Clinical application research on the properties and pathological components of artificial intelligence-assisted pulmonary ground glass nodules
Objective To study the clinical application value of artificial intelligence medical imaging-assisted diagnosis system in the diagnosis of benign and malignant pulmonary ground glass nodules(GGN)and the determination of pathological components.Methods A total of 44 eligible patients with GGN were randomly selected from patients who underwent chest CT examination for inclusion in the study.According to the pathological results,the all cases was divided into the adenocarcinoma group GGN and the inflammatory lesion group GGN.Then,the adenocarcinoma group(GGN)was split into a high-proportion group and a low-proportion group based on the proportion of adherent components.The measurement parameters of the two groups were recorded,including lesion length,average CT value,CT value standard deviation,compactness,sphericity,and patient age.SPSS 20.0 software was used to analyze the differences between the two groups.Receiver operating characteristic curves(ROC)analysis was performed on quantitative parameters with statistical significance to evaluate the ability of each measurement parameter to distinguish between benign and malignant GGN and to determine the pathological components of malignant GGN.At the same time,the optimal diagnostic threshold for the measurement parameter was calculated based on the maximum Youden's index(YI),and the area under the ROC curve(AUC),sensitivity,and specificity were obtained.P＜0.05 was considered the statistical difference.Finally,independent risk factors for distinguishing between benign and malignant GGN and determining the tissue composition of malignant GGN were identified based on the binary logistic regression model.Results ①In the comparison of GGN measurement data between the adenocarcinoma group and the inflammatory lesion group,the length and diameter of the GGN lesion,average CT value,and standard deviation of CT value in the adenocarcinoma group were significantly larger than those in the inflammatory lesion group(P＜0.05).The compactness and sphericity of the GGN in the adenocarcinoma group were significantly smaller than those in the inflammatory lesion group(P＜0.05).There was no significant difference in age between the two groups of GGN patients(P＞0.05).②In the comparison of GGN measurement data between the high-proportion adenocarcinoma group and the low-proportion adenocarcinoma group,it was only found that the average CT value of GGN in the high-proportion adenocarcinoma group was significantly lower than that in the low proportion adenocarcinoma group(P＜0.05).There was no significant difference in other aspects.③The binary logistic regression model analysis shows that the independent factor for distinguishing between adenocarcinoma GGN and inflammatory GGN is the length of the lesion.The average CT value is the independent distinguishing factor between high-proportion adenocarcinoma GGN and low-proportion adenocarcinoma GGN.Conclusion Quantitative analysis of CT features based on artificial intelligence medical imaging-assisted diagnosis system helps distinguish between benign and malignant GGN,and the best performance is achieved by combining various indicators for diagnosis.However,the ability to determine the pathological components of malignant GGN is limited,and a more accurate diagnosis still needs to be made by combining other clinical indicators for comprehensive judgment.The artificial intelligence medical imaging-assisted diagnostic system has significant clinical application value in diagnosing benign and malignant GGN and determining pathological components.The combination of various indicators has the best diagnostic efficiency.
万方数据
维普
