Detection of Lipid-Laden Atherosclerotic Plaque by Wavelet Analysis of Radiofrequency Intravascular Ultrasound Signals
In Vitro Validation and Preliminary In Vivo Application
Akihiro Murashige, MD,
Takafumi Hiro, MD, PhD*,
Takashi Fujii, MD, PhD,
Koji Imoto, MD,
Takashige Murata, MD,
Yusaku Fukumoto, MD and
Masunori Matsuzaki, MD, PhD
Division of Cardiovascular Medicine, Department of Medical Bioregulation, Yamaguchi University Graduate School of Medicine, Yamaguchi, Japan

View larger version (80K):
[in a new window]
|
Figure 1 Acquisition of radiofrequency intravascular ultrasound signals. There are 256 radial vectors of radiofrequency signal sampled around the intravascular ultrasound catheter.
|
|

View larger version (19K):
[in a new window]
|
Figure 2 Procedure of wavelet analysis. In this example, a wavelet is stretched twice and three times. At a time of t0, a high value of wavelet coefficient is provided (arrow), suggesting that a special wave pattern similar to the wavelet of scale 2 is included within the signal at the time. IVUS = intravascular ultrasound; RF = radiofrequency.
|
|

View larger version (58K):
[in a new window]
|
Figure 3 Representative examples of in vitro wavelet analysis of radiofrequency (RF) intravascular ultrasound (IVUS) signals from a lipid-laden plaque (A) and from a fibrous plaque without a lipid core (B). The upper panels show RF signals, the middle panels show the results of wavelet analysis, and the lower panels show the histologic specimen of the corresponding arterial cross section with Massons trichrome. In the time-scale domain color-coded mapping of wavelet analysis, an apparently different pattern of pink area from an RF signal vector of a lipid-laden plaque is observed between scale 20 and scale 30, compared with the fibrous plaque. F = fibrous area; L = lipid core.
|
|

View larger version (14K):
[in a new window]
|
Figure 4 Receiver operating curve analysis was performed with varying degrees of the wavelet coefficient in terms of the capability of the in vitro detection of lipid-laden plaque. This analysis revealed that the optimal value of this wavelet coefficient to discriminate a lipid-laden plaque was 0.6. C = the wavelet coefficient.
|
|

View larger version (61K):
[in a new window]
|
Figure 5 Representative examples of in vivo wavelet analysis of radiofrequency (RF) intravascular ultrasound (IVUS) signals from a lipid-laden plaque (A) and from a fibrous plaque without a lipid core (B). The left panels show conventional IVUS images, the middle panels show the results of wavelet analysis, and the right panels show the histologic cross sections of the corresponding directional coronary atherectomy specimen with hematoxylin-eosin and Azan stains. A similar pattern of color mapping was observed from the radiofrequency signal vector of a lipid-laden plaque, as seen in the in vitro study.
|
|
|