Using Texture Analysis of US Images to Differentiate Late Neck Lymph Nodes Metastasis to Nonmetastasis in Oral Squamous Cell Carcinoma

Abstract

ABSTRACT Purpose The purpose of this study was to perform texture analysis of late neck lymph nodes metastases (NLNMs) and non‐NLNMs in postoperative oral squamous cell carcinoma (SCC) on ultrasound (US) images and identify the optimal texture features for distinguishing these neck lymph nodes. Methods Fifty NLNMs and 53 non‐NLNMs were enrolled in this study. Seventy texture features were retrieved from the US images using the LIFEx software. The Mann Whitney U test was measurably utilized to survey on the off chance that there was a measurably noteworthy distinction within the textural characteristics between NLNMs and non‐NLNMs. The capacity of the surface highlights to recognize between NLNMs and non‐NLNMs was illustrated utilizing the receiver operating characteristic (ROC) analysis curves (ROC). Youden's J statistic was used to determine the cutoff positions in each ROC curve that maximized sensitivity and specificity. Results GLRLM of GLNU highlight appeared the foremost noteworthy contrast between these nodes ( p < 0.001). GLRLM of GLNU had AUC of 0.916, sensitivity of 0.92, and specificity of 0.792 when measured at the Youden cutoff value of 47.574. Conclusion Our results suggest that GLRLM of GLNU feature may be the best texture feature to differentiate from non‐NLNMs and to predict NLNMs in postoperative oral SCC.