Automated Motion Tracking of Vaginal Pessaries and Pelvic Floor Structures on Dynamic MRI.

INTRODUCTION AND HYPOTHESIS: Vaginal pessaries are a cost-effective, nonsurgical treatment for pelvic organ prolapse (POP), but limited understanding of pessary biomechanics and the inability of static MRI analyses to capture continuous device-tissue interactions hinder design innovation. While dynamic magnetic resonance imaging (MRI) offers insights into pelvic floor biomechanics, conventional analyses rely on static frame comparisons and cannot capture continuous device-tissue interactions. This study aimed to apply a validated, automated motion-tracking framework to dynamic MRI for frame-by-frame analysis of pessary kinematics and evaluate correlations between pessary displacement and changes in hiatus dimensions. METHODS: In this prospective pilot study, six individuals with anterior vaginal wall-predominant POP successfully using a ring pessary with support underwent dynamic 3D pelvic MRI at rest and during maximal Valsalva. A previously validated optical flow-based tracking algorithm was employed to quantify frame-by-frame motion of defined mid-sagittal regions of interest (ROIs), including pessary rims, perineal body, and anorectal angle. These regions of interest were chosen for their biomechanical relevance in capturing the interaction between the pessary and the structures that define the urogenital and levator hiatus size, key determinants of pessary retention. Pearson correlation was used to evaluate the relationship between distal pessary displacement and changes in urogenital and levator hiatus dimensions. RESULTS: The median age was 66.5 years (range 52-76) and median pessary use was 3 years (range 2-4); all patients used a size 3 or 4 ring pessary and performed self-maintenance. Two of six reported occasional prolapse of the pessary, and all achieved successful retention. Resting MRI showed the pessary positioned posterior and inferior to the pubic bone with elevation of both anterior and posterior vaginal walls in all patients, as compared to resting MRI without the pessary in situ. Automated tracking was successful for all participants. Frame-by-frame analysis demonstrated strong correlations between distal pessary translation and enlargement of the urogenital hiatus (r = 0.96 [95% CI 0.94-0.97]) and levator hiatus (r = 0.86 [0.81-0.91]). Correlations were stronger than those observed using start-to-end frame comparisons. CONCLUSIONS: This pilot study demonstrates the feasibility of automated frame-by-frame motion to quantify pessary-pelvic floor interactions during physiologic loading. This proof-of-concept establishes a foundation for larger studies to explore patient- and device-specific determinants of pessary function and failure, with the ultimate goal of informing personalized device design and improving clinical outcomes.