Assessing the Capability of ChatGPT, Google Bard, and Microsoft Bing in Solving Radiology Case Vignettes

Abstract

Background The field of radiology relies on accurate interpretation of medical images for effective diagnosis and patient care. Recent advancements in artificial intelligence (AI) and natural language processing have sparked interest in exploring the potential of AI models in assisting radiologists. However, limited research has been conducted to assess the performance of AI models in radiology case interpretation, particularly in comparison to human experts.

Objective This study aimed to evaluate the performance of ChatGPT, Google Bard, and Bing in solving radiology case vignettes (Fellowship of the Royal College of Radiologists 2A [FRCR2A] examination style questions) by comparing their responses to those provided by two radiology residents.

Methods A total of 120 multiple-choice questions based on radiology case vignettes were formulated according to the pattern of FRCR2A examination. The questions were presented to ChatGPT, Google Bard, and Bing. Two residents wrote the examination with the same questions in 3 hours. The responses generated by the AI models were collected and compared to the answer keys and explanation of the answers was rated by the two radiologists. A cutoff of 60% was set as the passing score.

Results The two residents (63.33 and 57.5%) outperformed the three AI models: Bard (44.17%), Bing (53.33%), and ChatGPT (45%), but only one resident passed the examination. The response patterns among the five respondents were significantly different (p = 0.0117). In addition, the agreement among the generative AI models was significant (intraclass correlation coefficient [ICC] = 0.628), but there was no agreement between the residents (Kappa = –0.376). The explanation of generative AI models in support of answer was 44.72% accurate.

Conclusion Humans exhibited superior accuracy compared to the AI models, showcasing a stronger comprehension of the subject matter. All three AI models included in the study could not achieve the minimum percentage needed to pass an FRCR2A examination. However, generative AI models showed significant agreement in their answers where the residents exhibited low agreement, highlighting a lack of consistency in their responses.

Keywords

artificial intelligence - Bard - Bing - ChatGPT - natural language processing - radiology - FRCR2A - fellowship

Publikationsverlauf

Artikel online veröffentlicht:
29. Dezember 2023

• References

• 1 Rathan R, Hamdy H, Kassab SE, Salama MNF, Sreejith A, Gopakumar A. Implications of introducing case based radiological images in anatomy on teaching, learning and assessment of medical students: a mixed-methods study. BMC Med Educ 2022; 22 (01) 723
• 2 Meng F, Kottlors J, Shahzad R. et al. AI support for accurate and fast radiological diagnosis of COVID-19: an international multicenter, multivendor CT study. Eur Radiol 2023; 33 (06) 4280-4291
• 3 Nijiati M, Zhang Z, Abulizi A. et al. Deep learning assistance for tuberculosis diagnosis with chest radiography in low-resource settings. J XRay Sci Technol 2021; 29 (05) 785-796
• 4 Akkus Z, Galimzianova A, Hoogi A, Rubin DL, Erickson BJ. Deep learning for brain MRI segmentation: state of the art and future directions. J Digit Imaging 2017; 30 (04) 449-459
• 5 Esmaeili M, Vettukattil R, Banitalebi H, Krogh NR, Geitung JT. Explainable artificial intelligence for human-machine interaction in brain tumor localization. J Pers Med 2021; 11 (11) 1213
• 6 Sinha RK, Deb Roy A, Kumar N, Mondal H. Applicability of ChatGPT in assisting to solve higher order problems in pathology. Cureus 2023; 15 (02) e35237
• 7 Vaishya R, Misra A, Vaish A. ChatGPT: is this version good for healthcare and research?. Diabetes Metab Syndr 2023; 17 (04) 102744
• 8 Korb KB, Nyberg EP, Oshni Alvandi A. et al. Individuals vs. BARD: experimental evaluation of an online system for structured, collaborative bayesian reasoning. Front Psychol 2020; 11: 1054
• 9 Rahsepar AA, Tavakoli N, Kim GHJ, Hassani C, Abtin F, Bedayat A. How AI responds to common lung cancer questions: ChatGPT vs Google Bard. Radiology 2023; 307 (05) e230922
• 10 Bhayana R, Krishna S, Bleakney RR. Performance of ChatGPT on a radiology board-style examination: insights into current strengths and limitations. Radiology 2023; 307 (05) e230582
• 11 Bhayana R, Bleakney RR, Krishna S. GPT-4 in radiology: improvements in advanced reasoning. Radiology 2023; 307 (05) e230987
• 12 McCoubrie P, McKnight L. Single best answer MCQs: a new format for the FRCR part 2a exam. Clin Radiol 2008; 63 (05) 506-510
• 13 Surry LT, Torre D, Trowbridge RL, Durning SJ. A mixed-methods exploration of cognitive dispositions to respond and clinical reasoning errors with multiple choice questions. BMC Med Educ 2018; 18 (01) 277
• 14 Tilmatine M, Hubers F, Hintz F. Exploring individual differences in recognizing idiomatic expressions in context. J Cogn 2021; 4 (01) 37
• 15. Khan B, Fatima H, Qureshi A. et al. Drawbacks of artificial intelligence and their potential solutions in the healthcare sector. Biomed Mater Devices 2023; 1-8
• 16 Agarwal M, Sharma P, Goswami A. Analysing the applicability of ChatGPT, Bard, and Bing to generate reasoning-based multiple-choice questions in medical physiology. Cureus 2023; 15 (06) e40977
• 17 Williams MC, Shambrook J. How will artificial intelligence transform cardiovascular computed tomography? A conversation with an AI model. J Cardiovasc Comput Tomogr 2023; 17 (04) 281-283
• 18 Korteling JEH, van de Boer-Visschedijk GC, Blankendaal RAM, Boonekamp RC, Eikelboom AR. Human- versus artificial intelligence. Front Artif Intell 2021; 4: 622364
• 19 Booth TC, Martins RDM, McKnight L, Courtney K, Malliwal R. The Fellowship of the Royal College of Radiologists (FRCR) examination: a review of the evidence. Clin Radiol 2018; 73 (12) 992-998
• 20 Ferrell B, Raskin SE, Zimmerman EB. Calibrating a transformer-based model's confidence on community-engaged research studies: decision support evaluation study. JMIR Form Res 2023; 7: e41516