The current status of large language models in summarizing radiology report impressions
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866912308399177728 |
|---|---|
| author | Hu, Danqing Zhang, Shanyuan Liu, Qing Zhu, Xiaofeng Liu, Bing |
| author_facet | Hu, Danqing Zhang, Shanyuan Liu, Qing Zhu, Xiaofeng Liu, Bing |
| contents | Large language models (LLMs) like ChatGPT show excellent capabilities in various natural language processing tasks, especially for text generation. The effectiveness of LLMs in summarizing radiology report impressions remains unclear. In this study, we explore the capability of eight LLMs on the radiology report impression summarization. Three types of radiology reports, i.e., CT, PET-CT, and Ultrasound reports, are collected from Peking University Cancer Hospital and Institute. We use the report findings to construct the zero-shot, one-shot, and three-shot prompts with complete example reports to generate the impressions. Besides the automatic quantitative evaluation metrics, we define five human evaluation metrics, i.e., completeness, correctness, conciseness, verisimilitude, and replaceability, to evaluate the semantics of the generated impressions. Two thoracic surgeons (ZSY and LB) and one radiologist (LQ) compare the generated impressions with the reference impressions and score each impression under the five human evaluation metrics. Experimental results show that there is a gap between the generated impressions and reference impressions. Although the LLMs achieve comparable performance in completeness and correctness, the conciseness and verisimilitude scores are not very high. Using few-shot prompts can improve the LLMs' performance in conciseness and verisimilitude, but the clinicians still think the LLMs can not replace the radiologists in summarizing the radiology impressions. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_02134 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | The current status of large language models in summarizing radiology report impressions Hu, Danqing Zhang, Shanyuan Liu, Qing Zhu, Xiaofeng Liu, Bing Computation and Language Large language models (LLMs) like ChatGPT show excellent capabilities in various natural language processing tasks, especially for text generation. The effectiveness of LLMs in summarizing radiology report impressions remains unclear. In this study, we explore the capability of eight LLMs on the radiology report impression summarization. Three types of radiology reports, i.e., CT, PET-CT, and Ultrasound reports, are collected from Peking University Cancer Hospital and Institute. We use the report findings to construct the zero-shot, one-shot, and three-shot prompts with complete example reports to generate the impressions. Besides the automatic quantitative evaluation metrics, we define five human evaluation metrics, i.e., completeness, correctness, conciseness, verisimilitude, and replaceability, to evaluate the semantics of the generated impressions. Two thoracic surgeons (ZSY and LB) and one radiologist (LQ) compare the generated impressions with the reference impressions and score each impression under the five human evaluation metrics. Experimental results show that there is a gap between the generated impressions and reference impressions. Although the LLMs achieve comparable performance in completeness and correctness, the conciseness and verisimilitude scores are not very high. Using few-shot prompts can improve the LLMs' performance in conciseness and verisimilitude, but the clinicians still think the LLMs can not replace the radiologists in summarizing the radiology impressions. |
| title | The current status of large language models in summarizing radiology report impressions |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2406.02134 |