Evaluation of different diffusion-weighted image techniques for head and neck radiation treatment: phantom and volunteer studies

PurposeTo quantitatively compare and assess the geometric distortion and apparent diffusion coefficient (ADC) accuracy of conventional single-shot EPI (SSEPI)-DWI, readout segmentation of long variable echo-trains (RESOLVE)-DWI, and BLADE-DWI techniques in both phantom and volunteer imaging studies. MethodsPhantom measurements were obtained using the QIBA DWI phantom at 0{degrees} and at ambient temperature (~17{degrees}). DW images were acquired using SSEPI-DWI, RESOLVE-DWI, and BLADE-DWI. Image geometric distortion factors (compression and dilation, shear distortion), and image shift factors were measured and compared with computed tomography images. ADC and signal-to-noise ratio (SNR) values were measured for the various DWI techniques. Images were also obtained from three healthy volunteers and three oropharynx cancer patients. Geometric distortion parameters and ADC values were analyzed in the following head and neck regions of interest (ROI): salivary glands, tonsils, and primary tumors (patient only). The metric evaluation included Dice similarity coefficient and Hausdorff distance mean. T2-weighted images were used as a reference to evaluate geometric distortion. ResultsIn the phantom experiment, the vials were prominently distorted on the SSEPI-DWI and RESOLVE-DWI images but were less distorted on the BLADE-DWI image, as determined by the overall effects of Image geometric distortion factors. The paired t-test showed no significant difference (0.15 < p < 0.88) in ADC values among the three DWI techniques; however, BLADE-DWI led to a lower SNR in vials. In the volunteer and patient studies, an ROI-based overlap metrics analysis of the salivary glands and gross tumor volumes were less distorted for BLADE-DWI than for EPI-based DWI, as determined by Wilcoxon paired signed-rank test. ConclusionBLADE-DWI demonstrated excellent geometric accuracy, with similar quantitative ADC values to those of EPI-based DW-MRI, thus potentially making this technique suitable for target volume delineation and functional assessment for head and neck radiation treatment in the future.