Early (72-hour) detection of radiotherapy-induced changes in an experimental tumor model using diffusion-weighted...

To assess and compare the potential of various diffusion-related magnetic resonance imaging (MRI) parameters to detect early radiotherapy (RT)-induced changes in tumors.

Nineteen tumors in a rat model were imaged on a clinical 3T system before and 72 hours after a single RT session. Diffusion imaging was performed using an echo planar sequence containing 16 b-factors and six gradient directions. This allowed us to perform a tensor analysis of mono- and biexponential decays and a q-space analysis. Parametric maps (both trace and fractional anisotropy) were reconstructed for: 2-point apparent diffusion coefficient (ADC), 16-point ADC, biexponential amplitudes and ADCs, and height, width, and kurtosis of the probability density function (PDF). A texture analysis yielded quantities such as average and contrast. The sensitivity of diffusion-related parameters was quantified in terms of the mean relative difference (when comparing pre- and post-RT status).

Traces and anisotropies display differences in response to RT. Average traces are most sensitive for ADCs and kurtosis. Average anisotropies are all very sensitive except the slow biexponential component. The best contrast (traces) was found for the ADCs and the width of the PDF.

ADC performed well, but high b-values analysis added extra sensitive parameters for monitoring early RT-induced changes.