(Sunday, 7/29/2018) 4:00 PM - 4:55 PM
Room: Room 207
Purpose: High b-value diffusion and dynamic contrast enhancement MRI have identified different image-phenotypes in glioblastoma (GBM). Different image-phenotypes have different characteristics and predictive values for outcomes, e.g., progression-free survival (PFS). A restricted diffusion model (RDM) has been developed to model intracellular diffusion restricted in sphere cells from diffusion weighted (DW) images acquired with bi-polar diffusion gradients. This study aims to assess cellularity and diffusion properties in different image-phenotypes in GBM and their predictive values for PFS using a RDM.
Methods: Thirty patients with histologically diagnosed GBM had DW images with 11 b-values from 0 to 2500 s/mm² on a 3T scanner. Four image-phenotype tumor volumes (TVs) were defined: TV_Gd as gross TV without surgical cavity on post-Gd T1-weighted images, TV_HCV as high intensity on b=3000 DW images, TV_CBV as high cerebral blood volume (CBV) on CBV maps, and TV_Overlap as overlap volume between TV_HCV and TV_CBV. The RDM was applied to the TVs to yield four parameters: cell radius (R), intracellular and extracellular diffusion coefficients (respective Din and Dex), and intracellular fractional volume Vin. The four parameters in different image-phenotypes were compared by Students’ t-test and assessed for predicting PFS using logistic regression.
Results: Vin values among four image-phenotype TVs were significantly different from each other (p<0.05), and had a descending order from TV_Overlap, to TV_HCV, TV_CBV and TV_Gd. Radius (R) values were also significantly different (p<0.05) from each other, except between TV_CBV and TV_Overlap. For 23 patients with available follow-up data, Din_CBV*TV_CBV was a significant parameter to predict early progression (within 6 months) from delayed progression (p=0.05).
Conclusion: Different image-phenotypes of GBM demonstrate different cellularity estimated by a RDM. One of these imaging approaches, Din_CBV*TV_CBV, is a significant predictor of early progression. The RDM will be further evaluated in a larger cohort with outcome data and tissue-pathology.
Keywords
Diffusion, Modeling, Quantitative Imaging
Taxonomy
Contact Email
