Fabrication and testing of a microchannel network phantom for MRI velocimetry

Magnetic resonance imaging (MRI) is a unique imaging modality in that both anatomy and flow can be imaged in non-opaque media. In medical diagnostics, MRI is used to map the function of brain regions by measuring the hemodynamic response, which is both distal to the site of neuronal activation and delayed with respect to the activation. An investigation of the relationship between fluid mechanics and the corresponding function magnetic resonance imaging (fMRI) activation map is examined through the design, fabrication, and testing of a unique microchannel network phantom. The phantom is fabricated out of polydimethylsiloxane (PDMS), an elastomer that is used extensively for microfluidics research. The channel size in the phantom is chosen such that similar flow characteristics are achieved to those in the capillaries of the human brain. The phantom is fabricated by bonding patterned layers of polydimethylsiloxane (PDMS) together to form a 3D flow path.; A rigorous analysis of artifacts due to shifts in k-space is presented using MR velocity images acquired from the flow through a Poiseuille flow phantom. It was seen from this analysis that the MRI-measured velocity maps are a superposition of a bilinear function onto the theoretical velocity maps. A scheme was developed to correct the phantom velocity images using points where the fluid was static. The measured flow rates are consistently higher than the flow rate imposed by the pump.; The MRI velocity maps scale with the Reynolds number of the flow. The theoretical average velocity distribution in a single channel is compared to the experimental distribution, with a consistent over-prediction of velocity near the channel edges. Slice thickness, order of image acquisition, partial volume effects, and acceleration effects are investigated, and show little effect on the measured flow rates. Examination of anatomical MR images show that the channels decreased in height during fabrication, which explains some of the over-prediction in flow rates. The remainder of the discrepancy is attributed to random error caused by a combination of a high value of the maximum encoded velocity and a low signal to noise ratio of the images.