Objective:To assess common errors in head and neck CT angiography (CTA) image reports using four types of large language models (LLM), namely GPT-4, DeepSeek, ERNIE Bot and SparkDesk, and to assess the feasibility of using existing LLMs to support quality control of radiology reports in Chinese.Methods:The study was a cross-sectional study. Totally 1 000 head and neck CTA image reports were randomly selected using the simple random sampling method from Xuanwu Hospital, Capital Medical University in 2023, including 500 primary reports and 500 finalized reports. Two radiologists collaboratively identified six types of errors in the reports: description errors, writing errors, left-right confusion errors, diagnostic omissions, logical sequence errors, and other errors. The overall quality of the reports was assessed using a 5-point Likert scale. Subsequently, GPT-4, DeepSeek, ERNIE Bot and SparkDesk models were employed to detect the same six types of errors in the imaging reports and to provide overall scoring. The results from manual review were considered the gold standard for calculating F1 score to evaluate model performance. Intra-class correlation coefficients ( ICC) were used to assess the consistency between manual scores and the overall scores from the four LLMs. Results:In the primary imaging reports, the proportions of manually detected errors were as follows: descriptive errors 2.6% (13/500), writing errors 0.6% (3/500), left-right confusion errors 0, diagnostic omissions 6.4% (32/500), logical sequence errors 5.2% (26/500), and other errors 0. In the finalized imaging reports, the proportions of errors across the six categories were 0.2% (1/500), 0, 0, 0, 0, and 0.2% (1/500), respectively. For error detection in the primary imaging reports, the F1 scores of GPT-4 for the six error types were 0.992, 0.997, 0.997, 0.967, 0.980, and 0.992, respectively. DeepSeek achieved F1 scores of 0.980, 0.955, 0.981, 0.920, 0.995, and 0.960; ERNIE Bot scored 0.982, 0.990, 1.000, 0.956, 0.976, and 0.999; and SparkDesk achieved 0.985, 0.995, 1.000, 0.961, 0.982, and 1.000. In the detection of errors in finalized imaging reports, GPT-4′s F1 scores were 0.994, 0.995, 0.998, 0.973, 0.989, and 0.993; DeepSeek scored 0.968, 0.965, 0.985, 0.971, 0.991, and 0.983; ERNIE Bot achieved 0.996, 0.992, 1.000, 0.983, 0.999, and 0.997; and SparkDesk achieved 0.999, 0.999, 1.000, 1.000, 1.000, and 0.999. The consistency between GPT-4, DeepSeek, and SparkDesk models and human ratings was moderate, with ICC values of 0.514, 0.560, and 0.515 respectively (all P0.001); in contrast, the overall score of ERNIE Bot showed poor consistency with human ratings, with an ICC of 0.221 ( P0.001). Conclusion:LLMs demonstrate high accuracy in detecting errors in head and neck CTA imaging reports. The overall scoring of report quality shows moderate consistency with manual assessments, indicating a certain feasibility for automated quality control in reporting.

Objective To observe the value of hypoperfusion intensity ratio(HIR)of CT perfusion(CTP)for predicting infarct core progression and prognosis of acute ischemic stroke(AIS).Methods Totally 271 AIS patients were retrospectively enrolled and divided into rapid progression group(group A,n=92)and slow progression group(group B,n=179)according to infarction growth rate(IGR).Clinical data,CTP parameters,treatment strategies and patients' outcome were compared between groups.Receiver operating characteristic curve was drawn,the area under the curve(AUC)was calculated to evaluate the efficacy of HIR for predicting rapid progression in infarct core of AIS.The mediating relationships among HIR,IGR and modified Rankin scale(mRS)90 days after treatment were analyzed.Results Significant differences of National Institute of Health stroke scale(NIHSS)score,Alberta stroke program early CT score(ASPECTS),also of interval time between onset and CTP,infarct core volume,hypoperfusion volume,HIR,whether intravenous thrombolysis and mRS score 90 days after treatments were found between groups(all P＜0.05).The AUC of HIR for predicting infarct core progression of AIS was 0.856,with sensitivity and specificity was 73.91％and 81.56％,respectively,when the optimal cutoff value was 0.42.IGR was a complete mediating variable between HIR and mRS score 90 days after treatment.Conclusion HIR of CTP could be used to effectively predict infarct core progression of AIS,which completely affected prognosis through mediating variable IGR.

Objective To observe the value of hypoperfusion intensity ratio(HIR)of CT perfusion(CTP)for predicting infarct core progression and prognosis of acute ischemic stroke(AIS).Methods Totally 271 AIS patients were retrospectively enrolled and divided into rapid progression group(group A,n=92)and slow progression group(group B,n=179)according to infarction growth rate(IGR).Clinical data,CTP parameters,treatment strategies and patients' outcome were compared between groups.Receiver operating characteristic curve was drawn,the area under the curve(AUC)was calculated to evaluate the efficacy of HIR for predicting rapid progression in infarct core of AIS.The mediating relationships among HIR,IGR and modified Rankin scale(mRS)90 days after treatment were analyzed.Results Significant differences of National Institute of Health stroke scale(NIHSS)score,Alberta stroke program early CT score(ASPECTS),also of interval time between onset and CTP,infarct core volume,hypoperfusion volume,HIR,whether intravenous thrombolysis and mRS score 90 days after treatments were found between groups(all P＜0.05).The AUC of HIR for predicting infarct core progression of AIS was 0.856,with sensitivity and specificity was 73.91％and 81.56％,respectively,when the optimal cutoff value was 0.42.IGR was a complete mediating variable between HIR and mRS score 90 days after treatment.Conclusion HIR of CTP could be used to effectively predict infarct core progression of AIS,which completely affected prognosis through mediating variable IGR.

Objective:To assess common errors in head and neck CT angiography (CTA) image reports using four types of large language models (LLM), namely GPT-4, DeepSeek, ERNIE Bot and SparkDesk, and to assess the feasibility of using existing LLMs to support quality control of radiology reports in Chinese.Methods:The study was a cross-sectional study. Totally 1 000 head and neck CTA image reports were randomly selected using the simple random sampling method from Xuanwu Hospital, Capital Medical University in 2023, including 500 primary reports and 500 finalized reports. Two radiologists collaboratively identified six types of errors in the reports: description errors, writing errors, left-right confusion errors, diagnostic omissions, logical sequence errors, and other errors. The overall quality of the reports was assessed using a 5-point Likert scale. Subsequently, GPT-4, DeepSeek, ERNIE Bot and SparkDesk models were employed to detect the same six types of errors in the imaging reports and to provide overall scoring. The results from manual review were considered the gold standard for calculating F1 score to evaluate model performance. Intra-class correlation coefficients ( ICC) were used to assess the consistency between manual scores and the overall scores from the four LLMs. Results:In the primary imaging reports, the proportions of manually detected errors were as follows: descriptive errors 2.6% (13/500), writing errors 0.6% (3/500), left-right confusion errors 0, diagnostic omissions 6.4% (32/500), logical sequence errors 5.2% (26/500), and other errors 0. In the finalized imaging reports, the proportions of errors across the six categories were 0.2% (1/500), 0, 0, 0, 0, and 0.2% (1/500), respectively. For error detection in the primary imaging reports, the F1 scores of GPT-4 for the six error types were 0.992, 0.997, 0.997, 0.967, 0.980, and 0.992, respectively. DeepSeek achieved F1 scores of 0.980, 0.955, 0.981, 0.920, 0.995, and 0.960; ERNIE Bot scored 0.982, 0.990, 1.000, 0.956, 0.976, and 0.999; and SparkDesk achieved 0.985, 0.995, 1.000, 0.961, 0.982, and 1.000. In the detection of errors in finalized imaging reports, GPT-4′s F1 scores were 0.994, 0.995, 0.998, 0.973, 0.989, and 0.993; DeepSeek scored 0.968, 0.965, 0.985, 0.971, 0.991, and 0.983; ERNIE Bot achieved 0.996, 0.992, 1.000, 0.983, 0.999, and 0.997; and SparkDesk achieved 0.999, 0.999, 1.000, 1.000, 1.000, and 0.999. The consistency between GPT-4, DeepSeek, and SparkDesk models and human ratings was moderate, with ICC values of 0.514, 0.560, and 0.515 respectively (all P0.001); in contrast, the overall score of ERNIE Bot showed poor consistency with human ratings, with an ICC of 0.221 ( P0.001). Conclusion:LLMs demonstrate high accuracy in detecting errors in head and neck CTA imaging reports. The overall scoring of report quality shows moderate consistency with manual assessments, indicating a certain feasibility for automated quality control in reporting.

With the advancement of medical technology，the quality of life of children with cancer is increasingly valued.In particular，the incidence of endocrine disorders is increasing year by year，and thyroid disorders in children with cancer have attracted more and more attention.The main treatment options for cancer are chemotherapy，radiation therapy and surgery.Immunotherapy and hematopoietic stem cell transplantation bring more hope for survival to children with cancer.Although these treatment methods can alleviate the disease，they cannot avoid causing damage to the thyroid gland.Therefore，in order to detect thyroid disorders in a timely manner，thyroid function needs to be monitored during treatment and a standardized follow-up plan needs to be developed.In this article，common thyroid disorders，risk factors and management of thyroid health are discussed.The aim is to improve physicians'understanding of thyroid diseases in children with cancer，detect abnormalities in a timely manner，and intervene to improve their long-term quality of life.

Objective To investigate the CT and MRI features of intraosseous myofibroma/myofibromatosis in pediatric patients.Methods The retrospective analysis involved the examination of clinical data and imaging findings from 15 children who were diagnosed with myofibroma/myofibromatosis of bone invasion through pathological means.Subsequently,the imaging characteristics were summarized.Results CT examinations were conducted on a total of 15 patients,with 2 of them also received enhanced scans.Additionally,MRI examinations were conducted on 5 patients,with 3 of them also underwent enhanced scans.Eleven patients were diagnosed with solitary type myofibroma,with 7 cases localized in the skull and the remaining lesions observed in the maxillofacial bone.Three patients exhibited the multicentric type without any involvement of visceral organs,while one patient presented with the multicentric type accompanied by visceral involvement.The lesions exhibited a uniform soft-tissue density on plain CT scan,predominantly located between the inner and outer layers of the bone.Additionally,they displayed swelling changes and osteolytic bone destruction,with some lesions showed residual bone shell.On MRI,the lesions exhibited a uniform signal,demonstrated an isointense or slightly hypointense signal on T1WI and an isointense or slightly hyperintense signal on T2WI.The lesions displayed significantly heterogeneous enhancement on CT and MRI.Conclusion The imaging manifestations of intraosseous myofibroma/myofibromatosis in pediatric patients exhibit certain characteristics,and the residual bone shell in the lesion is helpful for diagnosis,however,distinguishing it from Langerhans cell histiocytosis of the bone remains challenging,necessitating the reliance on pathological diagnosis.

As a common emotional and psychogenic disorder, anxiety disorder seriously threats the human physical and mental health. Ventral tegmental area (VTA) is the canter of the mesocortical limbic circuit, with extensive bidirectional connections to forebrain areas, and plays important role in regulating reward, motivation, cognition, and disgust. Besides, VTA is involved in anxiety regulation by forming functional connections with multiple brain regions and connecting external stimulus information and feedback output behaviours. This article briefly summarizes the different cell subsets of VTA and its involvement in anxiety-related neural circuits.

Objective:To observe the effects of four prostaglandin E2 (PGE2) receptors (EP 1-4R) on the activation of inflammasomes and cell damage in human retinal microvascular endothelial cells (hRMEC) in a high glucose environment. Methods:The hRMEC were divided into normal group and high glucose group, and they were cultured in Dulbecco modified Eagle medium containing 5.5 and 30.0 mmol/L glucose, respectively. Flow cytometry was used to observe the apoptosis rate of the high glucose group and the normal group; enzyme chain immunosorbent assay (ELISA) was used to detect the level of PGE2 in the culture supernatant of hRMEC cells. Western blot was used to detect the protein expression of cyclooxyganese (COX2) and EP 1-4R in hRMEC. Real-time fluorescent quantitative polymerase chain reaction (qRT-PCR) was used to detect the expression of EP 1-4R mRNA in hRMEC. After 72 h of culture, the cells in the high glucose group were divided into control group, PGE2 group, EP 1-4R agonist group, PGE2+EP 1-4R inhibitor group, and dimethylsulfoxide group. According to the group, each group was given the corresponding agonist or inhibitor to continue the culture for 24 h. QRT-PCR was used to detect the expression of nucleotide-binding oligomerization structure-like receptor protein (NLRP3) and pro-interleukin (IL)-1β mRNA in each group of cells. ELISA was used to detect the content of IL-1β and lactic dehydrogenase (LDH) in the cell culture supernatant. Western blot was used to detect the expression of cleaved Caspase-1 in each group of cells. At the same time, hRMEC in a high glucose environment was given IL-1β stimulation for 24 h, and the activity of LDH in the supernatant of the cell culture medium was detected. Results:The apoptotic rate, COX2 protein expression, and PGE2 protein content in hRMEC in the high glucose group were significantly higher than those in the normal group, and they were time-dependent. Compared with the normal group, the expression levels of EP 1R, EP 2R, EP 4R protein and mRNA in hRMEC in the high glucose group were higher than those in the normal group ( P<0.05). Compared with the control group, PGE2 group ( t=4.627, P<0.01), EP 1-4R agonist group ( t=3.889, 3.583, 2.445, 3.216; P<0.05) hRMEC NLRP3 mRNA expression level was significantly increased; the expression level of pro-IL-1β mRNA increased, however the difference was not statistically significant (PGE2 group: t=1.807, P>0.05; EP 1-4R agonist group: t=1.807, 1.477, 0.302, 1.926, P>0.05). Compared with the PGE2 group, the expression of NLRP3 mRNA in hRMEC in the PGE2+EP 2R inhibitor group was significantly reduced ( t=2.812, P<0.05); the expression of pro-IL-1β mRNA in hRMEC in the PGE2+EP 3R inhibitor group was significantly increased ( t=4.113, P<0.01). The protein content of IL-1β in the cell culture supernatant of the PGE2 group, EP 1R agonist group and EP 2R agonist group was significantly higher than that of the control group ( t=5.155, 4.136, 4.817; P<0.01). Compared with PGE2 group, the protein content of IL-1β in the cell culture supernatant of the PGE2+EP 2R inhibitor group and the PGE2+EP 4R inhibitor group were significantly lower than that of the PGE2 group ( t=1.964, 4.765; P<0.05). The expression of cleaved Caspase-1 in hRMEC in the PGE2 group and EP 2R agonist group was significantly higher than that in the control group ( t=5.332, 4.889; P<0.05). The expression of cleaved Caspase-1 in hRMEC in the PGE2+EP 2R inhibitor group was significantly lower than that of the PGE2 group ( t=6.699, P<0.01). The LDH activity in the cell culture supernatant of the PGE2 group and the EP 2R agonist group was significantly higher than that of the control group ( t=4.908, 4.225; P<0.05). The activity of LDH in the cell culture supernatant of the PGE2+EP 2R inhibitor group was significantly lower than that of the PGE2 group ( t=5.301, P<0.01). Compared with the control group, the LDH activity in the culture supernatant of hRMEC cells in the high glucose environment was significantly increased ( t=3.499, P<0.05). Conclusions:The four receptors of PGE2 can activate NLRP3 and its effector molecules to varying degrees. EP 2R mainly mediates hRMEC damage under high glucose environment.

Objective:To observe the effects of miR-146a on human retinal endothelial cell (HREC) under high glucose condition.Methods:Total of 57 cases diagnosed as diabetic mellitus and 40 cases with diabetic retinopathy (DR) in Wuxi People's Hospital Affiliated to Nanjing Medical University from October to December 2013.Forty-one healthy volunteers were enrolled and served as control group.The clinical data and venous blood samples of subjects were collected.HRECs were cultured in normal glucose (5.5 mmol/L) or high glucose medium (30 mmol/L). Real-time PCR was used to detect the expression of miR-146a.The cultured HRECs were transfected with miR-146a mimic, mimic negative control, inhibitor and inhibitor negative control by lipofectamine2000, respectively.The expression of miR-146a and intercellular cell adhesion molecule-1 (ICAM-1) mRNA was examined by real-time PCR and the expression of nuclear factor-кB (NF-кB) p65 and NF-кB p65 Ser536 was detected by Western blot assay. Results:The relative expression of miR-146a mRNA in the diabetic mellitus group and DR group was 0.36±0.08 and 0.27±0.08, respectively, which were significantly lower than 1.00±0.16 in the control group (both at P<0.01). The expression of miR-146a mRNA was 0.37±0.11 in the high glucose group, which was lower than 1.00±0.18 in the normal control group ( t=5.57, P<0.01). The relative expression of miR-146a mRNA in the miR-146a mimic group was 2 540.00±105.00, which was significantly higher than 61.00±17.90 in the miR-146a mimic control group; The relative expression of miR-146a mRNA in the miR-146a inhibitor group was 0.04±0.01, which was significantly lower than 0.88±0.04 in the miR-146a inhibitor control group ( t=23.23, 17.12; both at P<0.01). The relative expression of ICAM-1 mRNA in the miR-146a mimic group was 0.35±0.12, which was significantly lower than 1.00±0.13 in the miR-146a mimic control group; The relative expression of ICAM-1 mRNA in the miR-146a inhibitor group was 2.74±0.48, which was significantly higher than 1.00±0.16 in the miR-146a inhibitor control group ( t=3.58, 3.37; both at P<0.05). The relative expression of NF-кB p65 Ser536 in the miR-146a mimic group was 0.43±0.03, which was significantly lower than 1.07±0.09 in the miR-146a mimic control group ( t=6.74, P<0.01). The relative expression of NF-кB p65 Ser536 in the miR-146a inhibitor group was 2.08±0.12, which was significantly higher than 1.00±0.01 in the miR-146a inhibitor control group ( t=8.76; P<0.01). Conclusions:miR-146a can reduce inflammation of HREC in high glucose condition through inhibiting ICAM-1 expression and NF-кB phosphorylation.

Objective To investigate the genetic characteristics and prognostic influencing factors of the middleˉhighˉrisk patients with multiple myeloma (MM) based on Mayo Stratification of Myeloma and RiskˉAdapted Therapy (mSMART) consensus guidelines. Methods A total of 179 hospitalized MM patients in Fujian Provincial Hospital from June 2009 to October 2017 were collected. Eventually, 49 patients were included except for the patients who were unable to perform mSMART stratification. According to the mSMART stratification criteria, the patients were divided into lowˉrisk group (24 cases) and middleˉhighˉrisk group (25 cases). The genetic characteristics of the two groups were analyzed to explore the relationship between mSMART stratification and clinical features. KaplanˉMeier method and logˉrank test were used to make survival analysis; logistic regression analysis was used to analyze the prognostic influencing factors in highˉrisk patients. Results The incidence of CSK1B gene amplification was the highest in the lowˉrisk group (41.7%, 10/24), while in the middleˉhighˉrisk group, the incidence of RB1 gene deletion was the highest (88.0%, 22/25). In the lowˉrisk group and the middleˉhighˉrisk group, there were no statistical differences in bone destruction, hypercalcemia, renal damage, anemia, β2 microglobulin abnormality, albumin abnormality, lactate dehydrogenase abnormality, and plasma cell ratio abnormality (all P> 0.05). Survival analysis showed that the median survival time of the middleˉhighˉrisk group was lower than that of the lowˉrisk group (23.19 months vs. 39.71 months, P＝ 0.043). Multivariate logistic regression analysis found that anemia and bone destruction were risk prognostic influencing factors for mSMART stratification as a middleˉhighˉrisk group (P＝ 0.044, P＝ 0.002). Conclusion mSMART stratification could indicate the poor prognosis for the patients with middleˉhighˉrisk, and the anemia and bone destruction are risk prognostic influencing factors for patients with middleˉhighˉrisk stratification.