Large language models (LLMs) often struggle to fully capture the nuanced preferences and clinical judgement of radiologists in medical report summarization even when fine-tuned on massive medical reports. This could lead to the generated radiology reports lacking the professionalism and sufficient quality required in critical diagnostic decision-making. To investigate the alignment of LLMs with radiologists by reinforcement learning from AI feedback (RLAIF) for high-quality, clinically significant impression summarization in Chest CT reports. From the data set of our hospital, 94,844 chest CT reports (median age, 56 years [IQR, 46-69 years]; 29,350 male and 34,036 female patients) were included in this retrospective study. We created a comparative data set of chest CT reports annotated by radiologists, in which the impressions of the reports generated by fine-tuned LLMs and radiologists were further reviewed and ranked by Tongyi Qianwen (Qwen) and a radiologist. Using the comparative data set, we performed further fine-tuning by reinforcement learning to align LLMs with radiologists. The performance of the aligned LLMs was evaluated on a test set in terms of metrics including ROUGE, BLEU, Precision, Recall and F1 score. The results showed that an average agreement rate of the ranking between the radiologist and Qwen is up to 77.9 %. The aligned LLM achieved consistent performance improvement over its unaligned counterpart by 2.56 %, 1.77 % and 1.13 % in terms of Precision, Recall and F1 score, respectively (P < 0.001). In this study, the results demonstrated the effect of the explicit alignment of LLMs with radiologists using RLAIF in generating impressions of chest CT reports. Additionally, the study indicated the possibility that AI can serve as an alternative to humans in providing feedback for aligning LLMs with radiologists. This study confirms the benefit of reinforcement learning from AI feedback for impression summarization, paving the way for the application of AI feedback in clinical CT report summarization. It will serve as the basis for the future development of multi-model large language models in the field of radiology.

Objectives: To predict the progression of femoral head collapse in Association Research Circulation Osseous (ARCO) Stage 2-3A osteonecrosis based on the initial bone resorption lesion. Methods: A retrospective analysis of the location, attenuation, and maximum area in coronal position (MAC) of the initial bone resorption lesion in ARCO Stage 2 and 3A was conducted in 85 cases of osteonecrosis of the femoral head (ONFH). The cases were divided into rapid and slow progression groups according to whether femoral head collapse at follow-up was greater than 2 mm. The characteristics of the bone resorption lesion between the two groups were compared by analysis of variance. Receiver operating characteristic curve was used to analyze the MAC, regions of A2, and C1 of bone resorption lesion in predicting collapse progression. Results: The MAC of initial bone resorption lesion in rapid progression group (117.8 ± 72.1 mm2) was significantly larger than slow (53.1 ± 39.5 mm2) (p < 0.001). Regions of A2 and C1 involved were significantly higher in rapid than slow progression group. The area under the receiver operating characteristic curve of MAC, regions of A2 and C1 of bone resorption lesion to predict collapse progression were 0.81, 0.72 and 0.62 respectively. A threshold MAC of 49 mm2 had sensitivity of 86.1% and specificity of 61.9% to predict collapse progression. Conclusions: The MAC of initial bone resorption lesion in ARCO Stage 2-3A ONFH can predict the progression of femoral head collapse. If it is greater than 49 mm2 and located in regions of A2 and C1, the possibility of rapid progression is high, active monitoring and intervention should be recommended. Advances in knowledge: This study is the first to find that the maximum area in coronal position of initial bone resorption lesion in ARCO Stage 2 or 3A can predict progression of the femoral head collapse with a threshold of 49 mm2. If the maximum area is larger than 49 mm2 and located in the anterolateral or lateral column of the femoral head, the possibility of rapid collapse progression is high, therefore, monitoring should be strengthened and active intervention should be considered.