Original Article
Quantitative feature analysis of CT images of transbronchial dye markings mimicking true pulmonary ground-glass opacity lesions
Abstract
Background: Transbronchial dye marking is a preoperative localization technique aiding pulmonary resection. Post-marking computed tomography (CT) is performed to confirm the locations of the actual markings. This study aimed to evaluate the CT images of dye markings that present as ground-glass opacities (GGO), using quantitative feature analysis.
Methods: Thin-slice (1 mm) CT images of the dye markings and true ground glass nodule (GGN) lesions were obtained for quantitative analysis with gray-level co-occurrence matrix (GLCM) features. The quantification features including correlation, auto correlation, contrast, energy, entropy, and homogeneity were evaluated. Statistical analysis with boxplot was performed.
Results: GLCM features of multi-detector computed tomography (MDCT) images of the dye markings (n=13) and true GGN lesions (n=13) differed significantly in contrast, energy, entropy, auto correlation, and homogeneity. Cone beam computed tomographic (CBCT) image features of another group of dye markings (n=15) also showed a different distribution of feature values, than those of the MDCT images.
Conclusions: Quantitative analysis of the dye marking images revealed a discriminative variance, compared with those of the true GGN lesions. Furthermore, the image textures of dye markings on MDCT and CBCT also presented with obvious discrepancies.
Methods: Thin-slice (1 mm) CT images of the dye markings and true ground glass nodule (GGN) lesions were obtained for quantitative analysis with gray-level co-occurrence matrix (GLCM) features. The quantification features including correlation, auto correlation, contrast, energy, entropy, and homogeneity were evaluated. Statistical analysis with boxplot was performed.
Results: GLCM features of multi-detector computed tomography (MDCT) images of the dye markings (n=13) and true GGN lesions (n=13) differed significantly in contrast, energy, entropy, auto correlation, and homogeneity. Cone beam computed tomographic (CBCT) image features of another group of dye markings (n=15) also showed a different distribution of feature values, than those of the MDCT images.
Conclusions: Quantitative analysis of the dye marking images revealed a discriminative variance, compared with those of the true GGN lesions. Furthermore, the image textures of dye markings on MDCT and CBCT also presented with obvious discrepancies.
