"about":{"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.","site":"https://codewithpapers.app","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","syntology":{"site":"https://syntology.ai","developers":"https://syntology.ai/developers","mcp":{"server":"https://syntology.ai/mcp","transport":"streamable-http","server_card":"https://syntology.ai/.well-known/mcp/server-card.json","auth":{"type":"trial token, no account","trial_token":"https://syntology.ai/api/oauth/trial/token","method":"POST","docs":"https://syntology.ai/developers"}},"have":"https://syntology.ai/api/graph/have?x=<method, arXiv id or title> (free, answers coverage only)","paper_base":"https://syntology.ai/paper/","atlas_base":"https://app.syntology.ai/?focus="},"machine_readable":[{"url":"https://codewithpapers.app/llms.txt","what":"the machine catalog: every machine-readable file, counted"},{"url":"https://codewithpapers.app/index/manifest.json","what":"paper-to-code index by arXiv id, with Syntology's counts"},{"url":"https://codewithpapers.app/search/manifest.json","what":"site search index (titles, authors) and its files"},{"url":"https://codewithpapers.app/download","what":"bulk files: Syntology's layer, described there"},{"url":"https://codewithpapers.app/build_manifest.json","what":"the build record: inputs, counts, exclusions, probes"}]},"url":"/paper/long-tailed-classification-of-thorax-diseases","title":"Long-Tailed Classification of Thorax Diseases on Chest X-Ray: A New Benchmark Study","arxiv_id":"2208.13365","date":"2022-08-29","proceeding":null,"authors":["Gregory Holste","Song Wang","Ziyu Jiang","Thomas C. Shen","George Shih","Ronald M. Summers","Yifan Peng","Zhangyang Wang"],"abstract":"Imaging exams, such as chest radiography, will yield a small set of common findings and a much larger set of uncommon findings. While a trained radiologist can learn the visual presentation of rare conditions by studying a few representative examples, teaching a machine to learn from such a \"long-tailed\" distribution is much more difficult, as standard methods would be easily biased toward the most frequent classes. In this paper, we present a comprehensive benchmark study of the long-tailed learning problem in the specific domain of thorax diseases on chest X-rays. We focus on learning from naturally distributed chest X-ray data, optimizing classification accuracy over not only the common \"head\" classes, but also the rare yet critical \"tail\" classes. To accomplish this, we introduce a challenging new long-tailed chest X-ray benchmark to facilitate research on developing long-tailed learning methods for medical image classification. The benchmark consists of two chest X-ray datasets for 19- and 20-way thorax disease classification, containing classes with as many as 53,000 and as few as 7 labeled training images. We evaluate both standard and state-of-the-art long-tailed learning methods on this new benchmark, analyzing which aspects of these methods are most beneficial for long-tailed medical image classification and summarizing insights for future algorithm design. The datasets, trained models, and code are available at https://github.com/VITA-Group/LongTailCXR.","url_abs":"https://arxiv.org/abs/2208.13365v1","url_pdf":"https://arxiv.org/pdf/2208.13365v1.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":"long-tailed-classification-of-thorax-diseases","repo_url":"https://github.com/vita-group/longtailcxr","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"long-tail-learning","task_name":"Long-tail Learning"},{"task_slug":"medical-image-classification","task_name":"Medical Image Classification"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[],"datasets_introduced":[{"slug":"mimic-cxr-lt","name":"MIMIC-CXR-LT","full_name":"long-tailed version of MIMIC-CXR"},{"slug":"nih-cxr-lt","name":"NIH-CXR-LT","full_name":"Long-tailed (LT) NIH ChestXRay14"}],"methods_introduced":[],"results":[{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Decoupling (cRT)","rank_in_archive_order":1,"of":15,"metrics":{"Balanced Accuracy":"0.296"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Reweighted LDAM-DRW","rank_in_archive_order":2,"of":15,"metrics":{"Balanced Accuracy":"0.275"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Class-balanced LDAM-DRW","rank_in_archive_order":3,"of":15,"metrics":{"Balanced Accuracy":"0.267"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Reweighted LDAM","rank_in_archive_order":4,"of":15,"metrics":{"Balanced Accuracy":"0.243"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Reweighted Focal Loss","rank_in_archive_order":5,"of":15,"metrics":{"Balanced Accuracy":"0.239"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Decoupling (tau-norm)","rank_in_archive_order":6,"of":15,"metrics":{"Balanced Accuracy":"0.230"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Class-balanced Softmax","rank_in_archive_order":7,"of":15,"metrics":{"Balanced Accuracy":"0.227"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Class-balanced LDAM","rank_in_archive_order":8,"of":15,"metrics":{"Balanced Accuracy":"0.225"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Reweighted  Softmax","rank_in_archive_order":9,"of":15,"metrics":{"Balanced Accuracy":"0.211"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Class-balanced Focal Loss","rank_in_archive_order":10,"of":15,"metrics":{"Balanced Accuracy":"0.191"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"MixUp","rank_in_archive_order":11,"of":15,"metrics":{"Balanced Accuracy":"0.176"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Focal Loss","rank_in_archive_order":12,"of":15,"metrics":{"Balanced Accuracy":"0.172"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Softmax","rank_in_archive_order":13,"of":15,"metrics":{"Balanced Accuracy":"0.169"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"Balanced-MixUp","rank_in_archive_order":14,"of":15,"metrics":{"Balanced Accuracy":"0.168"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-mimic-cxr-lt","task":"Long-tail Learning","dataset":"MIMIC-CXR-LT","model":"LDAM","rank_in_archive_order":15,"of":15,"metrics":{"Balanced Accuracy":"0.165"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Decoupling (cRT)","rank_in_archive_order":1,"of":15,"metrics":{"Balanced Accuracy":"0.294"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Reweighted LDAM-DRW","rank_in_archive_order":2,"of":15,"metrics":{"Balanced Accuracy":"0.289"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Class-balanced LDAM-DRW","rank_in_archive_order":3,"of":15,"metrics":{"Balanced Accuracy":"0.281"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Reweighted LDAM","rank_in_archive_order":4,"of":15,"metrics":{"Balanced Accuracy":"0.279"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Class-Balanced Softmax","rank_in_archive_order":5,"of":15,"metrics":{"Balanced Accuracy":"0.269"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Reweighted Softmax","rank_in_archive_order":6,"of":15,"metrics":{"Balanced Accuracy":"0.260"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Class-balanced LDAM","rank_in_archive_order":7,"of":15,"metrics":{"Balanced Accuracy":"0.235"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Class-Balanced Focal Loss","rank_in_archive_order":8,"of":15,"metrics":{"Balanced Accuracy":"0.232"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Decoupling (tau-norm)","rank_in_archive_order":9,"of":15,"metrics":{"Balanced Accuracy":"0.214"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Reweighted Focal Loss","rank_in_archive_order":10,"of":15,"metrics":{"Balanced Accuracy":"0.197"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"LDAM","rank_in_archive_order":11,"of":15,"metrics":{"Balanced Accuracy":"0.178"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Balanced-MixUp","rank_in_archive_order":12,"of":15,"metrics":{"Balanced Accuracy":"0.155"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Focal Loss","rank_in_archive_order":13,"of":15,"metrics":{"Balanced Accuracy":"0.122"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"MixUp","rank_in_archive_order":14,"of":15,"metrics":{"Balanced Accuracy":"0.118"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-nih-cxr-lt","task":"Long-tail Learning","dataset":"NIH-CXR-LT","model":"Softmax","rank_in_archive_order":15,"of":15,"metrics":{"Balanced Accuracy":"0.115"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2208.13365","atlas_url":"https://app.syntology.ai/?focus=2208.13365","mcp":null,"developers":"https://syntology.ai/developers","held":true,"reason":"arxiv_id_in_atlas","checked_against":{"atlas_ids":549449,"atlas_ids_sha256":"afd72399dafd9906e9062927f70655952385a3d75488558e1f0484e950ab09b7","read_at":"2026-10-05T01:46:21Z"},"agent_calls":[{"tool":"get_citation_path","arguments":{"paper_1":"2208.13365"},"arguments_in":{"paper_2":"another paper's arXiv id or title"},"call":"get_citation_path(paper_1=\"2208.13365\", paper_2=\"…\")"},{"tool":"get_concepts_for_paper","arguments":{"arxiv_id":"2208.13365"},"call":"get_concepts_for_paper(arxiv_id=\"2208.13365\")"}],"by_repo":[{"repo":"vita-group/longtailcxr","kind":"official","samples":0,"ran":0,"instrument":0,"unverified":0}],"by_repo_is":"one row per repository in the archive's code links, Syntology's graph links or the samples; samples 0 means no sample is linked to this paper, a statement about Syntology's coverage, not about the repository: with harvested_for_other_papers present, Syntology harvested the repository for that many other papers, and without it, Syntology harvested nothing from it; placed_by_identical_code counts samples whose link records no repository, placed here because identical code was harvested from this repository for another paper; ran counts executions on a synthesized input and instrument counts failures of Syntology's instrument, not of the code"},"code_links_note":null,"arxiv_metadata":null,"syntology_extracted_results":null,"record_sha256":"a0f6d52c3a0ad5e7bedd7563b84a50a11f3fbc360c7d6046a0fdc5c225eec05f","record_changed_at":"2026-09-28","record_changed_at_basis":"first_hashed"}