Radiology
PulseAugur coverage of Radiology — every cluster mentioning Radiology across labs, papers, and developer communities, ranked by signal.
1 天有情绪数据
-
MedExpMem 通过经验记忆提高 VLM 诊断准确性
研究人员开发了 MedExpMem，这是一个旨在增强医学领域视觉语言模型 (VLM) 诊断能力的新颖框架。该系统允许 VLM 从自身的诊断失败中学习，通过经验记忆积累专业知识，而不仅仅是静态知识。MedExpMem 将这种经验组织成区分性注释，以指导鉴别推理，从而在放射学基准测试中将准确性提高了 7.0%。
-
Doctors face growing AI deepfake crisis, AMA calls for new laws
Doctors are increasingly becoming the subjects of AI-generated deepfake videos used to promote questionable products and spread misinformation, leading to concerns about public trust in the medical field. The American M…
-
RadLite微调小型LLM，用于CPU可部署的放射学AI
研究人员开发了RadLite，一种用于放射学任务的30-40亿参数小型语言模型（SLM）微调方法。该方法利用Qwen2.5-3B-Instruct和Qwen3-4B等模型的LoRA微调，显著提高了九种不同放射学应用的性能。所得模型足够小，可以量化并在消费级CPU上部署，为资源受限的临床环境提供了实用的解决方案。
-
@ thomasfuchs @ emilymbender This article is about a study of the use of # LLM # AI in radiology and other medical areas. The study’s conclusions confirm what w
Linguist Emily Bender argues that the current chatbot interface for large language models (LLMs) is not beneficial, despite their utility in tasks like transcription and translation. She criticizes the design choice of …