Analysis of pairwise correlations in multi-parametric PET/MR data for biological tumor characterization and treatment individualization strategies

作者:Leibfarth Sara*; Simoncic Urban; Moennich David; Welz Stefan; Schmidt Holger; Schwenzer Nina; Zips Daniel; Thorwarth Daniela
来源:European Journal of Nuclear Medicine and Molecular Imaging, 2016, 43(7): 1199-1208.
DOI:10.1007/s00259-016-3307-7

摘要

Purpose The aim of this pilot study was to explore simultaneous functional PET/MR for biological characterization of tumors and potential future treatment adaptations. To investigate the extent of complementarity between different PET/MR-based functional datasets, a pairwise correlation analysis was performed. Methods Functional datasets of N=15 head and neck (HN) cancer patients were evaluated. For patients of group A (N=7), combined PET/MR datasets including FDG-PET and ADC maps were available. Patients of group B (N=8) had FMISO-PET, DCE-MRI and ADC maps from combined PET/MRI, an additional dynamic FMISO-PET/CT acquired directly after FMISO tracer injection as well as an FDG-PET/CT acquired a few days earlier. From DCE-MR, parameter maps K-trans, v(e) and v(p) were obtained with the extended Tofts model. Moreover, parameter maps of mean DCE enhancement, (Delta S) over bar (DCE), and mean FMISO signal 0-4 min p.i., (A) over bar (FMISO), were derived. Pairwise correlations were quantified using the Spearman correlation coefficient (r) on both a voxel and a regional level within the gross tumor volume. Results Between some pairs of functional imaging modalities moderate correlations were observed with respect to the median over all patient datasets, whereas distinct correlations were only present on an individual basis. Highest inter-modality median correlations on the voxel level were obtained for FDG/FMISO (r = 0.56), FDG/(A) over bar (FMISO) (r = 0.55), (A) over bar (FMISO)/(Delta S) over bar (DCE) (r = 0.46), and FDG/ADC (r = -0.39). Correlations on the regional level showed comparable results. Conclusion The results of this study suggest that the examined functional datasets provide complementary information. However, only pairwise correlations were examined, and correlations could still exist between combinations of three or more datasets. These results might contribute to the future design of individually adapted treatment approaches based on multiparametric functional imaging.

  • 出版日期2016-7