Quality Assurance Assessment of Diffusion-Weighted and T2-Weighted Magnetic Resonance Imaging Registration and Contour Propagation for Head and Neck Cancer Radiotherapy

Background/Purpose: Adequate image registration of anatomic and functional MRI scans is necessary for MR-guided head and neck cancer (HNC) adaptive radiotherapy planning. Despite the quantitative capabilities of diffusion-weighted imaging (DWI) MRI for treatment plan adaptation, geometric distortion remains a considerable limitation. Therefore, we systematically investigated various deformable image registration (DIR) algorithms to co-register DWI and T2-weighted (T2W) images. Materials/Methods: We compared post-acquisition registration algorithms from three software packages (ADMIRE, Velocity, and 3D Slicer) applied to T2W and DWI MRI images in twenty HNC patients. In addition, we investigated implicit rigid registration (no algorithm applied) as a control comparator. Ground truth segmentations of radiotherapy structures (tumor and organs at risk) were generated by a physician expert on both image sequences. Three additional experts provided segmentations for five cases for interobserver variability studies. For each registration approach, structures were propagated from T2W to DWI images. These propagated structures were then compared with ground truth DWI structures using the Dice similarity coefficient (DSC), false-negative DSC, false-positive DSC, surface DSC, 95% Hausdorff distance, and mean surface distance. Results: 19 left submandibular glands, 18 right submandibular glands, 20 left parotid glands, 20 right parotid glands, 20 spinal cords, 9 brainstems, and 12 tumors were delineated. ADMIRE, the atlas-based auto segmentation DIR algorithm, demonstrated improved performance over implicit rigid registrations for most comparison metrics and structures (Bonferroni-corrected p < 0.05), while Velocity and 3D Slicer algorithms did not. Moreover, the ADMIRE methods significantly improved performance in individual and pooled analysis compared to all other methods. Interobserver variability analysis revealed no significant difference between observers (p > 0.05). Conclusions: Certain deformable registration software packages, such as those provided by ADMIRE, may be favorable for registering T2W and DWI images. These results are important to ensure the appropriate selection of registration strategies for MR-guided radiotherapy.