Optimization of phase-contrast MRI shunt flow measurement across multiple scanners for the management of hydrocephalus

OBJECTIVE Ventricular shunts divert CSF in patients with hydrocephalus, often providing lifesaving intervention. Untreated shunt failure can lead to increased intracranial pressure and subsequent neurological injury. Phase-contrast MRI (PC-MRI) is a noninvasive technique to assess shunt function by measuring shunt flow. This work assists in clinical implementation of PC-MRI by quantifying and accounting for sources of variation in shunt flow measurements. METHODS Shunt phantoms were used to provide consistency between scanners, with flow rates spanning a physiological range. Ten MRI scanner/coil combinations with variable field strength (0.55–7T), manufacturer (Siemens and Philips), coil design (8-, 16-, 20-, and 32-channel), gradient strength (22, 33, 40, 80, and 200 mT/m), and gradient slew rate (55, 125, 160, 200, and 1000 T/m/sec) were tested. Sequence parameters were optimized for each scanner, and flow rates were extracted using three analytical methods that calculate flow: 1) mean fluid velocity, 2) maximum fluid velocity, and 3) a parabolic model of laminar flow. PC-MRI measurements were plotted against known flow rates to compare accuracy and precision across scanners. A phantom incorporating four commonly used catheter models was designed to assess variability introduced by catheter-related magnetic susceptibility artifact. RESULTS The authors show that PC-MRI can be successfully performed across a wide range of MRI scanners, including the first measurements of shunt flow at low (0.55T) and high (7T) magnetic field strength. The comparison dataset between scanners from the single-catheter phantom included 400 independent acquisitions and 1600 unique postprocessed flow values. The highest accuracy was achieved using a parabolic fit of the shunt flow profile combined with background phase correction. Advanced modeling reduced the mean absolute error ± standard deviation between measured and true flow rates from 2.0 ± 2.4 to 0.88 ± 1.1 mL/hr. These resulting deviations were at the observed range of physiological variation, even on low-field scanners. The authors performed additional scans to quantify manageable sources of variability, including shim placement, image plane angle, and human analytical error. Differences between catheter models did not significantly affect measured flow values. CONCLUSIONS PC-MRI enables accurate shunt flow measurement across diverse MRI platforms (10 scanner/coil combinations tested and all found to function), including at field strengths of 0.55T and 7T. Standardized protocols reduce error and will help support expanded use of the technique.