Objective:To apply artificial intelligence（AI）in classifying ovarian tumors on ultrasound images，and compare the diagnostic results of several sonographers with varying seniority levels.Methods:A total of 645 patients diagnosed with adnexal masses via gynecological ultrasound examination at Qilu Hospital of Shandong University from January 2021 to December 2024 were enrolled. Three deep learning architectures，i.e.，Alexnet，Densenet121，and Resnet50 were developed and used to internally test the classification effectiveness of ovarian tumors，while the optimal model was selected for external testing. Two junior sonographers and two senior sonographers were recruited to independently diagnose ovarian tumors in the external test dataset. Subsequently，the benign and malignant results of the model's predictions were disclosed to each sonographer，and their revised diagnoses on the same external test data in combination with the best AI model were recorded.Results:The optimal model achieved an accuracy of 0.941，sensitivity of 0.936，and specificity of 0.944 on the internal test dataset，and maintained robust performance on the external test dataset with accuracy of 0.891，sensitivity of 0.880，and specificity of 0.907. Compared to junior sonographers，the optimal model demonstrated significantly higher sensitivity in discriminating benign from malignant ovarian tumors（0.880 vs. 0.723，0.602；all P<0.05）. No statistically significant difference was observed in diagnostic accuracy between the optimal model and senior sonographer 1（ P=0.05）. With assistance from the optimal model，junior sonographers achieved significant improvements in both sensitivity and specificity（sensitivity：0.723 vs. 0.843，0.602 vs. 0.819；specificity：0.778 vs. 0.833，0.685 vs. 0.741；all P<0.05）. Conclusions:The optimal model achieves comparable performance to that of senior sonographers in ovarian tumor classification. With model assistance，the diagnostic performance of junior sonographers is significantly improved.

Objectives:To investigate the effect of non-optimal temperature exposure during pregnancy on the risk for preterm birth and identify the susceptible exposure window. At the same time, the interaction between non-optimal temperature and pollutants exposure during pregnancy on preterm birth was analyzed, in order to provide strong clues for the influence of non-optimal temperature exposure during pregnancy on the risk for preterm birth.Methods:A total of 1 852 pregnant women were recruited from September 2021 to June 2023 in Fujian Provincial Maternal and Child Health Care Center. Questionnaire survey was conducted, and their health records were analyzed. The permanent address of each pregnant woman was matched with Fifth Generation European Centre for Medium-Range Weather Forecasts Atmospheric Reanalysis of the Global Climate and a geo-statistical combination model based on satellite remote sensing data collection, then follow-up for pregnancy outcome was conducted. Distributed lag nonlinear model was used to assess the association between exposure to non-optimal temperature during pregnancy and the risk for preterm birth and a multiplicative interaction model was used to assess the interaction between exposure to pollutants and non-optimal temperatures during pregnancy on the risk for preterm birth.Results:After adjusting for potential confounders such as maternal age, occupation, Gross Domestic Product of the region, pre-pregnancy preconception BMI, newborn sex, the weekly susceptibility windows of extreme low temperature ( P1, P3, P5) were week 1-22 , and the weekly susceptibility windows of extreme high temperature ( P95, P97, P99) were week 27 and week 32-36. Extreme low temperature [ P1 ( OR=1.147, 95% CI: 1.041-1.265), P5 ( OR=1.284, 95% CI: 1.035-1.501)] and extreme high temperature [ P97 ( OR=1.146, 95% CI: 1.039-1.263), P99 ( OR=1.216, 95% CI: 1.099-1.345)] exhibited multiplicative interaction with PM 2.5. Conclusions:Exposure to non-optimal temperature during pregnancy was associated with an increased risk for preterm birth. The susceptible exposure windows of extreme low temperature were mainly in early and mid-pregnancy, and the susceptible exposure windows of extreme high temperature were mainly in late-pregnancy. Exposure to non-optimal temperatures and pollutants during pregnancy was associated with an increased risk for preterm birth.

OBJECTIVE: To investigate the technical key points and effectiveness of ultrasonic bone scalpel-assisted anterior controllable antedisplacement and fusion (ACAF) for treating cervical ossification of the posterior longitudinal ligament (OPLL). METHODS: Between June 2022 and December 2024, 11 OPLL patients underwent ultrasonic bone scalpel-assisted ACAF. The cohort included 8 males and 3 females, aged 49-74 years (mean, 56.7 years). The OPLL classification included 5 cases of mixed-type, 4 cases of segmental-type, and 2 cases of continuous-type cases. Ossification involved 2-5 spinal segments (mean, 3.2). Disease duration ranged from 2 to 18 months (mean, 6.2 months). The operation time, intraoperative blood loss, and complications were recorded. Pain improvement was assessed using the visual analogue scale (VAS) score, and neurological function was evaluated using Japanese Orthopaedic Association (JOA) score. Postoperative cervical CT and MRI were performed to measure spinal canal encroachment rate, spinal canal area, and spinal cord sagittal diameter. RESULTS: All operations were successfully completed. The operation time ranged from 174 to 360 minutes (mean, 255.9 minutes). The intraoperative blood loss ranged from 170 to 530 mL (mean, 345.9 mL). The C ₅ nerve root palsy occurred in 1 patient. No cerebrospinal fluid leakage, aggravated spinal cord injury, or recurrent/superior laryngeal nerve injuries occurred. All patients were followed 3-12 months (mean, 7.2 months). At last follow-up, VAS scores significantly decreased and JOA scores significantly increased compared to preoperative values ( P<0.05). According to the JOA improvement rate, the effectiveness was rated as excellent in 2 cases, good in 8, and fair in 1, with an excellent and good rate of 90.9%. Radiological re-examination revealed no implant loosening, screw breakage, or aggravated spinal stenosis. Postoperative spinal canal encroachment rate significantly decreased, while spinal canal area and spinal cord sagittal diameter significantly increased compared to preoperative measurements ( P<0.05). CONCLUSION For the treatment of cervical OPLL via ACAF, the intraoperative application of ultrasonic bone scalpel-assisted osteotomy enables precise vertebral groove creation and mobilization of the vertebra-ossification complex, thereby enhancing surgical safety and achieving satisfactory short-term effectiveness.
Humans ; Middle Aged ; Male ; Female ; Ossification of Posterior Longitudinal Ligament/diagnostic imaging* ; Aged ; Cervical Vertebrae/diagnostic imaging* ; Spinal Fusion/instrumentation* ; Treatment Outcome ; Ultrasonic Surgical Procedures/instrumentation* ; Operative Time

[Objective]To evaluate the utility of left atrial(LA)volume and strain measured by 4D auto left atrial quantification(4D auto LAQ)in differentiating pre-capillary from post-capillary pulmonary hypertension(PH),and to compare its discriminative performance with echocardiographic pulmonary to left atrial global strain ratio(ePLAGS).[Methods]A total of ninety-eight subjects with intermediate to high probability of PH were prospectively enrolled.Clinical history and laboratory data were collected.All patients underwent comprehensive transthoracic echocardiography,and LA volume and strain parameters were measured by dedicated commercial software for LA 4D analysis.[Results]Based on pulmonary arterial wedge pressure,patients were divided into pre-capillary PH group[n=39;mean age(53±24)years]and post-capillary PH group[n=59;mean age(57±18)years].Compared to the pre-capillary PH group,the post-capillary PH group showed significantly higher LAVImax,LAVImin and LAVIpreA but markedly lower LASr and LAScd.Multivariate logistic regression identified LAVImax[OR:1.40;95%CI:(1.052,1.872);P=0.021]and LAScd[OR:1.76;95%CI:(1.183,2.489);P=0.004]as independent predictors of post-capillary PH.ROC analysis demonstrated that LAVImax(AUC=0.82,P＜0.001)and LAScd(AUC=0.78,P＜0.001)had strong discriminating power for predicting post-capillary PH group,with optimal cutoff values of 35.69 mL/m2(sensitivity 86%,specificity 74%)and-9%(sensitivity 80%,specificity70%).[Conclusion]LAVImax and LAScd measured with 4D auto LAQ are robust parameters for distinguishing pre-capillary PH from post-capillary PH.

Objective To explore the impact of irregular shape of head CT hematomas on postoperative re-bleeding and prognosis of patients with spontaneous intracerebral hemorrhage(ICH)who undergo craniotomy for hematoma evacuation.Methods We retrospectively enrolled consecutive patients with ICH who underwent craniotomy for hematoma evacuation in the Department of Neurosurgery of the Second People's Hospital of Guiyang Affiliated to Guizhou Medical University from January 2019 to June 2024.Baseline and clinical data were collected form the patients,including age,gender,smoking history,drinking history,hypertension,diabetes,history of anticoagulant use,admission systolic and diastolic blood pressure,admission National Institutes of Health stroke scale(NIHSS)score,Glasgow coma scale(GCS)score,time from onset to the first head CT,neutrophil-to-lymphocyte ratio(NLR),and platelet-to-lymphocyte ratio(PLR).Admission head CT scans were used to assess hematoma shape(regular or irregular),hematoma location(basal ganglia,lobar,multifocal),hematoma volume,perihematomal edema volume,the presence of midline shift,and intraventricular extension.Volume of the hematoma was assessed 2 days after surgery.Postoperative re-bleeding is defined as an increase in the volume of the hematoma by 12.5 ml compared to the previous postoperative CT scan within 2 weeks after surgery,or the reappearance of high-density areas in the focal area of the head CT scan during follow-up after complete hematoma clearance.Conduct patients follow-ups via telephone at 6 months postoperatively to assess their modified Rankin scale(mRS)scores.The sliding dichotomy method was applied to define prognosis based on the patients' baseline characteristics and disease severity.The prognostic score was calculated using formula:10 × admission GCS score-age-0.64 × admission hematoma volume.A prognostic score＞27.672 was considered potentially favorable,while a score ≤ 27.672 was considered potentially unfavorable.For patients with a potentially favorable prognosis,an mRS score of 0-2 was defined as a good outcome,and a score of 3-6 as a poor outcome.For those with a potentially unfavorable prognosis,an mRS score of 0-3 was defined as a good outcome,and a score of 4-6 as a poor outcome.In the comparison of baseline and clinical data between patients with regular and irregular hematoma shapes,factors with P＜0.05 were included in propensity score matching(PSM)to adjust for confounding variables.A 1∶1 matching was performed using the nearest neighbor method with a caliper value set to 0.25.Variables with statistically significant differences between groups after PSM matching were included in the multivariate Logistic regression analysis to identify influencing factors for postoperative re-bleeding and poor prognosis in ICH patients undergoing craniotomy hematoma evacuation.The predictive value of irregular hematoma shape for postoperative rebleeding and poor prognosis in ICH patients was analyzed using receiver operating characteristic(ROC)curve analysis.Results(1)A total of 440 ICH patients were enrolled,including 342 males and 98 females,aged from 20 to 84 years with a mean age of(56±12)years.Statistically significant differences were observed in baseline and clinical data between patients with regular and irregular hematoma shapes before PSM,including age,admission GCS score,NIHSS score,NLR,proportion of patients with hematoma rupture into ventricles,preoperative hematoma volume,proportion of patients with midline shift,preoperative volume of hematoma surrounding edema,proportion of patients with hematoma located in multiple sites,and postoperative 2-day hematoma volume(all P＜0.05).After propensity score matching of these factors,298 ICH patients were included in the statistical analysis,comprising 228 males and 70 females,with an age range of 20 to 84 years and a mean age of(57±12)years.Following PSM,no statistically significant differences were observed in the baseline and clinical characteristics between patients with irregular and regular hematoma shapes(all P＞0.05).(2)After propensity score matching,28 patients experienced postoperative re-bleeding while 270 did not.Significant differences were observed between the two groups in the following factors:proportion of patients with a history of anticoagulant use,admission PLR,NLR,irregular hematoma shape,and hematoma volume at 2 days after operation(all P＜0.05).No statistically significant differences were found in the remaining baseline and clinical characteristics(all P＞0.05).Using postoperative re-bleeding as the dependent variable and incorporating factors with P＜0.05 from the univariate analysis as independent variables,multivariate Logistic regression analysis identified irregular hematoma shape(OR,2.821,95％CI 1.142-6.968,P=0.025)and larger hematoma volume at 2 days post-operation(OR,1.062,95％CI 1.026-1.099,P＜0.01)as independent risk factors for re-bleeding following intracranial hematoma evacuation in ICH patients.ROC curve analysis demonstrated that irregular hematoma shape predicted postoperative re-bleeding with an area under the curve(AUC)of 0.62,showing a sensitivity of 71.4％and a specificity of 52.2％.(3)After propensity score matching,174 patients had poor prognosis while 124 had good prognosis.Significant intergroup differences were observed in age,admission GCS score,NIHSS score,irregular hematoma shape,proportion of patients with hematomas located in the basal ganglia and cerebral lobes,and hematoma volume at 2 days post-operation(all P＜0.05).No statistically significant differences were found in the remaining baseline and clinical characteristics(all P＞0.05).Using poor prognosis as the dependent variable and incorporating factors with P＜0.05 from univariate analysis as independent variables,multivariate Logistic regression analysis revealed that advanced age(OR,1.039,95％CI 1.015-1.064,P=0.002),high admission NIHSS score(OR,1.068,95％CI 1.025-1.113,P=0.002),irregular hematoma shape(OR,2.675,95％CI 1.582-4.524,P＜0.01),and larger hematoma volume at 2 days post-operation(OR,1.033,95％CI 1.002-1.064,P=0.038)were independent risk factors for poor prognosis.Conversely,lobar hematoma location(OR,0.192,95％CI 0.073-0.504,P＜0.01)was identified as a protective factor.ROC curve analysis showed that irregular hematoma shape predicted poor prognosis after intracranial hematoma evacuation with an AUC of 0.61,demonstrating a sensitivity of 59.2％and specificity of 62.9％.Conclusion Irregular hematoma shape on head CT is an independent risk factor for both postoperative re-bleeding and poor prognosis in ICH patients undergoing craniotomy for hematoma evacuation.

Objectives:To investigate the effect of non-optimal temperature exposure during pregnancy on the risk for preterm birth and identify the susceptible exposure window. At the same time, the interaction between non-optimal temperature and pollutants exposure during pregnancy on preterm birth was analyzed, in order to provide strong clues for the influence of non-optimal temperature exposure during pregnancy on the risk for preterm birth.Methods:A total of 1 852 pregnant women were recruited from September 2021 to June 2023 in Fujian Provincial Maternal and Child Health Care Center. Questionnaire survey was conducted, and their health records were analyzed. The permanent address of each pregnant woman was matched with Fifth Generation European Centre for Medium-Range Weather Forecasts Atmospheric Reanalysis of the Global Climate and a geo-statistical combination model based on satellite remote sensing data collection, then follow-up for pregnancy outcome was conducted. Distributed lag nonlinear model was used to assess the association between exposure to non-optimal temperature during pregnancy and the risk for preterm birth and a multiplicative interaction model was used to assess the interaction between exposure to pollutants and non-optimal temperatures during pregnancy on the risk for preterm birth.Results:After adjusting for potential confounders such as maternal age, occupation, Gross Domestic Product of the region, pre-pregnancy preconception BMI, newborn sex, the weekly susceptibility windows of extreme low temperature ( P1, P3, P5) were week 1-22 , and the weekly susceptibility windows of extreme high temperature ( P95, P97, P99) were week 27 and week 32-36. Extreme low temperature [ P1 ( OR=1.147, 95% CI: 1.041-1.265), P5 ( OR=1.284, 95% CI: 1.035-1.501)] and extreme high temperature [ P97 ( OR=1.146, 95% CI: 1.039-1.263), P99 ( OR=1.216, 95% CI: 1.099-1.345)] exhibited multiplicative interaction with PM 2.5. Conclusions:Exposure to non-optimal temperature during pregnancy was associated with an increased risk for preterm birth. The susceptible exposure windows of extreme low temperature were mainly in early and mid-pregnancy, and the susceptible exposure windows of extreme high temperature were mainly in late-pregnancy. Exposure to non-optimal temperatures and pollutants during pregnancy was associated with an increased risk for preterm birth.

Objective To explore the impact of irregular shape of head CT hematomas on postoperative re-bleeding and prognosis of patients with spontaneous intracerebral hemorrhage(ICH)who undergo craniotomy for hematoma evacuation.Methods We retrospectively enrolled consecutive patients with ICH who underwent craniotomy for hematoma evacuation in the Department of Neurosurgery of the Second People's Hospital of Guiyang Affiliated to Guizhou Medical University from January 2019 to June 2024.Baseline and clinical data were collected form the patients,including age,gender,smoking history,drinking history,hypertension,diabetes,history of anticoagulant use,admission systolic and diastolic blood pressure,admission National Institutes of Health stroke scale(NIHSS)score,Glasgow coma scale(GCS)score,time from onset to the first head CT,neutrophil-to-lymphocyte ratio(NLR),and platelet-to-lymphocyte ratio(PLR).Admission head CT scans were used to assess hematoma shape(regular or irregular),hematoma location(basal ganglia,lobar,multifocal),hematoma volume,perihematomal edema volume,the presence of midline shift,and intraventricular extension.Volume of the hematoma was assessed 2 days after surgery.Postoperative re-bleeding is defined as an increase in the volume of the hematoma by 12.5 ml compared to the previous postoperative CT scan within 2 weeks after surgery,or the reappearance of high-density areas in the focal area of the head CT scan during follow-up after complete hematoma clearance.Conduct patients follow-ups via telephone at 6 months postoperatively to assess their modified Rankin scale(mRS)scores.The sliding dichotomy method was applied to define prognosis based on the patients' baseline characteristics and disease severity.The prognostic score was calculated using formula:10 × admission GCS score-age-0.64 × admission hematoma volume.A prognostic score＞27.672 was considered potentially favorable,while a score ≤ 27.672 was considered potentially unfavorable.For patients with a potentially favorable prognosis,an mRS score of 0-2 was defined as a good outcome,and a score of 3-6 as a poor outcome.For those with a potentially unfavorable prognosis,an mRS score of 0-3 was defined as a good outcome,and a score of 4-6 as a poor outcome.In the comparison of baseline and clinical data between patients with regular and irregular hematoma shapes,factors with P＜0.05 were included in propensity score matching(PSM)to adjust for confounding variables.A 1∶1 matching was performed using the nearest neighbor method with a caliper value set to 0.25.Variables with statistically significant differences between groups after PSM matching were included in the multivariate Logistic regression analysis to identify influencing factors for postoperative re-bleeding and poor prognosis in ICH patients undergoing craniotomy hematoma evacuation.The predictive value of irregular hematoma shape for postoperative rebleeding and poor prognosis in ICH patients was analyzed using receiver operating characteristic(ROC)curve analysis.Results(1)A total of 440 ICH patients were enrolled,including 342 males and 98 females,aged from 20 to 84 years with a mean age of(56±12)years.Statistically significant differences were observed in baseline and clinical data between patients with regular and irregular hematoma shapes before PSM,including age,admission GCS score,NIHSS score,NLR,proportion of patients with hematoma rupture into ventricles,preoperative hematoma volume,proportion of patients with midline shift,preoperative volume of hematoma surrounding edema,proportion of patients with hematoma located in multiple sites,and postoperative 2-day hematoma volume(all P＜0.05).After propensity score matching of these factors,298 ICH patients were included in the statistical analysis,comprising 228 males and 70 females,with an age range of 20 to 84 years and a mean age of(57±12)years.Following PSM,no statistically significant differences were observed in the baseline and clinical characteristics between patients with irregular and regular hematoma shapes(all P＞0.05).(2)After propensity score matching,28 patients experienced postoperative re-bleeding while 270 did not.Significant differences were observed between the two groups in the following factors:proportion of patients with a history of anticoagulant use,admission PLR,NLR,irregular hematoma shape,and hematoma volume at 2 days after operation(all P＜0.05).No statistically significant differences were found in the remaining baseline and clinical characteristics(all P＞0.05).Using postoperative re-bleeding as the dependent variable and incorporating factors with P＜0.05 from the univariate analysis as independent variables,multivariate Logistic regression analysis identified irregular hematoma shape(OR,2.821,95％CI 1.142-6.968,P=0.025)and larger hematoma volume at 2 days post-operation(OR,1.062,95％CI 1.026-1.099,P＜0.01)as independent risk factors for re-bleeding following intracranial hematoma evacuation in ICH patients.ROC curve analysis demonstrated that irregular hematoma shape predicted postoperative re-bleeding with an area under the curve(AUC)of 0.62,showing a sensitivity of 71.4％and a specificity of 52.2％.(3)After propensity score matching,174 patients had poor prognosis while 124 had good prognosis.Significant intergroup differences were observed in age,admission GCS score,NIHSS score,irregular hematoma shape,proportion of patients with hematomas located in the basal ganglia and cerebral lobes,and hematoma volume at 2 days post-operation(all P＜0.05).No statistically significant differences were found in the remaining baseline and clinical characteristics(all P＞0.05).Using poor prognosis as the dependent variable and incorporating factors with P＜0.05 from univariate analysis as independent variables,multivariate Logistic regression analysis revealed that advanced age(OR,1.039,95％CI 1.015-1.064,P=0.002),high admission NIHSS score(OR,1.068,95％CI 1.025-1.113,P=0.002),irregular hematoma shape(OR,2.675,95％CI 1.582-4.524,P＜0.01),and larger hematoma volume at 2 days post-operation(OR,1.033,95％CI 1.002-1.064,P=0.038)were independent risk factors for poor prognosis.Conversely,lobar hematoma location(OR,0.192,95％CI 0.073-0.504,P＜0.01)was identified as a protective factor.ROC curve analysis showed that irregular hematoma shape predicted poor prognosis after intracranial hematoma evacuation with an AUC of 0.61,demonstrating a sensitivity of 59.2％and specificity of 62.9％.Conclusion Irregular hematoma shape on head CT is an independent risk factor for both postoperative re-bleeding and poor prognosis in ICH patients undergoing craniotomy for hematoma evacuation.

[Objective]To evaluate the utility of left atrial(LA)volume and strain measured by 4D auto left atrial quantification(4D auto LAQ)in differentiating pre-capillary from post-capillary pulmonary hypertension(PH),and to compare its discriminative performance with echocardiographic pulmonary to left atrial global strain ratio(ePLAGS).[Methods]A total of ninety-eight subjects with intermediate to high probability of PH were prospectively enrolled.Clinical history and laboratory data were collected.All patients underwent comprehensive transthoracic echocardiography,and LA volume and strain parameters were measured by dedicated commercial software for LA 4D analysis.[Results]Based on pulmonary arterial wedge pressure,patients were divided into pre-capillary PH group[n=39;mean age(53±24)years]and post-capillary PH group[n=59;mean age(57±18)years].Compared to the pre-capillary PH group,the post-capillary PH group showed significantly higher LAVImax,LAVImin and LAVIpreA but markedly lower LASr and LAScd.Multivariate logistic regression identified LAVImax[OR:1.40;95%CI:(1.052,1.872);P=0.021]and LAScd[OR:1.76;95%CI:(1.183,2.489);P=0.004]as independent predictors of post-capillary PH.ROC analysis demonstrated that LAVImax(AUC=0.82,P＜0.001)and LAScd(AUC=0.78,P＜0.001)had strong discriminating power for predicting post-capillary PH group,with optimal cutoff values of 35.69 mL/m2(sensitivity 86%,specificity 74%)and-9%(sensitivity 80%,specificity70%).[Conclusion]LAVImax and LAScd measured with 4D auto LAQ are robust parameters for distinguishing pre-capillary PH from post-capillary PH.

Objective:To apply artificial intelligence（AI）in classifying ovarian tumors on ultrasound images，and compare the diagnostic results of several sonographers with varying seniority levels.Methods:A total of 645 patients diagnosed with adnexal masses via gynecological ultrasound examination at Qilu Hospital of Shandong University from January 2021 to December 2024 were enrolled. Three deep learning architectures，i.e.，Alexnet，Densenet121，and Resnet50 were developed and used to internally test the classification effectiveness of ovarian tumors，while the optimal model was selected for external testing. Two junior sonographers and two senior sonographers were recruited to independently diagnose ovarian tumors in the external test dataset. Subsequently，the benign and malignant results of the model's predictions were disclosed to each sonographer，and their revised diagnoses on the same external test data in combination with the best AI model were recorded.Results:The optimal model achieved an accuracy of 0.941，sensitivity of 0.936，and specificity of 0.944 on the internal test dataset，and maintained robust performance on the external test dataset with accuracy of 0.891，sensitivity of 0.880，and specificity of 0.907. Compared to junior sonographers，the optimal model demonstrated significantly higher sensitivity in discriminating benign from malignant ovarian tumors（0.880 vs. 0.723，0.602；all P<0.05）. No statistically significant difference was observed in diagnostic accuracy between the optimal model and senior sonographer 1（ P=0.05）. With assistance from the optimal model，junior sonographers achieved significant improvements in both sensitivity and specificity（sensitivity：0.723 vs. 0.843，0.602 vs. 0.819；specificity：0.778 vs. 0.833，0.685 vs. 0.741；all P<0.05）. Conclusions:The optimal model achieves comparable performance to that of senior sonographers in ovarian tumor classification. With model assistance，the diagnostic performance of junior sonographers is significantly improved.

Objective To investigate the effect of malignant cerebellar hemorrhage on 3-month prognosis of small spontaneous cerebellar hemorrhage.Methods Clinical data of 380 consecutive patients with spontaneous cerebellar hemorrhage admitted in Emergency Department of the Affil-iated Hospital of Guizhou Medical University,Neurosurgery Department of Jinyang Hospital Af-filiated to Guizhou Medical University,and Neurosurgery Department of the Second Affiliated Hospital of Guizhou Medical University from April 2014 to March 2023 were collected and retro-spectively analyzed,and finally,70 patients who met the requirements of small amount of sponta-neous cerebellar hemorrhage were enrolled in this study.They were assigned into benign cerebel-lar hemorrhage group(43 cases)and malignant cerebellar hemorrhage group(27 cases).Accord-ing to their clinical outcomes in 3 months after onset,they were divided into a good prognosis group(51 cases)and a poor prognosis group(19 cases).General clinical data,imaging data,com-plications,inflammatory indicators and prognosis were collected.After collinear diagnosis was used to exclude factors with collinear influence,the independent correlation between good progno-sis and poor prognosis was analyzed by binary logistic regression model.Finally,ROC curve was plotted to analyze the significant data.Results The maximum diameter of hematoma was signifi-cantly larger in the malignant cerebellar hemorrhage group than the benign group(P=0.021).The patients of the poor prognosis group had larger proportion of malignant cerebellar hemor-rhage,and higher neutrophil percentage,WBC count and NLR than those of the good prognosis group(P＜0.05,P＜0.01).Multivariate logistic regression analysis showed that malignant cere-bellar hemorrhage was an independent predictor of poor prognosis in 3 months(OR=6.218,95％CI:1.140-17.623,P=0.013).The sensitivity,specificity,positive predictive value,negative pre-dictive value and Youden index of malignant cerebellar hemorrhage in predicting the 3-month prognosis of patients were 63.2％,70.6％,44.4％,83.7％and 0.338,respectively,and the AUC value was 0.669.Conclusion Malignant cerebellar hemorrhage is an independent predictor of 3-month prognosis in patients with small spontaneous cerebellar hemorrhage.The patients with malignant cerebellar hemorrhage have poor prognosis than those with benign cerebellar hemorrhage.