Room: Exhibit Hall
Purpose: To study the computed tomography (CT) radiomic features of primary hepatocellular carcinoma (HCC) by quantifying the features of â€œfast forward and fast outâ€? imaging manifestations and identifying the normal liver tissue and HCC at different phases using the CT radiomics.
Methods: Contrast-enhanced CT images of 53 HCC patients were enrolled in this study retrospectively. The gross target volume (GTV) and partial normal liver tissue were selected as regions of interest (ROIs) in different contrast-enhanced CT phases which included plain scan phase (PSH), arterial phase (AP) and portal venous phase (PVH). The radiomic feature differences between the GTV and normal liver tissue were compared among the CT contrast-enhanced phases. And the ability to identify HCC was analyzed by ROC curve.
Results: A total of 55 radiomic features were extracted from each ROI. The number of radiomic features which rate of change was greater than 30% that differed significantly between the PSH and AP, AP and PVH, and AP and PVH was 0, 2, and 4, respectively, for normal liver tissue and 23, 31 and 12, respectively, for the GTV. There were 49, 46, and 38 features that changed significantly between the GTV and normal liver tissue in the PH, AP and PVP, respectively, and 35 common features in this comparison. In all, 13, 22, and 1 radiomic features in the PSP, AP, and PVP, respectively, exhibited an area under the receiver operating characteristic curve >0.7, sensitivity and specificity>70%.
Conclusion: Quantitative analysis of the imaging manifestations features in HCC and normal liver tissue by radiomics technology in different phases can provide a reliable reference for the accurate diagnosis and treatment of HCC.