MULTI-PARAMETRIC SONOGRAPHIC STRATIFICATION OF UTERINE FIBROIDS: FRAMEWORK DEVELOPMENT, VALIDATION AND CLINICAL PHENOTYPIC MAPPING

Abstract

The traditional diagnostic models evaluate uterine fibroids using isolated sonographic variables rather than an integrated matrix. To address this issue, a prospective cross sectional pilot study was conducted to develop a Multi-Parametric Sonographic Stratification Framework (MPSF) that predicts clinical phenotypes which are heavy menstrual bleeding, pelvic pain and dysmenorrhea, by combining structural, topographical and hemodynamic metrics. To handle unequal variances and limited sample size (n=36), inherent to a pilot cohort, the statistical pipeline used Welch’s ANOVA, Fisher-Freeman-Halton exact test, Multiple Correspondence Analysis and Multinomial Logistic Regression (MLR). Unsupervised MCA clustered the categorical inputs, capturing 50.36% of the cumulative structural variance and identified three distinct clinical zones. In addition to that, the MLR model produced a high explanatory variance (McFadden Pseudo-R2= 0.6472). Crucially, preliminary data isolated the intra-tumoral Resistive Index as a primary differentiator for severe mechanical pressure profiles. While the pilot cohort size precludes calculating definitive odds ratio for high impedance boundaries. The strength of this initial correlation indicates that multidimensional mapping outperforms isolated clinical observations. The MPSF establishes a mathematically rigorous diagnostic blue print, demonstrating substantial predictive potential that indicates large-scale validation to permanently optimize patient trajectories and surgical triage.