The predictive value of artificial intelligence-assisted high-resolution CT in degree of invasion of early lung adenocarcinoma with subsolid nodules
10.3969/j.issn.1002-1671.2024.04.010
- VernacularTitle:人工智能辅助下高分辨率CT对亚实性结节早期肺腺癌浸润程度的预测价值
- Author:
Ping LI
1
;
Hailiang WANG
;
Yiping GAO
;
Xiaohua ZHANG
Author Information
1. 嘉兴市中医医院放射科,浙江 嘉兴 314000
- Keywords:
subsolid nodules;
high-resolution computed tomography;
artificial intelligence
- From:
Journal of Practical Radiology
2024;40(4):557-561
- CountryChina
- Language:Chinese
-
Abstract:
Objective To investigate the predictive value of artificial intelligence(AI)-assisted high-resolution computed tomo-graphy(HRCT)in degree of invasion of early lung adenocarcinoma with subsolid nodules(SSN).Methods Ninety patients(a total of 90 SSN)with pulmonary SSN on AI-assisted HRCT scans were selected as the study subjects.According to the degree of invasion in pathology,the cases were divided into invasive adenocarcinoma(IAC)group(48 cases)and non-invasive adenocarcinoma(non-IAC)group(42 cases).The differences in general information,pulmonary SSN morphological characteristics,and related parameters were compared between the two groups.The influencing factors were analyzed using a multiple logistic regression model,and a nomo-gram prediction model was constructed.The diagnostic efficacy of the model was evaluated using receiver operating characteristic(ROC)curves,calibration curves,and clinical decision curves.Results Between the two groups,there were statistically significant differences in age,nodular nature,shape,tumor-lung interface,vacuole sign,burr sign,air-bronchial sign,vascular bunching sign and pleural indentation sign.Compared with the non-IAC group,the CT maximum value,energy,variance of CT value,CT average value,maximum surface area,mean long and short diameter,3D long diameter,surface area,volume,mass and entropy of the IAC group significantly increased,while the CT minimum value,compactness and sphericity of the IAC group significantly decreased.The results of multiple logistic regression model showed that burr sign,vascular bunching sign,tumor-lung interface,mass,pleural indentation sign,and CT average value were independent risk factors affecting the development of SSN into early IAC.Finally,a nomogram pre-diction model was constructed based on risk factors,and the results of ROC curves,calibration curves and clinical decision curves showed that the predictive model had good diagnostic efficacy.Conclusion HRCT imaging features assisted by AI have a high pre-dictive value for the degree of invasion in early lung adenocarcinoma with SSN,and focusing on the burr sign,vascular bunching sign,tumor-lung interface,mass,pleural indentation sign,and CT average value can improve clinical treatment and prognosis for patients.
