Table 1 Researches of large language models (LLMs) in radiology in 2023
No | Author(Year) | Methods | Sample size | Type of LLMs | Prompt | Scoring method |
|---|---|---|---|---|---|---|
1 | Mago and Sharma (2023) | Asking ChatGPT about radiographic anatomical landmarks, oral and maxillofacial pathologies, and their radiographic features | A questionnaire consisting of 80 questions | ChatGPT3 | No prompt | 4-point modified Likert scale |
2 | Doshi et al. (2023) | Simplify radiology reports | 254 radiology reports | ChatGPT3.5, ChatGPT4.0, Google Bard, and Microsoft Bing | Three types of prompts | Readability scores |
3 | Jeblick et al. (2023) | Simplify radiology reports | 3 radiology reports | ChatGPT | One prompt | 5-point Likert scale |
4 | Lyu et al. (2023) | Translate radiology reports into plain language | 62 chest CT and 76 brain MRI reports | ChatGPT4 | Seven types of prompts | 5-point Likert scale |