3111 - Quantification of Postural Effects on Head and Neck Anatomy Using Upright and Supine Low-Field Open MRI

Purpose/Objective(s): Upright radiotherapy offers potential dosimetric advantages, yet quantitative data on posture-related anatomical changes compared to standard supine positioning remain scarce. Understanding these variations is critical, because the altered organ geometry and overlap would directly impact dose distributions and toxicity risks (e.g., xerostomia, dysphagia). We hypothesized that posture induces systematic changes in organ geometry and relative position, measurable via deformable image registration (DIR) between upright and supine magnetic resonance images (MRI).

Materials/Methods: Ten healthy volunteers underwent 0.5T open MRI of the head-and-neck (HN) region in both upright and supine postures. MRI scans utilized a 3D gradient-echo sequence (TR/TE=17/7ms), and relevant organs at risk (OARs) were delineated by a senior physician. Rigid registration isolated global repositioning differences first, whilst subsequent DIR extracted deformations between upright and supine images. To ensure reliable motion estimation between the two postures, the accuracy of the dense deformation vector field (DVF) was evaluated by comparing the warped supine images/OARs to the corresponding upright reference using image similarity metrics and structure Dice similarity coefficients (DSC). These validated DVFs were used to quantify resulting postural related geometric changes across the HN region.

Results: DIR provided highly reliable supine-to-upright correspondence compared to rigid alignment alone. Substantial improvements in both image similarity (mean absolute error: 165±38 to 103±35; mutual information: 0.50±0.10 to 0.95±0.13; normalized cross-correlation: 0.62±0.09 to 0.92±0.03) and median DSC of all OARs (0.48±0.09 to 0.68±0.05) validated the accuracy of the dense DVFs for subsequent analysis. All subjects (N=10) exhibited pronounced variations in both OAR shape and location during supine-to-upright transition. Most notably, upright positioning maintained a larger separation between the submandibular glands (SMG) and the pharyngeal constrictor muscles (PCM). For supine, the SMGs demonstrated symmetric displacement towards PCM, with the left and right glands shifting laterally (1.5±1.3 mm and 3.2±1.7 mm; p<0.01) and posteriorly (3.8±2.0 mm and 4.4±1.5 mm; p<0.001) respectively. In subjects (N=7) with minimal neck flexion variations between the two postures, upright positioning provided a pronounced increase in displacement of the posterior oral cavity wall toward the superior PCM (1.23±0.75 mm, p=0.0049).

Conclusion: High-quality DIR of sequential upright and supine MRI challenges HN postural stability assumptions by revealing systematic gravity-induced deformations, notably in the submandibular glands and oral cavity. Characterizing these variations provides the empirical knowledge critical for defining posture-specific planning strategies to ensure high-precision upright treatment.