Dear AuntMinnie Member,
Manually protocoling CT exams can take up time radiologists could be using for their interpretive responsibilities -- making implementation of large language models (LLMs) for this task appealing, according to a study that we're highlighting in this edition of our CT Insider.
A Canadian team found that the LLM GPT-4o selected optimal abdominal and pelvic CT exam protocols 96.2% of the time compared with radiologists' 88.3%. Click here to get the full details.
Once you've read that article, take a look at our coverage of a study that found that adding endovascular recanalization to regular medical care doesn't necessarily improve symptomatic nonacute intracranial artery occlusion (sNAIAO) treatment, as well as a story on how lung cancer screening reporting reveals incidental finding risk classes and a report on how clinical protocols help inform the use of cervical spine CT.
In more AI-related CT imaging news, check out our coverage of a study that assessed the performance of a radiologic ternary classification model for differentiating lung lesions on CT images, another that addressed the question of whether AI can salvage suboptimal CT studies, and a third that explored the promise of deep-learning image reconstruction for improving CT venography performance.
You'll also want to read our story on research that found low lung cancer rates on chest CT after incidental neck imaging, as well as our write-up of a study that explored what causes lung cancers to be missed on screening. Plus, see our article on a sobering report from the Harvey L. Neiman Health Policy Institute (HPI) regarding how much contrast media was administered to Medicare beneficiaries between 2011 and 2024.
Our CT content area is chock full of articles that outline the modality's benefits, and we invite you to visit it regularly. If you have CT-related topics you'd like us to consider, please contact me.
Kate Madden Yee
Senior Editor
AuntMinnie.com