Abstract
Computational fluid dynamics (CFD) has been widely used to study cerebrospinal fluid (CSF) flow in Chiari Malformation Type I (CM-I). However, most approaches rely on limited patient-specific detail, and it remains unclear whether such minimal input is sufficient to yield physiologically realistic flow predictions. In this study, we construct a series of MRI-based models of the craniocervical CSF space in a CM-I patient, complemented with representations of microanatomical features derived from ex vivo measurements and patient MRI data, to assess how CFD predictions are influenced by the choice of boundary conditions in the numerical integrations and the inclusion or omission of nerve roots and denticulate ligaments in the anatomical model. Our results reveal that while increasing patient-specific detail in boundary conditions improves agreement with velocity fields measured via phase-contrast MRI, key flow features—most notably anterior–posterior compartmentalization and bidirectional patterns during flow reversal—only emerge when denticulate ligaments are included in the model. In contrast, inclusion of nerve roots has a more localized effect on the velocity field and a modest impact on pressure drops. Our findings not only clarify how more detailed boundary conditions and improved anatomical fidelity affect velocity and pressure predictions, but also provide a mechanistic explanation for flow patterns commonly observed in CM-I that have remained unexplained, highlighting the critical role of denticulate ligaments. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12987-026-00780-y.