大语言模型在肿瘤诊断中的文字报告与医学影像应用研究进展

打开文本图片集

Abstract：Largelanguagemodels（LMs）areemergingartificialintellgence technologieswithstrongtextandimageprocessing capabilities，oferingcriticalsupportfortheintelligenttransformationofhealthcareandimprovingclinicaleficiencand quality.Tisreviewummaesthecurrntaplations，techicalaturesndfuturediectionsofsiancedigosis， focusingon twokeyscenarios：automatedanalysisof textualreports （e.g.，imagingpathology，andcasesummaries）and multimodal diagnosiscombining text and medical images.Findings show thatLLMs now perform atalevel comparable to generalresident physicians incancerdiagnosis butarestillincapableofmaking specializedandprecisejudgments.Theyalso exhibitapplication-specifictraits，suchasparameter-eficientmodelsadaptedforgrasrots-levelscenarioanddivergent versatilityinmultilingualreportanalysis.Futureefortsshouldprioritizedevelopingspecialized，practicalmeicalLMs throughoptiizedfie-tuningstrategies，constructionofghqualityChinesemedicaldatasets，andintegrationwithision languagemodelstopromotetheclinicalapplicationof thesemodelsandincreasetheaccesibilityofhealthcareresources. Keywords： large language models; artificial intelligence; cancer diagnosis; pathology; medical imaging

大语言模型（LLMs是一种基于深度学习的人工智能技术，“大"字表示模型训练数据以及参数量远超既往模型，LLMs的核心功能是帮助计算机理解人类语言。（剩余21095字）