3D multimodal cardiac data reconstruction using computerized tomographic angiography and x-ray angiography registration
Computerized tomographic angiography (3D data representing coronary arteries) and X-ray angiography (2D X-ray image sequences providing information about coronary arteries and their stenosis) are standard and popular assessment tools utilized for medical diagnosis of coronary artery diseases. At...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2016
|
| Online Access: | http://psasir.upm.edu.my/id/eprint/69309/1/FSKTM%202016%201%20IR.pdf http://psasir.upm.edu.my/id/eprint/69309/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Computerized tomographic angiography (3D data representing coronary
arteries) and X-ray angiography (2D X-ray image sequences providing
information about coronary arteries and their stenosis) are standard and
popular assessment tools utilized for medical diagnosis of coronary artery
diseases. At present, the results of both modalities are analyzed individually by
medical specialists, and it is difficult for them to mentally connect the details of
these two techniques. This research aims to propose a new framework for
coronary artery registration in both modalities to provide the 3D position of the
stenosis diagnosed in X-ray angiography images.
In this study, coronary arteries from two modalities are registered in order to
create a 3D reconstruction and visualization of stenosis position. Over the last
decade, some algorithms have been developed to register coronary arteries
from the above modalities. However, most of them failed to register coronary
arteries in order to estimate stenosis points from X-ray angiography on
computerized tomographic angiography modalities. In this research work, we
report on new contributions to the major parts of a 3D multimodal cardiac data
reconstruction system.
The first contribution is to propose a fast, accurate and fully automatic method
for coronary artery segmentation and labeling from X-ray angiography, based
on the proposed modified Starlet Wavelet Transform for segmentation, so that
the method can be utilized for registering with other modalities. The average
accuracy, sensitivity, specificity, and precision values of the proposed method
in LCA angiograms from the data sets are 0.96934, 0.86014, 0.98439, and
0.87797, respectively, and in RCA angiograms, the values are 0.97425,
0.89962, 0.99587, and 0.93021, respectively. Obviously, the proposed method
is robust in all performance metrics. Also, by comparing the results, it shows
that the proposed method has minimum artifacts and it also segmented the
whole parts of the coronary arteries. Furthermore, the running time for the proposed method is much better than other methods, as the whole process is
done in less than 1 s.
The second contribution is related to propose a fast and accurate method for
3D main coronary artery segmentation and labeling from CT Angiography
based on the proposed Intersection Tracking method and Improved vesselness
filter. By conducting experiments on the clinical data sets, it is proven that the
proposed method improved the ability of 3D coronary artery segmentation and
labeling from computerized tomographic angiography by increasing the overall
overlap evaluation to 95.76%. In addition, since no registration is needed prior
to applying the proposed intersection tracking method, and also segmentation
and labeling are applied at the same time in this work, hence, it is faster
compared to previous methods.
The feature-based coronary artery registration in computerized tomographic
angiography and X-ray angiography by using the provided features in both
modalities is the topic of the third contribution in this work. Tests using the
clinical data sets demonstrated that the proposed method aided the specialists
to find the location of stenosis lesion and also to determine the visual
relationship between the corresponding coronary arteries with the mean value
3D distance of 3.375mm and standard deviation of ±1.4137mm in a maximum
processing time of 0.2 s for each coronary artery registration.
The proposed research work is applicable and portable for common personal
computers, as well as with respect to the standard medical acquisition
methods. Moreover, the medical acquisition standards remain unchanged in
this work, which means that no calibration in the acquisition devices is
required, and it can be applied to most computerized tomographic angiography
and X-ray angiography devices. Another benefit of this work would be each
corresponding coronary arteries from these two modalities can be registered
individually and finally, the results can be combined and displayed as a whole
coronary arterial tree including stenosis. |
|---|