Room: Exhibit Hall | Forum 8
Purpose: The purpose of this study is to analyze the correlation between CT radiomic features and metastatic cervical lymph nodes of nasopharyngeal carcinoma (NPC).
Methods: The CT and PET/CT images of 11 NPC patients with cervical lymph node metastasis were analyzed. Gross tumor volume of nasopharynx (GTVnx) and cervical lymph node were countoured, which was defined as regions of interest (ROIs). Metastatic lymph nodes (MLN) were identified by PET/CT images. 11 GTVnx and 108 lymph nodes (31 were MLN, 77 were non-metastatic lymph nodes (NMLN)) were obtained. The radiomic features and macro features including lymph node volume, maximum cross section diameter (MCSD) were extracted from GTVnx, MLN and NMLN. And the differences of features were compared.
Results: (1) The average volume of MLN was 4.8cmÂ³ and the MCSD was from 1.1cm to 3.7cm, which were significantly greater than NMLN (average volume=0.5 cmÂ³, MCSD was from 0.4cm to 1.7cm, p <0.05). (2) 36 radiomic features were statistically significant different between MLN and NMLN. In the comparation of GTVnx with MLN and NMLN, there were 13 same radiomic features which were statistically significant different. (3) Radiomic features of MLN varied and increased, include Convex Hull Volume, Convex Hull Volume 3D, Mass, Number Of Voxel, Surface Area and Volume (Shape, AUC=0.941-0.956, Youden index=0.767-0.823). consistent with the macro features, such as volume (AUC=0.982, Youden index=0.890) and MCSD (AUC=0.982, Youden index=0.877). (4) Features which can reflect the homogeneity of MLN increased, included Intensity Direct-Energy (AUC=0.954, Youden index=0.818), Neighbor Intensity Difference 3-Busyness (AUC=0.941, Youden index=0.818) and Shape-Compactness1 (AUC=0.950, Youden index=0.875).
Conclusion: Some CT radiomic features can quantify the performance of metastatic cervical lymph node in NPC, which can provide qualitative reference for identifying the lymph node metastasis of NPC and predicting the outcome.
Not Applicable / None Entered.