Objective To evaluate the performance of two versions of ChatGPT, GPT4 and GPT3.5, on the Final FRCR (Part A) also referred to as FRCR Part 2A radiology exam. The primary objective is to assess whether these large language models (LLMs) can effectively answer radiology test questions while providing accurate explanations for the answers. Methods The evaluation involves a total of 281 multiple choice questions, combining the 41 FRCR sample questions found on The Royal Collage of Radiologists website and 240 questions from a supplementary test bank. Both GPT4 and GPT3.5 were given the 281 questions with the answer choices, and their responses were assessed for correctness and accuracy of the explanations provided. The 41 FRCR sample questions difficulty was ranked into “low order” and “high order” questions. A significance level of p<0.05 was used. Results GPT4 demonstrated significant improvement over GPT3.5 in answering the 281 questions, achieving 76.5% correct answers compared to 52.7%, respectively (p<0.001). GPT4 demonstrated significant improvement over GPT3.5 in providing accurate explanations for the 41 FRCR sample questions, with an accuracy of 65.9% and 31.7% respectively (p=0.002). The difficulty of the question did not significantly affect the models’ performances. Conclusion The findings of this study demonstrate a significant improvement in the performance of GPT4 compared to GPT3.5 on FRCR style examination. However, the accuracy of the provided explanations might limit the models’ reliability as learning tools. Advances in Knowledge The study indirectly explores the potential of LLMs to contribute to the diagnostic accuracy and efficiency of medical imaging while raising questions about the current LLMs limitations in providing reliable explanations for radiology related questions hindering its uses for learning and in clinical practice. Highlights ChatGPT4 passed an FRCR part 2A style exam while ChatGPT3.5 did not. ChatGPT4 showed significantly higher correctness of answers and accuracy of explanations. No significant difference in performance was observed between “high order” and “lower order” questions. Explanation accuracy was lower than correct answers rate limiting the Models’ reliability as learning tools.

Pattern recognition remains a cornerstone of radiologic interpretation, as it facilitates a confident and comprehensive differential diagnosis. Certain pathologies present with specific and highly recognizable patterns on imaging modalities. These patterns can resemble familiar real-life phenomena, including cosmic bodies that surround us. We present in this article a compilation of radiologic signs across various modalities that take inspiration from cosmic phenomena. For each sign, we summarize its defining imaging appearance, typical clinical context, and common pitfalls; where available, we note diagnostic performance (e.g., sensitivity/specificity) to guide appropriate weighting in practice. By coupling memorable imagery with succinct clinical guidance, this pictorial review aims to support a faster, more accurate pattern recognition that is applicable in both low-resource and tertiary care settings, while recognizing that these signs function as educational aids rather than validated diagnostic tests. In familiarizing themselves with these classic signs, training radiologists can benefit from an engaging and memorable way of recognizing various pathological conditions.