Apparent diffusion coefficient-dependent voxelwise computed diffusion-weighted imaging: An approach for improving SNR and reducing T2 shine-through effects

Abstract

Purpose To introduce and evaluate a method for signal-to-noise ratio (SNR) improvement and T2 shine-through effect reduction in diffusion-weighted magnetic resonance imaging (DWI). Materials and Methods The proposed method uses quantitative information given by the voxel apparent diffusion coefficient (ADC) to derive voxelwise-computed DWI (vcDWI). Behavior of signal intensity variations was simulated and correlated with measurements using a dedicated phantom for DWI allowing for independent adjustment of T2-relaxivity and diffusivity. Signal-to-noise (SNR) and contrast-to-noise (CNR) ratios were measured and compared to the method of computed DWI (cDWI). Image signal was correlated with ADCs to appreciate the extent of T2 shine-through effects. Additionally, the proposed method was retrospectively applied to whole-body DWI data of 20 patients with metastatic malignancies. vcDWI was compared to cDWI and measured DWI with respect to image quality, lesion detectability, and lesion diffusivity assessment. Results Theoretically predicted signal intensity variations showed a high correlation with measured phantom data (r > 0.96). The proposed method yielded lower background signal intensity variation and higher contrast (+144%) and CNR (+358%) for diffusion-restricted phantom compartments than cDWI. Signal intensities of vcDWI showed an increased inverse correlation with phantom ADC values compared to cDWI (r = -0.86 vs. r = -0.73). Application to patient data showed higher image quality (P < 0.001) and lesion detectability (P = 0.011) using vcDWI compared to cDWI, and higher confidence for the correct identification of diffusion-restricted lesions compared to measured DWI (80/80 vs. 60/81; P = 0.013). Conclusion vcDWI is a promising approach for the reduction of T2 shine-through effects and improvement of SNR and CNR in DWI. The clinical significance of these improvements, especially regarding lesion detection, needs to be evaluated in larger prospective clinical studies.