|
SPIE:
Sarang Lakare, Dongqing
Chen, Lihong Li, Arie Kaufman, Mark Wax, and
Zhengrong Liang
Summary:
Purpose: Virtual
colonoscopy aims to provide a safe and comfortable technique
to screen the entire colon for detection of polyps (cancerous
growth). An accurate diagnosis requires a clean, un-obstructed
view of the colon lumen during thevirtual fly-through.
One way to achieve a clean colon lumen is to perform
physical bowel cleansing prior to the image scan. This
method, although effective, is highly uncomfortable
for the patient. Electronic cleansing provides an alternative
solution by removing the residual material from the
scanned dataset before the dataset is used for the virtual
fly-through. This work aims at developing a fully automatic
solution for electronic cleansing that is fast and reliable.
Methods:
Prior to obtaining CT images for virtual colonoscopy,
the patient undergoes bowel preparation of mild laxatives
and a low residue diet.This also includes four 250 cc
doses of 2.1% w/v barium sulfate
suspension taken with meals and two doses of 60 cc of
gastroview (diatrizoate meglumine and diatrizoate sodium
solution) taken the night prior to and the morning of
the procedure. CT images are obtained after
the patient's colon is distended with CO2 (1 - 2 L)
given through a rectal tube using standard virtual colonoscopy
parameters and are reconstructed into a 3D dataset.
The dataset is automatically classified into different
regions using approximate thresholding. Special rays
which are designed to detect intersection between distinct
regions, known as segmentation rays, are then shot through
the dataset. These rays detect intersections and mark
the partial volume regions. In our case, the partial
volume region between air and tagged residual fluid
is undesirable. Hence, when a segmentation ray detects
this type of partial volume region, we remove that region
and mark it as air. The region between soft-tissue and
tagged fluid is also a partial volume region. Because
we need to remove the tagged fluid, we convert this
region to a partial volume region between air and soft-tissue.
The segmentation rays which detect a partial volume
region between soft-tissue and tagged fluid, convert
the fluid region into air region while maintaining the
partial volume between the two. Finally, the remaining
tagged fluid is converted to air. Results: We tested
our fully automatic electronic cleansing algorithm on
both volunteer and patient datasets. Figure 1(a) shows
a CT slice and Figure 1(b) shows the same slice after
applying our electronic cleansing algorithm. We also
compare our results (Figure 2c) with thresholding (Figure
2a) and vector quantization (Figure 2b). The average
time taken for electronic cleansing (for datasets with
300-400 slices of 512x512 resolution each) is 1 min
on a Linux workstation (900MHz AMD).
New or breakthrough work: In our electronic
cleansing approach, we use a new segmentation technique
based on segmentation rays. These rays are specially
designed to analyze the intensity profile as they traverse
through the dataset. When this intensity profile matches
any of the pre-defined profiles, the rays perform image
reconstruction. We use these rays to detect the itersection
between air and tagged fluid and between tagged fluid
and soft-tissue. Since partial volume is always at the
intersections, it is easily
eliminated.
Conclusion:
We presented an electronic colon cleansing technology
based on segmentation rays. The advantage of segmentation
rays over other segmentation approaches is in the detection
of partial volume regions. Segmentation rays can accurately
detect partial volume regions and remove them if needed.
Once partial volume is eliminated, removal of other
unwanted regions (e.g., tagged fluid) is straight-forward
(e.g., by using thresholding). This approach to electronic
cleansing is extremely fast as it requires minimal computation.
Keywords:
electronic cleansing, segmentation rays, virtual
colonoscopy, virtual endoscopy, massive polyp screening,
early cancer detection, volumetric segmentation, interactive
visualization.
|
