Characterization of parotid gland tumors using diffusion-relaxation correlation spectrum imaging: a preliminary study

Abstract

Aim: To assess the performance of diffusion-relaxation correlation spectrum imaging (DR-CSI) in the characterization of parotid gland tumors.

Materials and methods: Twenty-five pleomorphic adenomas (PA) patients, 9 Warthin's tumors (WT) patients and 7 malignant tumors (MT) patients were prospectively recruited. DR-CSI (7 b-values combined with 5 TEs, totally 35 diffusion-weighted images) was scanned for pre-treatment assessment. Diffusion (D)-T2 signal spectrum summating all voxels were built for each patient, characterized by D-axis with range 0∼5 × 10^-3 mm²/s, and T2-axis with range 0∼300ms. With boundaries of 0.5 and 2.5 × 10^-3 mm²/s for D, all spectra were divided into three compartments labeled A (low D), B (mediate D) and C (high D). Volume fractions acquired from each compartment (V_A, V_B, V_C) were compared among PA, WT and MT. Diagnostic performance was assessed using receiver operating characteristic analysis and area under the curve (AUC).

Results: Each subtype of parotid tumors had their specific D-T2 spectrum. PA showed significantly lower V_A (8.85 ± 4.77% vs 20.68 ± 10.85%), higher V_B (63.40 ± 8.18% vs 43.05 ± 7.16%), and lower V_C (27.75 ± 8.51% vs 36.27 ± 11.09) than WT (all p<0.05). V_B showed optimal diagnostic performance (AUC 0.969, sensitivity 92.00%, specificity 100.00%). MT showed significantly higher V_A (21.23 ± 12.36%), lower V_B (37.09 ± 6.43%), and higher V_C (41.68 ± 13.72%) than PA (all p<0.05). Similarly, V_B showed optimal diagnostic performance (AUC 0.994, sensitivity 96.00%, specificity 100.00%). No significant difference of V_A, V_B and V_C was found between WT and MT.

Conclusions: DR-CSI might be a promising and non-invasive way for characterizing parotid gland tumors.