Intravoxel incoherent motion MR imaging analysis for diagnosis of placenta accrete spectrum disorders: A pilot feasibility study

Background: Placenta accreta spectrum (PAS) disorders occur when the placenta adheres abnormally to the uterine myometrium and can have devastating effects on maternal health due to risks of massive postpartum hemorrhage and possible need for emergency hysterectomy. PAS can be difficult to diagnose using routine clinical imaging with ultrasound and structural MRI. Objective: To determine feasibility of using intravoxel incoherent motion (IVIM) magnetic resonance imaging (MRI) analysis in the diagnosis of the placenta accreta spectrum disorders in pregnant women. Methods: A total of 49 pregnant women were recruited including 14 with pathologically confirmed cases of PAS and 35 health controls without prior cesarean delivery and no suspected PAS by ultrasound. All women underwent diffusion-weighted imaging with an 8 b-value scanning sequence. A semi-automated method for image processing was used, creating a 3D object map, which was then fit to a biexponential signal decay curve for IVIM modeling to determine slow diffusion (D_s), fast diffusion (D_f), and perfusion fraction (P_f). Results: Our results demonstrated a high degree of model fitting (R² ≥ 0.98), with P_f significantly higher in those with PAS compared to healthy controls (0.451 ± 0.019 versus 0.341 ± 0.022, p = 0.002). By contrast, no statistical difference in the D_f (1.70 × 10⁻² ± 0.38 × 10⁻² versus 1.48 × 10⁻² ± 0.08 × 10⁻² mm²/s, p = 0.211) or D_s (1.34 × 10⁻³ ± 0.10 × 10⁻³ versus 1.45 × 10⁻³ ± 0.007 × 10⁻³ mm²/s, p = 0.215) was found between subjects with PAS and healthy controls. Conclusions: The use of MRI, and IVIM modeling in particular, may have potential in aiding in the diagnosis of PAS when other imaging modalities are equivocal. However, the widespread use of these techniques will require generation of large normative data sets, consistent sequencing protocols, and streamlined analysis techniques.