Diffusion‐weighted MRI for tumour volume delineation: Comparison with morphological MRI

Abstract

Diffusion-weighted magnetic resonance imaging (dwMRI) is sensitive to tissue microstructure on the cellular level and may therefore help to define biological tumour subvolumes and add complementary information to morphology-based cancer treatment protocols and therapy monitoring. The purpose of this study was therefore to evaluate the potential of dwMRI as compared with morphological MRI (mMRI) for tumour volume delineation using a nude rat human tumour xenograft model. Sixteen tumour-bearing rats (10 H1299, six FaDu) were examined with mMRI (T2-weighted true fast imaging with steady precession (TrueFISP), T1-weighted fast low angle shot (FLASH), T2-weighted dual echo steady state (DESS)) and echo-planar dwMRI in a clinical scanner at 1.5 T. For each method, we compared tumour volume and intra- and inter-observer variability of tumour outer edge delineation (disregarding intra-tumoural structure) as well as tumour signal-to-noise ratio (SNR) and tumour-to-muscle contrast-to-noise ratio (CNR). Tumours were visualised with significantly higher SNR and CNR in dwMRI. Median tumour volumes as measured by dwMRI (3.5 cm(3)) and mMRI (TrueFISP: 3.3 cm(3); FLASH: 3.3 cm(3); DESS: 3.2 cm(3)) were not significantly different and significantly correlated. Related to partial volume effects, the intra- and inter-observer variability of dwMRI (intra/inter: 12%/12%) was larger than for mMRI (TrueFISP: 4%/4%; FLASH: 5%/5%; DESS: 5%/5%). In conclusion, dwMRI allows tumour delineation with overall volume estimation comparable with mMRI approaches but slightly higher observer variability. Thus, besides tumour outline, it may potentially supplement morphology-based therapy planning and monitoring with additional biological information.