X-RAY CT

X-ray CT is a radiological imaging system first developed by Hounsfield (1973). The attenuation of two-dimensional fan beams of X-rays that penetrate a sample is measured by an array of detectors. These X-ray projection data from various directions are obtained by stacking contiguous two-dimensional images (Fig. F4). The degree of X-ray attenuation depends on the density and atomic number of atoms composing the samples. Higher density and higher atomic numbers result in higher attenuation of X-rays.

An X-ray CT scanner (W2000) at the Geological Survey of Japan (Hitachi Medical Co., Tokyo, Japan) was used in the present study (Fig. F4A). A target (Mo-W alloy) in an X-ray tube produces X-rays by collision with electrons accelerated at 120 kV with a 150-mA current. X-rays that have penetrated the sample are measured by 768 detectors. The in-plane resolution (voxel size) of the X-ray CT is 0.31 mm. The X-ray CT reports the data in the form of the so-called CT number (Nct) defined as

Nct = (N - Nw)/Nw x 1000,

where N is the linear X-ray absorption coefficient of the sample and Nw is the linear absorption coefficient of the standard reference, pure water. The Nct of water appears with a value of 0 and that of a nonattenuating material, such as air, appears with a value of -1000. The Nct is a function of the density, state, and chemical composition of the material in any voxel.

The typical imaging parameters used are as follows:

Reconstruction filter = 6.
X-ray tube current = 120 mA (cube samples) or 100 mA (whole rounds).
Voltage = 150 kV (cube samples) or 120 kV (whole rounds).
Slice thickness = 1.0 mm (thickness of the X-ray fan beam).
Voxel size = 0.313 x 0.313 x 1.0 mm3.
Scan time = 4.0 s.
Imaging diameter = 160 mm.
Pixel matrix size = 512 x 512.

In addition, T.B.C. compensation (a compensation filter) is applied to remove the high-attenuation artifact that generally appears in CT images from the outer rim of a sample.

The output CT images are digitized as TIFF formatted, 16-bit, gray-scale image files. The original data are converted to 8-bit color images because the operating systems of popular personal computers such as Macintosh and PC cannot generally support 16-bit images. The converted 256-color values (8 bit) are linearly interpolated over a range of Ncts from 0 to 4000. Each color grade in an X-ray CT image consequently represents a 15.625-Nct range (e.g., Fig. F5).

A series of eight slice images was obtained from each cube sample. The space between each slice is 3 mm (Fig. F4B), and each slice has a 1-mm thickness (Fig. F4C). For the whole-round cores, there is no space between each slice and each slice has a 1-mm thickness (Fig. F6A). The image from the middle slice (Fig. F4B) on both cubes and whole-round samples was used for the calculation of the average Nct. Because one slice image has 60 pixels x 60 pixels, the average is from 3600 Nct. The results are presented in Tables T1, T2, and T3.

Processed CT images from whole-round core samples of representative lithologies are shown in Fig. F5. Lamination can be observed in the radiolarian porcellanite and chert and nannofossil chalk samples. The high attenuation of X-rays (red) displayed in Figure F5G appears to be associated with a relatively altered portion of the sample. Although altered basalt has a lower density than fresh basalt, the relatively high Nct indicates the presence of atoms of a high atomic number. We speculate that this is due to absorption of heavy metals (e.g., Zn) onto the surface of clay minerals.

Three-dimensional X-ray attenuation distribution can be reconstructed using a series of slice images (Fig. F4B). An example of such a reconstructed image is shown in Figure F6A, and a cross-sectional image perpendicular to coring direction (z-vertical) is shown in Figure F6B. X-ray CT images of selected cube samples are shown in Fig. F7.

The relationships between the average Nct from basalt cube samples and shipboard wet bulk density values (Shipboard Scientific Party, 2000a; see Table T11, p. 204, in the "Site 801" chapter) are shown in Figure F8. Samples with higher density generally correlate with higher Nct samples (R = 0.89), as expected. There are some exceptions such as basalt Sample 185-801C-33R-1, 11-13 cm, which has a relatively low density (2.49 g/cm3) and a relatively high Nct (3875). Again, this indicates an association of atoms of high atomic number with alteration minerals.

NEXT