3D LCSM Analysis Compared to 2D Optical Imaging

2D Optical Image

This is a 2D optical microscope image provided courtesy of the JSC curation facility to assist us in our work on track 152. Note that the image is 'squashed' in the y direction (up/down on the screen).

3D LCSM Imaging

Results here are for track #152. The full LCSM data set covering this entire track is ~60 GB, in 23 sequential volumes. Here, we show just 3 of those volumes, stitched together into one volume. Where the stitching occurs, there is a contrast change, or jump. The images are in the x-y plane of the volume (z is in/out of the screen). This part of the track (the entry area) was imaged at 130 nanometers (nm)/pixel, the rest at ~78 nm/pxl, in the x-y directions. The resolution in the z direction (in/out of the screen) is lower by ~2.5x than the resolution in the x-y direction due to limitations imposed by the wavelength of the light (488nm). A better point-spread function approximation would allow doubling or tripling of this resolution in z.
All the images are at 12-bit depth scaled to 16-bit depth because the tif format cannot hold 12-bit images. This data has been deconvolved only, from the raw LCSM output. No image 'tweaking' or enhancement has been performed. This is umpublished data collected in late April 2009, actively being worked on at the AMNH.

A low-resolution field scan of track #82 is available here: track82-field scan


Some of these results were presented at AGU (Spring 2009, Toronto) at a special session, "3-dimensional Micro-imaging Techniques of Geomaterials", and will appear in a publication based on these proceedings. The LCSM and Tomography community attendees were particularly interested in our 'stitching' techniques, which are new to LCSM imaging and data manipulation.
A paper comparing LCSM to synchrotron X-ray microtomography (XR-CMT) has been submitted to MaPS on 26-June-2009.

Reference Image:

This REFERENCE IMAGE is a projection of the entire track volume at 1/16 scale, showing entry hole volume.
The regions that are available below as tiff stacks are in the blue-green boxes.


These files are stacks of TIF format images, which is the standard way 3D images are handled.
Although TIF files are not compressed, each stack of images is stored in a compressed format.
The WinRAR program for large data files can be obtained from cnet.com (Google winrar download).
We are working on incorporating these data into the CATMAID format (a web-based system for massive LCSM datafiles), however that program uses a variant of MySQL that is not supported on our servers at AMNH, so this is a wetware/software facetime problem.

Results are given at several resolutions suitable for various computer architectures:

FULL resolution, 16-bit (1.3GB rar, 4.5 GB stack) This will not load or view in a 32-bit operating system.
WARNING: This file will crash or otherwise not work without advanced memory management techniques on small computers.
FULL resolution, 16-bit, .rar file (1.3 GB; not native to Windows)
1/2 resolution, 16-bit (0.42 GB rar, 1.2 GB stack; each voxel in the image is 16 in the raw data) 1/2 resolution, 16-bit, .rar file (412 MB; not native to Windows)
1/2 resolution, 16-bit, .zip file (466 MB; zip file limit is 2GB)
1/4 resolution, 8-bit (0.07 GB rar; each voxel in the image is 16 in the raw data) 1/2 resolution, 8-bit, .rar file (67 MB; not native to Windows)
1/2 resolution, 8-bit, .zip file (65 MB) << USE THIS FOR WINDOWS PC !!!

How to view the 3D images using ImageJ software.

1) Un-zip (or uncompress) the stack in its own directory (folder). 2) Start ImageJ.
3) File -> Import -> Image Sequence, click on first file of the sequence to load it. Change 'increment' to load only partial stacks.
4) If memory allocation is an issue, see MEMORY HELP .
5) Move image stack slider at the bottom of the image window to a slice that shows interesting features.
6) Image -> Adjust -> Brightness/Contrast: to adjust image stack and highlight features of interest.
7) button "LUT" allows changing to a myriad variety of contrast look up tables. We recommend 'Red Hot' and 'Rainbow RGB'.
7) play. Adjusting 'maximum' has highest value.
The relative area represented by each pixel at full, 1/2, and 1/4 resolutions is illustrated here: RESOLUTIONS (full=orange, 1/2=blue, 1/4=gray). That is, if 16 pixels of full resolution image are the size of the big gray square, they are all lumped together to become a single pixel at 1/4 resolution.
. Note that the suggested steps 6, and 7, above are very simplistic contrast manipulation tools. We use other techniques for analytical work.

Notes on the LCSM result:
- there is a very real crack transverse to the axis of the crack. This extend through the entire keystone in the z direction (in/out of screen).
- the track is non-circular in cross-section, and gets more so further down the track.
- there is brighter reflection (more material deposited) on the bottom edge of the track than the top.

Memory management techniques:

- load every other tif image (increment = 2 in step 3, above).
- use the "virtual stack" option when loading into ImageJ (this reads from disk, so it runs slowly).
- use an advanced volume-rendering software (e.g., Imaris, MIMICS, etc.). These are expensive.

This collection of pages describing the work of Denton Ebel and collaborators is provided by Denton Ebel as a free public service. Denton Ebel, the Dept. of Earth and Planetary Sciences, and all of their parent organizations make no guarantees as to the accuracy, timeliness or availability of this service or data provided by it.

We cannot be held responsible for any circumstances resulting from its use, inavailability, or possible inaccuracy. We reserve the right to suspend or discontinue this service or portions thereof at any time.

We also reserve the right to deny access to any individual or organization that we determine is abusing this service. Be aware that if we are forced to terminate your access, all access from your location may be terminated as a side effect, therefore preventing others at your location from using our service. Examples of abuse include automated transfers resulting in excessive data requests (because it hinders others from accessing the service) and attempting to gain access to documents and host machines not intended for public use.

Last modified June 25, 2009