Objective To explore the value of combining dual-energy CT (DECT) energy spectrum curve slope with CT value on evaluating the invasion of early lung cancer.

Methods A retrospective case-control study was conducted to retrospectively analyze the clinical data and DECT imaging data of 90 patients (44 males and 46 females, aged 34-84 years, with an average of 57.89±11.52 years) who underwent chest DECT plain scan combined with enhanced scan before surgery and were diagnosed as stage I early lung adenocarcinoma by postoperative histopathological examination in Shaanxi Provincial People’s Hospital between January 2024 and October 2025. The lesion DECT quantitative parameters lesion location, maximum diameter, plain scan CT value, slope of energy spectrum curve (λHU), iodine concentration (IC) and effective atomic number (Z_eff) and normalized iodine concentration (NIC) in arterial and venous phases were measured and recorded. According to the histopathological examination results, the patients were classified into adenocarcinoma in situ (AIS) group (n=19), microinvasive carcinoma (MIA) group (n=25) and invasive adenocarcinoma (IAC) group (n=46). The differences in parameters among the three groups were compared. AIS patients and MIA patients were combined into the non-IAC group (n=44) and were compared with the IAC group (n=46). Independent-samples t test (homogeneity of variance) was used for comparison between the two groups. One-way analysis of variance was applied for comparison among multiple groups, and SNK-q test was utilized for pairwise comparison between groups. The difference between the non-IAC group and the IAC group was compared by chi-square test. The parameters with statistically significant differences in univariate analysis were included in the collinearity diagnosis, and the degree of collinearity of the variables was evaluated by variance inflation factor (VIF). Variables with VIF>10 were determined with significant collinearity and were eliminated. Non-collinearity indicators were incorporated into multivariate logistic regression analysis to screen independent predictive risk factors for identifying early IAC. ROC curve was drawn to evaluate the diagnostic efficiency of each parameter and combined prediction model on predicting IAC. DeLong test evaluated the differences in the diagnostic efficiency of each factor and combined prediction model.

Results There were no statistical difference in lesion location among the three groups (χ²=0.226, P=0.893), while the differences in maximum lesion diameter, CT value of plain scan as well as λHU, Z_eff, IC and NIC in arterial phase and venous phase were statistically significant (F=95.373, 34.588, 50.293, 52.073, 20.929, 32.192, 55.745, 52.656, 61.093, 54.569, all P<0.001). No statistical difference was found in lesion location between the IAC group and the non-IAC group (χ²=0.206, P=0.650), while there were statistical differences in the maximum lesion diameter, plain scan CT value and arterial phase and venous phase λHU, IC, Z_eff and NIC (t=13.384, 6.578, 11.068, 9.462, 3.731, 6.277, 8.926, 9.180, 9.416, 8.998, all P<0.001). Univariate analysis results revealed that the above parameters were all associated with IAC (all P<0.05). After excluding λHU in venous phase and IC, NIC and Z_eff in arterial phase and venous phase by collinearity diagnosis, multivariate logistic regression identified maximum lesion diameter, plain scan CT value and arterial phase λHU as independent predictors for IAC (all P<0.05). ROC curve analysis results indicated that the AUCs of maximum lesion diameter and plain scan CT value were 0.857 and 0.932 respectively, and the AUC of the combined prediction model was 0.966, with a sensitivity of 91.30% and a specificity of 90.91%. The predictive efficiency of the combined model was significantly superior to that of any single parameter.

Conclusion Preoperative DECT scan is helpful to judge the invasion of patients with early lung cancer. Plain scan CT value and arterial phase λHU are effective parameters for predicting IAC. The combined evaluation of the two parameters can significantly enhance the accuracy of preoperative diagnosis.