Objective: This study aimed to explore the added value of histogram analysis of the ratio of initial to final 90-second time-signal intensity AUC (AUCR) for differentiating local tumor recurrence from contrast-enhancing scar on follow-up dynamic contrast-enhanced T1-weighted perfusion MRI of patients treated for head and neck squamous cell carcinoma (HNSCC).
Materials and methods: AUCR histogram parameters were assessed among tumor recurrence (n = 19) and contrast-enhancing scar (n = 27) at primary sites and compared using the t test. ROC analysis was used to determine the best differentiating parameters. The added value of AUCR histogram parameters was assessed when they were added to inconclusive conventional MRI results.
Results: Histogram analysis showed statistically significant differences in the 50th, 75th, and 90th percentiles of the AUCR values between the two groups (p < 0.05). The 90th percentile of the AUCR values (AUCR90) was the best predictor of local tumor recurrence (AUC, 0.77; 95% CI, 0.64-0.91) with an estimated cutoff of 1.02. AUCR90 increased sensitivity by 11.7% over that of conventional MRI alone when added to inconclusive results.
Conclusion: Histogram analysis of AUCR can improve the diagnostic yield for local tumor recurrence during surveillance after treatment for HNSCC.
Keywords: MRI; differential diagnosis; head and neck squamous cell carcinoma; local tumor recurrence; perfusion imaging.