Semiautomatic analysis of phase contrast magnetic resonance imaging of cerebrospinal fluid flow through the aqueduct of Sylvius

Objective: Quantification of the cerebrospinal fluid (CSF) flow through the aqueduct of Sylvius by means of magnetic resonance imaging (MRI) is subject to interobserver variability due to the region of interest (ROI) selection. Our objective is to develop a semiautomatic measurement method to achieve reproducible quantitative analysis of CSF flow rate and stroke volume. Material and methods: MR examinations were performed using a 1.5 T scanner with a phase contrast sequence (velocity. encoding [V-enc] of 20 cm/s, FOV = 160, 3 mm slice thickness, image matrix size 256 x 256, TR = 53 ms, TE = I I ms, NSA = 2, flip angle = 15 degrees and 23 frames per cardiac cycle with peripheral retrospective pulse gating). Our method was developed using MATLAB R7. Errors introduced by background offset and possible aliased pixels were automatically detected and corrected if necessary in order to calculate the flow parameters that characterize CSF dynamics. The semiautomatic seed method reproducibility was evaluated and compared with the radius method by two observers analysing 21 healthy subjects. Results: The measurements using the semiautomatic seed method reduced the interobservers variability (intra-class correlation [ICC] = 1.0 for stroke volume and for volumetrie flow rate) versus the radius method (ICC = 0.46 for stroke volume and 0.65 for flow rate). Normal stroke volume (39.19 +/- 20.13 mu l/cyele), flow rate (3.81 +/- 2.81 ml/min), maximal mean systolic velocity (5.27 +/- 1.3 cm/s) and maximal mean diastolic velocity (4.20 +/- 1.4 cm/s) were calculated with the half moon and aliasing corrected seed method. Conclusions: Semiautomatic measurements (seed method with half moon background and aliasing correction) allow a generalization of the calculus of flow parameters with great consistency and independency of the operator.