Developed Methods for Angiography/IVUS Image Data Fusion
Developed methods
The final goal of this research in progress is to achieve
geometrically correct 3D volume rendering of IVUS pullback data. The 3D
trajectory of the IVUS catheter and catheter twist will be derived from
biplane angiography imaging.
The proposed method consists of the following main steps:
1. Biplane angiography system calibration and correction
of geometric distortion.
- Using a calibration phantom (ball) and a grid of markers attached to
the two image intensifiers, images are geometrically corrected to remove
all typical biplane imaging artifacts (e.g., pincushion distorsion) and
three-dimensional geometry of the biplane system is determined. Knowledge
of the three-dimensional geometry makes possible to determine correspondence
of points between the two angiographic projections.
- The eight markers are automatically identified in each of the two
calibration images. The identification of these markers creates reference
points for the geometric correction.
- For
a live JAVA demo performing automatic detection of the calibration markers,
follow this LINK.
- Original calibration images of a ball with markers (two projections,
note the 8 markers forming a square grid, note two elongated markers determining
ball diameter):
2. Detection of IVUS catheter core in the individual biplane
images.
- In each of the geometrically corrected angiographic images
(example shows pig heart #2), an operator
determines a region of interest (ROI) and the catheter is automatically
determined.
Steps 1 and 2 are demonstrated by a JAVA program.
- After the start of the JAVA program, the previously determined marker
positions and biplane system parameters are used to geometrically correct
the angiograms and calculate 3D geometry.
- An example
of the data file that contains the biplane angiography system setup
parameters and that is imported to the JAVA program is provided.
- Once the interaction is enabled (watch for the message on screen),
clicking an either angiogram will cause the display of a pair of epipolar
lines. Corresponding points can thus be determined.
- The program is prompting the user to determine two corresponding points
of the IVUS catheter and mark the ROI for catheter detection in each view.
- Automated catheter detection follows.
- All geometric parameters of the biplane system as well as all locations
of the catheter core are exported in the JAVA window and used as input
for the next steps.
- An example
of the data file exported from the JAVA program is provided.
3. 3-D trajectory of the IVUS pullback.
4. Determination of the time-location function of the imaging
tip of the IVUS catheter.
5. Calculation of the IVUS pullback data twist.
6. Determination of the absolute positioning of the IVUS pullback
data.
- Position of the catheter inside the vessel lumen is not known. Therefore
at least one landmark must be used to determine rotational positioning
of the IVUS image data with respect to the 3D course of the imaged vessel.
- In the current implementation, the landmark position is determined
interactively.
7. Mapping of IVUS data in the 3D space.
8. Border detection and quantitative analysis of IVUS pullback
data in 3-D.
- IVUS sequence segmentation has been developed by the University of
Iowa group in the past.
- After a segmentation has been performed on the IVUS frames, its results
can be added into the VRML scene. For pig heart #2, the
extracted inner and outer contours
are included into the VRML scenes with
the catheter path or with the
IVUS frames. Furthermore, a fly
through animation can be performed (Note: follow Viewpoints in
reverse order to fly from proximal to distal).
- Using IVUS segmentation together with geometrically correct 3-D geometry
facilitates accurate quantitative measurements like determination of plaque
volume, plaque distribution, plaque composition, etc.
- These important topics are focus of our ongoing research.
Questions, comments, and suggestions are welcome. Please direct any
correspondence to Milan Sonka
by email: milan-sonka@uiowa.edu
.
Last Modified: October 14, 1997
![[Go Back]](../IMAGES/prev.gif)
Go back to the main page
