An easy method for the evaluation of visual tumor heterogeneity of (18)F-FDG PET/CT using 10-step color scale: A correlative study in patients with lung cancer

Lung Cancer
28/07/2020

Hell J Nucl Med. 2020 Jul 27:s002449912105. doi: 10.1967/s002449912105. Online ahead of print.

ABSTRACT

OBJECTIVE: The purpose of this study was to assess the correlation between visual assessment by the peritumoral halo layer (PHL) method and analysis with texture and intensity histogram metrics for evaluating tumor heterogeneity (TH) in fluorine-18-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) of patients with lung cancer.

SUBJECTS AND METHODS: We retrospectively reviewed the 64 patients with records of 18F-FDG PET/CT prior to treatment. After tumor segmentation by the PHL method, three visual patterns: the existence of cold defect, the location of hottest core, and number of irregular layers can be obtained on 18F-FDG PET/CT images. To examine the correlation, first-order entropy was extracted by texture analysis and the area under curve-cumulative SUV-volume histogram (AUC-CSH) value was used in the intensity histogram analysis. A correlation between the visual assessment and the parameters of the metrics analysis was evaluated. We also evaluated the inter-correlation among the visual assessment and correlation between metabolic tumor volume (MTV) and parameters of TH.

RESULTS: A significant correlation was noted between visual assessment and quantitative indices of texture and intensity histogram analysis with high inter-observer agreement. Additionally, a significant inter-correlation between the existence of cold defect, location of hottest core, and the number of irregular layers was observed. Metabolic tumor volume was correlated with all of the parameters in the 18F-FDG PET/CT images, such as AUC-CSH, entropy, existence of cold defect, location of hottest core, and number of irregular layers.

CONCLUSION: In this study, we provide a new visual assessment criterion of TH with a strong inter-observer agreement. The visual analysis using the PHL method could be a potential marker for evaluating heterogeneous tumors.