Transcript Poster Presentation

Optimization of Phase Contrast Imaging
Luke Powers1, Alfred Luk1, Christopher Weaver1, Jonathan Fermo1
Advisors: Paul King, Ph.D.1; Frank E. Carroll, M.D. 2; Edwin Donnelly, M.D., Ph.D.2; Robert Traeger; Gary Shearer3
1
Department of Biomedical Engineering, Vanderbilt University, Nashville, TN
2 Vanderbilt University Medical Center Department of Radiology and Radiological Sciences
3 W.M. Keck Foundation Free-Electron Laser Center, Vanderbilt University
Project Overview
Design
Testing & Results
The monochromatic x-ray source was undergoing maintenance and could not be used for the testing of the
device. Imaging was conducted in the x-ray lab at the Free-Electron Laser (FEL) Center using:
Phase Contrast Radiography (PC-R) Overview
• Molly (Molybdenum) X-Ray Tube
Traditional radiography utilizes differences in x-ray absorption
to develop an image. Phase Contrast Radiography (PC-R)
utilizes differences in the refraction and diffraction of the x-ray
beam at interfaces as it transverses the object, resulting in an
edge effect enhancement seen on the image produced. The
basis of phase contrast imaging effects originates from
differences in the x-ray index of refraction, n, which is given
by
n=1–δ–iβ
where the imaginary component, β, is utilized in absorption
imaging and the real component, δ, is utilized in PC-R.
Therefore this technique is dependent
on different physical properties than
those of traditional absorption imaging
techniques. This difference gives PCR the potential to detect edges that
would appear invisible on
conventional radiography.
Produces a polychromatic beam with a focal spot of 100 μm and energy peaks at mostly 17.5 & 19.5 keV.
Can serve as a monochromatic source by isolating a particular energy with a mosaic crystal
4
• MAR 345 Digital Detector
3
2
1
100 micron resolution mode. Computer controlled, raw scan data was transmitted from detector to
computer. Raw image data converted into .TIFF files and analyzed using Image J
• Plastic Cuvette
Transparent plastic, 1.3 X 1.3 cm, filled with water. Chosen over other objects to image due to sharp edges
present within the object in order to better detect edge effects. Placed in middle of rotational stage
The experiment was composed of 7 scanning sessions. The R1 distance = 70 cm and the R2 distance ranging
from 90 – 30 cm, with 10 cm movement increments per session.
Figure A: PC-R image of a breast cancer
specimen. The tumor (white arrow heads) is
invading the chest wall (white arrows).
Figure B: The "phase-only" image further reveals
strands of tumor invasion (black arrows) not
apparent in Figure A.
PC-R provides a new method for soft-tissue imaging contrast and has significant
potential for early cancer detection in mammography and specimen radiography.
However there is much future work to be done in terms of characterizing the
parameters required for various tissues to successfully move PC-R into a clinical
setting. A system to accurately position a specimen for PC-R testing would assist in
the optimization of such parameters.
Illustration of edge
effect enhancement
1. MAR 345 stage:
Range = 100 cm
Resolution = 12.7 µm/step
• Horizontal stage:
Range = 6 cm
Resolution = 25 µm/step
• Vertical stage:
Range = 5 cm
Resolution = 25 µm/step
• Rotational stage:
Range = 400 grad
Resolution = 0.0167 grad/step
PC-R Considerations
•
•
•
•
•
•
Power Supply
Optical interrupters/cables (4)
Microcontroller chip
Microstepper drivers (4)
Interface boards (4)
iPocket ethernet controller
Plastic cuvette on rotational stage.
X-ray image at R2 = 90 cm.
Contains both absorption & phase
components.
Magnification of cuvette edge.
Contrast adjusted to show edges.
Yellow line indicates ROI.
Control System Overview
Spatial Coherence
Unlike x-ray absorption, phase contrast effects depend on the spatial coherence (d) of
the x-ray wave front. It is proportional to the x-ray wavelength (λ) and source-to-object
distance (R1), and inversely proportional to the source focal spot size (f). A larger
“spatial coherence length” yields more pronounced phase contrast effects.
(Right) Plot of x-ray intensity data on ROI drawn on
magnified cuvette edge. Edge enhancement apparent in
plot. (Above) Edge effect of phase contrast image
compared to that of an absorption only image.
Phase Extraction
• I(x)z=0 = Absorption component
z


I ( x )  I ( x ) z  0 1 
 " ( x) 
 2

Conclusion
• 1+λz/2π Φ”(x) = Phase component
•λ
= Wavelength
The PC-R images of the cuvettes tested for the presence of enhanced edge effects. The image taken at a R2
distance of 90 cm contained the characteristic edge effect profile in the intensity plot. The next step would be
to extract the phase component from the image and make adjustments for magnification effects.
• z = Object to detector distance
The goal is to divide out the absorption component and be left with a “phase only” image. This is
accomplished by taking two images: one being the absorption component only and the other containing both
components. The absorption only image is taken when the object is touching the detector (z=0). When z is
increased to get the second image, the new image is larger than the first.
Each stage’s motion is controlled through a PBASIC program, which is
stored on the BS2sx microcontroller chip. The stages are given 3
commands: home, forward, or reverse. The other input needed is how
far the stage must move. The optical interrupters are used to detect the
home position. PBASIC always makes sure the stage is moving to the
desired location from only one direction in order to prevent inaccuracies
due to hysteresis buildup in the stage’s ball screw.
The image with both absorption and phase components
must be scaled down to the size of the absorption only
image. Poor accuracy when scaling will create false
edges in the PC-R image. Therefore, phase extraction
requires the positioning of the two images to be
accurate and precise.
Magnification due to similar triangles
Device Description
Main objective:
To construct a device for computer-controlled automatic movement of an object and
detector to be used in phase contrast imaging research
Future Research:
LabVIEW VI Front Control Panel
The stages are then individually controlled through a LabVIEW Virtual
Instrument (VI), which communicates with the BS2sx microcontroller
chip via Ethernet connection. The user inputs the distance desired and
the direction, while the VI keeps track of the current location of the
stages. The VI also acts as a filter for various error-causing commands
and prevents them from being sent to the BS2sx microcontroller chip.
Cost/Market Analysis
$1,373.93
Estimated Cost of Items Available from FEL
$6,277.00
• Constructing reproducible images without false edges
Estimated Total Cost of Parts
$7,650.93
• Imaging using scattering, defines pixel resolution < 150 microns
Estimated Market Value of Device
• Computed Tomography
Based on the collected results, the error in each stage is negligible. The resolution achievable for each stage
was high enough to be used in parameter optimization for phase contrast imaging.
Social Impact
Breast cancer is the most common type of cancer among women in the United States today and is the leading
cause of cancer death among women age forty to forty-nine. The National Cancer Institute estimates that,
based on current rates, 13.2 % of women born today will be diagnosed with breast cancer at some time in their
lives.
PC-R has significant potential for the early detection of cancers in mammography, due to its ability to detect
objects invisible on conventional radiography. Development of this procedure may reduce the death rate of
breast cancer patients significantly.
Cost of Items Purchased
• Optimizing parameters for PC imaging of specific objects/tissues
This data confirms that our device can be used to successfully generate images with identifiable phase
components. An analysis of the errors in each stage revealed that the error for the horizontal and vertical
stages is 10 μm. The error in the detector stage is 0.001 inches (about 25 μm). The error in the rotational
stage is 0.0167 gradians. The optical interrupter has a differential distance of 25 μm. This would only provide
error for the MAR 345 stage, since it has a resolution of 12.7 μm/step. The optical interrupter only affects
absolute distance and not relative distance.
$30,000
Acknowledgements
This project was made possible with the resources available at the W.M. Keck Foundation FEL Center, the
Vanderbilt BME Department, and help from the following people:
Gary Shearer, Dr. Frank Carroll, Robert Traeger, Scott Degenhardt, Dr. Edwin Donnelly, Dr. Paul King