Potential of spectral imaging generated by contrast-enhanced dual-energy CT for lung cancer histopathological classification - A preliminary study.

Abstract

Purpose: The potential of spectral images, particularly electron density and effective Z-images, generated by dual-energy computed tomography (DECT), for the histopathologic classification of lung cancer remains unclear. This study aimed to explore which imaging factors could better reflect the histopathological status of lung cancer.

Method: The data of 31 patients who underwent rapid kV-switching DECT and subsequently underwent surgery for lung cancer were analyzed. Virtual monochromatic images (VMIs) of 35 keV and 70 keV, virtual non-contrast images (VNC), iodine content images, electron density images, and effective Z-images were reconstructed for the following analyses: 1) correlation with the ratio of the lepidic growth pattern in the whole tumor and 2) comparisons with the four histological groups: well-differentiated adenocarcinoma (WDA), moderately differentiated adenocarcinoma (MDA), and poorly differentiated adenocarcinoma (PDA) and squamous cell carcinoma (SCC).

Results: There were significant correlations between the ratio of lepidic growth pattern and 70 keV, 35 keV, VNC, and electron density images (r = -0.861, P < 0.001; r = -0.791, P < 0.001; r = -0.869, P < 0.001; r = -0.871, P < 0.001, respectively). There were significant differences in the 70 keV, 35 keV, VNC, and electron density images in the Kruskal-Wallis test (P = 0.001, P = 0.006, P < 0.001, P < 0.001, respectively). However, there were no significant differences in iodine content or effective Z-images.

Conclusions: Electron density images generated by spectral imaging may be better indicators of the histopathological classification of lung cancer.

Clinical relevance: Electron density images may have an added value in predicting the histopathological classification of lung cancer.

Key points: •The role of electron density and effective Z-images obtained using dual-energy CT in lung cancer classification remains unclear.•Electron density and virtual non-contrast images correlated better with the ratio of lepidic growth patterns in lung cancer.•Electron density imaging is a better indicator of the histopathological classification of lung cancer than effective Z-imaging.