Ability of ChatGPT to generate competent radiology reports for distal radius fracture by use of RSNA template items and integrated AO classifier

Wolfram A Bosbach; Jan F Senge; Bence Nemeth; Siti H Omar; Milena Mitrakovic; Claus Beisbart; András Horváth; Johannes Heverhagen; Keivan Daneshvar

doi:10.1067/j.cpradiol.2023.04.001

Ability of ChatGPT to generate competent radiology reports for distal radius fracture by use of RSNA template items and integrated AO classifier

Curr Probl Diagn Radiol. 2024 Jan-Feb;53(1):102-110. doi: 10.1067/j.cpradiol.2023.04.001. Epub 2023 Apr 17.

Authors

Wolfram A Bosbach¹, Jan F Senge², Bence Nemeth³, Siti H Omar⁴, Milena Mitrakovic⁵, Claus Beisbart⁶, András Horváth⁷, Johannes Heverhagen⁵, Keivan Daneshvar⁵

Affiliations

¹ Department of Diagnostic, Interventional and Pediatric Radiology (DIPR), Inselspital, Bern University Hospital, University of Bern, Switzerland. Electronic address: WolframAndreas.Bosbach@Insel.CH.
² Department of Mathematics and Computer Science, University of Bremen, Bremen, Germany; Max-Planck Dioscuri Centre for Topological Data Analysis, Warsaw, Poland.
³ Department of Diagnostic, Interventional and Pediatric Radiology (DIPR), Inselspital, Bern University Hospital, University of Bern, Switzerland; Faculty of Information Technology and Bionics, Pázmány Péter Catholic University, Budapest, Hungary.
⁴ Department of Diagnostic, Interventional and Pediatric Radiology (DIPR), Inselspital, Bern University Hospital, University of Bern, Switzerland; Department for Radiology, Kuala Lumpur Hospital, Ministry of Health of Malaysia, Kuala Lumpur, Malaysia.
⁵ Department of Diagnostic, Interventional and Pediatric Radiology (DIPR), Inselspital, Bern University Hospital, University of Bern, Switzerland.
⁶ Institute of Philosophy, University of Bern, Bern, Switzerland; Center for Artificial Intelligence in Medicine, University of Bern, Bern, Switzerland.
⁷ Faculty of Information Technology and Bionics, Pázmány Péter Catholic University, Budapest, Hungary.

PMID: 37263804
DOI: 10.1067/j.cpradiol.2023.04.001

Abstract

The amount of acquired radiology imaging studies grows worldwide at a rapid pace. Novel information technology tools for radiologists promise an increase of reporting quality and as well quantity at the same time. Automated text report drafting is one branch of this development. We defined for the present study in total 9 cases of distal radius fracture. Command files structured according to a template of the Radiological Society of North America (RSNA) and to Arbeitsgemeinschaft Osteosynthese (AO) classifiers were given as input to the natural language processing tool ChatGPT. ChatGPT was tasked with drafting an appropriate radiology report. A parameter study (n = 5 iterations) was performed. An overall high appraisal of ChatGPT radiology report quality was obtained in a score card based assessment. ChatGPT demonstrates the capability to adjust output files in response to minor changes in input command files. Existing shortcomings were found in technical terminology and medical interpretation of findings. Text drafting tools might well support work of radiologists in the future. They would allow a radiologist to focus time on the observation of image details and patient pathology. ChatGPT can be considered a substantial step forward towards that aim.

MeSH terms

Diagnostic Imaging
Humans
North America
Radiography
Radiology*
Wrist Fractures*