MRI
The first sample, a 4.0-mm glass bead pack, was used to test the
Mathematica notebook workflow described in the materials and methods
section. The high connectivity and simplicity of this porous medium
makes it ideal for testing the Mathematica notebook for accuracy in
porosity calculations and PCA. The larger pore space between individual
beads allows for water molecules to yield T2 values
longer than those they would have in a rock sample due to less
confinement. These conditions prompted the use of the RARE-Inv-Rec pulse
sequence to reproduce the structural features of the simple bead sample.
Figure 4 shows the results of image collection with RARE-Inv-Rec (Figure
4A) and the application of the Mathematica notebook to the glass bead
sample (workflow steps 1-6, Figure 4B-4E). The set of MRI images was
uploaded into the notebook and some slices in the set were selected. The
selected 2D slices were used to construct a 3D rendering of the sample.
The 3D rendering was then taken through a series of cropping stages
including Cartesian and cylindrical cropping (Figure 4B and Figure 4C).
Once the 3D rendering was finalized through the cropping stages, a set
of filters was applied. These filters include color convert from RBG to
greyscale, sharpen, and a Gaussian filter (Figure 4D). As shown by the
color scale to the right of the image in Figure 4D, the higher the
proton density in the sample, the more yellow the color in the figure
will be. This also translates to pixel values between 0 and 1, which are
later used in the binarization process. The bright yellow pixels n
Figure 4D represent the water that fills the pore space around the
beads, whereas the beads appear as empty spaces. Lower proton density
can also be observed near the surface of the beads. The high
connectivity of the fluid expected from this simple is also clearly
detected. Once the filters are applied to the cropped 3D image, the 3D
image is then transformed to its binary form (Figure 4E), which allows
for further analysis to be applied to the image via PCA. The binary
transformation was applied using threshold values between 0.05 and 1.0
in 0.05 increments. This process permits a comprehensive analysis of the
sample.