Multiparametric voxel-based analyses of standardized uptake values and apparent diffusion coefficients of soft-tissue tumours with a positron emission tomography/magnetic resonance system: Preliminary results

Abstract

Objectives

To investigate the usefulness of voxel-based analysis of standardized uptake values (SUVs) and apparent diffusion coefficients (ADCs) for evaluating soft-tissue tumour malignancy with a PET/MR system.

Methods

Thirty-five subjects with either ten low/intermediate-grade tumours or 25 high-grade tumours were prospectively enrolled. Zoomed diffusion-weighted and fluorodeoxyglucose (¹⁸FDG)-PET images were acquired along with fat-suppressed T2-weighted images (FST2WIs). Regions of interest (ROIs) were drawn on FST2WIs including the tumour in all slices. ROIs were pasted onto PET and ADC-maps to measure SUVs and ADCs within tumour ROIs. Tumour volume, SUVmax, ADCminimum, the heterogeneity and the correlation coefficients of SUV and ADC were recorded. The parameters of high- and low/intermediate-grade groups were compared, and receiver operating characteristic (ROC) analysis was also performed.

Results

The mean correlation coefficient for SUV and ADC in high-grade sarcomas was lower than that of low/intermediate-grade tumours (−0.41 ± 0.25 vs. −0.08 ± 0.34, P < 0.01). Other parameters did not differ significantly. ROC analysis demonstrated that correlation coefficient showed the best diagnostic performance for differentiating the two groups (AUC 0.79, sensitivity 96.0%, specificity 60%, accuracy 85.7%).

Conclusions

SUV and ADC determined via PET/MR may be useful for differentiating between high-grade and low/intermediate-grade soft tissue tumours.

Key Points

• PET/MR allows voxel-based comparison of SUVs and ADCs in soft-tissue tumours.

• A comprehensive assessment of internal heterogeneity was performed with scatter plots.

• SUVmax or ADCminimum could not differentiate high-grade sarcoma from low/intermediate-grade tumours.

• Only the correlation coefficient between SUV and ADC differentiated the two groups.

• The correlation coefficient showed the best diagnostic performance by ROC analysis.