Click here to

Session: Imaging Data Science for Treatment Assessment [Return to Session]

Voxel-Based Estimation of Microstructure Characteristics of Head and Neck Cancers Using Time-Dependent MR Diffusion Weighted Imaging and a Random Walk with Barriers Model

S Nejad-Davarani1*, M Mierzwa1, T Feiweier2, K Casper1, Y Cao1, (1) University of Michigan, Ann Arbor, MI, (2) Siemens Healthcare GmbH, Erlangen, Germany


TH-C-202-5 (Thursday, 7/14/2022) 10:00 AM - 11:00 AM [Eastern Time (GMT-4)]

Room 202

Purpose: To investigate a novel approach to characterize microstructure and diffusion parameters and their early changes in response to Chemoradiation Therapy (CRT) in Head and Neck cancers (HNCs) based upon MR-based microstructure imaging and a Random Walk with Barriers model (RWBM).

Methods: Six patients with HNCs had MRI scans prior to and at week 2 during CRT (2wk). Diffusion weighted (DW) images were acquired on a 3T scanner (MAGNETOM Skyra, Siemens) at 4 diffusion times (td= 5.3 to 46.6 ms), b-values = 50-1300 s/mm², and voxel size = 2.34 X 2.34 X 7.5 mm³ using a prototype spin echo sequence with oscillating and pulsed diffusion gradients. After low-pass filtering and co-registration, DW maps at all td and b-values were fitted to the Short-Time Limit and Long-Time Limit RWBM to estimate free (D₀) and restricted bulk (D∞) diffusion coefficients, cell membrane permeability (κ) and cell volume to surface ratio (V/S), by a robust iterative linear least squares fitting. Model selection at each voxel was performed to optimize data sets that satisfied the model condition. Descriptive statistics of these parameters in primary (GTVp) and nodal gross tumor volumes (GTVn) were calculated.

Results: Averaged pre-RT values of D₀, D∞, V/S and κ were 1.6 µm²/ms, 0.77 µm²/ms, 1.4 µm, and 0.12 µm/ms in GTVp, and 2.1 µm²/ms, 0.91 µm²/ms, 1.6 µm and 0.13 µm/ms in GTVn, respectively. D∞ and V/S values showed heterogeneity in GTVs. D∞ and κ values in GTVp increased significantly at 2wk (p<0.05). In GTVn, significant increases were observed for D∞ and V/S (p<0.05).

Conclusion: Tissue Microstructure parameters estimated using RWBM and DWI could be novel imaging biomarkers for therapy response assessment in HNC with further development and validation.


Diffusion, MRI, Tissue Characterization


IM- MRI : Diffusion MRI

Contact Email