CT features and histogram analysis of non-contrast images for differentiating malignant and benign mediastinal lymph...

To evaluate the diagnostic value of CT features and histogram analysis in distinguishing between malignant and benign mediastinal lymph nodes in patients with non-small cell lung cancer (NSCLC).

This retrospective study analyzed non-contrast chest CT images from 40 NSCLC patients, comprising 80 pathology-proven mediastinal lymph nodes (46 benign, 34 metastasis). Morphologic features, including size, shape, margins, and internal composition, were independently assessed by two radiologists. Histogram analysis was conducted using the Synapse Vincent system with six parameters: mean attenuation, mean positive pixel (MPP), standard deviation (SD), skewness, kurtosis, and entropy. Statistical analysis included the Mann-Whitney test for continuous data, Fisher's exact test for categorical data, and receiver-operating characteristic (ROC) curve analysis to assess diagnostic accuracy, with statistical significance set at p < 0.05.

Malignant lymph nodes demonstrated significantly larger sizes (p < 0.001), ill-defined margins (p = 0.024), irregular shapes (p < 0.001), and the presence of necrotic areas (p < 0.001). A nodal size cutoff of 13.0 mm and volume of 1.632 ml were strongly associated with malignancy, yielding high diagnostic accuracy with sensitivities of 70.6% and 73.5% and specificities of 95.7% and 87.0%, respectively. Significant differences were observed between benign and malignant lymph nodes in several CT histogram parameters, including mean attenuation (p = 0.004), skewness (p = 0.041), kurtosis (p = 0.005), and entropy (p < 0.001). The integrating all CT histogram parameters yielded an area under the curve (AUC) of 0.870 for differentiating between benign and malignant lymph nodes.

The combination of morphologic CT features and CT histogram analysis offers a robust method for differentiating malignant from benign mediastinal lymph nodes in NSCLC patients, potentially enhancing diagnostic accuracy and informing treatment strategies.