ObjectiveTo explore the efficiency of Ki-67 expression and CT imaging features in predicting the degree of invasion of lung adenocarcinoma. MethodsThe clinical data of 217 patients with pulmonary nodules, who were diagnosed as suspicious lung cancer by multi-disciplinary treatment clinic of pulmonary nodules in our hospital from September 2017 to August 2021, were retrospectively analyzed. There were 84 males and 133 females, aged 52 (25-84) years. The patients were divided into two groups according to the infiltration degree, including an adenocarcinoma in situ and microinvasive adenocarcinoma group (n=145) and an invasive adenocarcinoma group (n=72). ResultsThere was no statistical difference in the age and gender between the two groups (P>0.05). The univariate analysis showed that CK-7, P63, P40 and CK56 expressions were not different between the two groups (P=0.172, 0.468, 0.827, 0.313), while Napsin A, TTF-1 and Ki-67 expressions were statistically different (P=0.002, 0.020, <0.001). The multivariate analysis showed that Ki-67 expression was statistically different between the two groups (P<0.001). Ki-67 was positively correlated with malignant features of CT images and the degree of lung adenocarcinoma invasion (P<0.05). Ki-67 and CT imaging features alone could predict the degree of lung adenocarcinoma invasion, but their sensitivity and specificity were not high. Ki-67 combined with CT imaging features could achieve a higher prediction efficiency.ConclusionCompared with Ki-67 or CT imaging features alone, the combined prediction of Ki-67 and imaging features is more effective, which is of great significance for clinicians to select the appropriate operation occasion.
Objective To discuss the main pathological types and imaging features of pulmonary nodules highly suspected to be malignant in clinical practice but pathologically confirmed to be benign. Method A retrospective analysis was conducted on the clinical data of the patients with pulmonary nodules, who were initially highly suspected of malignancy but were pathologically confirmed as benign, treated at the First Affiliated Hospital of Xiamen University from December 2020 to April 2023. Based on the results of preoperative discussions, the patients were divided into a benign group and a suspicious malignancy group. Results Finally, 232 patients were collected, including 112 males and 120 females, with an average age of 51 years. There were 127 patients in the benign group, and 105 patients in the suspicious malignancy group. There was no statistical difference in age, gender, symptoms, smoking history, and tumor history between the two groups (P>0.05). However, there were statistical differences in nodule density, CT values, margins, shapes, andmalignant signs between the two groups (P<0.05). The analysis showed that in suspiciously malignant pulmonary nodules, the solid group was mainly characterized by collagen nodules and fibrous tissue hyperplasia (33.3%), tuberculosis (20.3%), and fungal infection (18.5%), while the non-solid group was primarily composed of collagen nodules and fibrous tissue hyperplasia (39.2%) and atypical adenomatous hyperplasia (17.6%). ConclusionThe benign pulmonary nodules suspected of malignancy are pathologically characterized by the presence of collagen nodules and fibrous tissue hyperplasia, tuberculosis, atypical adenomatous hyperplasia, and fungal infections. In terms of imaging features, they typically present as non-solid nodules, accompanied by signs of malignancy such as spiculation, lobulation, cavitation, and pleural retraction.
Following the rapid advancement of artificial intelligence technologies, especially the development of large language models like ChatGPT, the field of medical clinical practice is undergoing an unprecedented technological revolution. These advanced technologies, through efficient processing and analysis of large datasets, not only provide medical professionals with auxiliary diagnoses and treatment suggestions but also significantly enhance the quality and efficiency of medical education. This study conducts a comprehensive analysis and review of the applications of large language models in various aspects, including clinical inquiry, history collection, medical literature writing, clinical decision support, optimization of medical portal websites, patient health management, medical education, academic research, and scientific writing. However, the application of these technologies is not without flaws and presents several limitations and ethical challenges. This paper focuses on challenges related to technological errors, academic dishonesty, abuse risks, over-reliance, possibilities of misdiagnosis and treatment errors, and issues of accountability. In conclusion, large language models demonstrate tremendous potential in the integration and advancement of medical practices. Nevertheless, while fully harnessing the benefits brought by ChatGPT, it is essential to acknowledge and address these ethical challenges to ensure that the application of ChatGPT in the medical field is responsible and effective.