Texture analysis of 18F-FDG PET images for the detection of cervical lymph node metastases in patients with oral squamous cell carcinoma

Abstract

Objectives: This study aimed to determine whether texture analysis of positron emission tomography (PET) images from patients with oral squamous cell carcinoma (OSCC) can accurately diagnose cervical lymph node (LN) metastases. Methods: We conducted a retrospective chart review of 166 patients with newly pathologically diagnosed OSCC. The patient inclusion criteria were LNs exhibiting high 18F-fluorodeoxyglucose uptake [maximum standardised uptake value (SUVmax) ≥2.5] and/or LNs diagnosed as metastatic cervical LNs by a board-certified nuclear medicine physician, irrespective of the SUVmax value. The patient exclusion criteria were LNs with a metabolic tumour volume (MTV) of ≤1.0 cm3. The primary predictor variables were 31 textural indices, SUVmax and MTV. The diagnostic performance was evaluated by receiver operating characteristic (ROC) curve analysis. Results: Analyses of 91 cervical LNs (48 positives, 43 negatives) indicated that long-run emphasis (LRE) could demonstrate the highest area under the ROC curve (AUC) of 0.809. Comparison of the AUCs between SUVmax and LRE indicated that the AUC of LRE was marginally higher than that of SUVmax (p = 0.067). Furthermore, the diagnostic performance of the combination of SUVmax and LRE was more accurate than that of SUVmax alone (p = 0.04). Conclusions: The combination of SUVmax and LRE may improve the diagnostic accuracy of determining cervical LN metastases.

Keywords

• Oral cancer
• Lymph node
• Heterogeneity
• Positron emission tomography
• Textural indices