| BackgroundFor centuries, clinicians make use of palpation to perceive the anatomical changes of human bodies to help the diagnosis. One of the major underlying principles is that tissue pathologic changes always lead to the alteration of tissue stiffness. However, the sensitivity of taction receptors in fingers is very limited. Vast minor lesions or deep pathologic changes are far beyond the scope of these receptors. It is not practical and feasible to diagnose every pathologic conditions via exploratory laparotomy. Although imaging techniques such as computerized tomography and magnetic resonance imaging have enhanced the sensitivity to detect minor or deep lesions, they can not provide the information of tissue stiffness. As mentioned above, the elastic modulus among varied human tissues can be four order of magnitude while traditional imaging techniques can only demonstrate a very small range of such difference. As a major principle of medical ultrasonography, pathologic changes will cause the changes of tissue acoustic characteristics. However, such acoustic difference would be affect by the settings of ultrasound energy, frequency and attenuation. Moreover, for diffused lesions with homogenous changes of acoustic characteristics, ultrasonography may be less sensitive due to lack of distinctions of acoustic characteristics. Thus, a technique capable of demonstrating the elastic characteristics of human tissues will be of vital significance.Elastography is born under such background. This thechnique holds it as the theoretical principles that in case of uniaxial mechanical load, every element withinthe elastic media will undergo different levels of axial strain according to their stiffness. The strain of stiffer materials will be greater than that of softer materials. From 1960s, relative elastography researches have been carried on breast, kidney and eyeballs etc.It has been well evidenced recently that plaque rupture and sequent thrombosis are major etiologies of acute cardiovascular events. The stability of atherosclerotic plaques is determined by the size of lipid core, the thickness of fibrous cap and the extent of focus inflammation. Meanwhile, these factors are also the determinants of the molecular structures of plaques and the latter plays an important role on tissue elasticity. Both static and dynamic mathematic models have elucidated that drastic displacement or distortion of the plaque, especially the fibrous cap, is the key predictor of plaque rupture. Thus, the study of strain distribution patterns of atherosclerotic plaques has the potential value on screening vulnerable plaques, elucidating the mechanism of vulnerability and predicting the location of rupture.Artery vasculatures move periodically caused by blood flow pulsation and this feature meets the basic requirement of elastography. Since 1990s, the intravascular ultrasound elastography (IVUSE) has been a new frontier and research focus. The IVUSEphy take advantage of mechanical characteristics of intravascular ultrasound (IVUS) and provide a technique to evaluate the biomechanics of vessel walls and atherosclerotic plaques. One-dimensional and two-dimensional elastography and palpogram have been developed with vessel phantoms, animal arteries and human arteries obtained from autopsy in vitro. All the work demonstrates the potential value of this promising technique in differentiating biomechanical vulnerable spots and in evaluating plaque vulnerability.However, there are no commercially available products of elastography software. In spite of the ability of demonstrating the elastic characteristics of artery walls and intima, the present studies ignore the information of adventitia, the role of which in the occurrence and progression of atherosclerosis is being put more and more emphasis on. In addition, spatial information helps the localization of plaques and thorough understanding their geometric shape. It can be deduced thatthree-dimensional IVUSE based on three-dimensional intravascular ultrasonography will contribute greatly to pan-coronary high-strain pattern detection as well as screening the vulnerable plaques more accurately. Unfortunately, rare research has been reported on this area. Virtual endoscopy (VE) is a brand new technique developed recently. The feature of convenient human-computer interaction, dynamic navigator and unique perspective possessed by VE, enable this techniques provide more accurate spatial information of local and global intima. However, no research has been reported on the combination of VE and IVUSE. Objects(1) to study the theoretical principles underlying the IVUSE and the key techniques involved in the construction of elastograms, such as image manipulation, image rendering and quantitative analyzing.(2) to develop two-dimensional IVUSE capable of imaging the elastic characteristics of vessel walls, intima and adventitia and three-dimensional IVUSE with virtual endoscopic perspective.(3) to validate the feasibility and accuracy of the developed software with vessel mimicking phantoms and animal arteries both in vitro and in vivo.Methods1. software development: based on the theoretical framework of elastogram, the whole program was developed under Microsoft? C++ Ver 6.0 compatible with Windows? XP operating system. The basic moduli include pretreatment, border detection, displacement estimation, strain calculation, color coding, three-dimensional intravascular ultrasonogram reconstruction, and at last, the IVUSE and three-dimensional IVUSE construction. (1) image pretreatment: take use of the different profiles of blood speckles and tissue borders in frequency domain, to reduce the blood noise by means of fast Fourier transformation. Then, manipulate all image frames via image averaging techniques. The grey scales of pixel in the processing image could be mapped to the outgoing images with equalizing function. The final adjustment of the parameters in the averaging procedures was made interactively according to the quality of outcomes. (2) automatic border detection: a discretedynamic contour (DDC) model was used in this study. Three steps - surface model initialization, border segmentation and border connection, were included in the DDC model algorithm. (3) displacement estimation: a block matching algorithm based on speckle tracking technique was utilized to estimate the magnitude and direction of the displacement of the tag being tracked. (4) strain calculation: radial strain was calculated with the equation s = AL / Lo under the assumptions of homogeneity, isotropy and elastomer, where Lo - the initial length of the tag from the center of the image, i.e. the radius of the tag. AL - increment of Lo after pressurization. (5) elastogram construction: constrain the speckle tracking range within the intima thickness using a smaller matching block, the strain values of tags on intima could be obtained. Thus a palpogram was constructed after the strain value was color-coded and overlapped on the lumen border of raw IVUS images. An adventitia-palpogram was constructed with the same principle. For strain values in the vessel wall, an interpolation of intima and adventitia was used and in turn, an elastogram was constructed. (6) three-dimensional intravascular ultrasonogram reconstruction: a similar algorithm as DDC model was applied to match the neighboring cross-section images. Based on such matching, the series images acquired from IVUS automatic pullback were stacked. The vertices of the DDC model were connected with three-dimensional cross-correlation coefficient algorithm. (7) three-dimensional IVUSE construction : stacking every two-dimensional elastogram along the reconstructed IVUS images resulted in three-dimensional IVUSE.2. The validation of the imaging feasibility and accuracy of the developed software: (1) study materials: five vessel-mimicking phantoms made from PVA cryogels, two abdomen aorta arteries from atherosclerotic rabbit models, two femoral arteries from healthy canine and six kidney arteries, five carotid arteries and three external iliac arteries from atherosclerotic mini-swine models. (2) IVUS examination: the phantoms and rabbit abdomen aorta arteries were examined in vitro. The IVUS images of different lumen pressure were acquired after preloading and the distances of automatic pullback were recorded. The canine external iliac arteries and all the arteries from mini-swine model were examined in vivo. The images were stored anddistances of automatic pullback were recorded. The intima and adventitia borders were detected by the software available on the IVUS mainframe. The lumen and vessel area values were recorded. For each cross-sections enrolled in this study, the intima and adventitia borders were rebuilt manually and the results of areas were also recorded. The recorded images were stored on CD-R after DICOM format transformation.3. Pathological examination: the arteries were fixed with Linger-formaldehyde, frozen-embedded and stained with HE and Masson staining techniques. The pathological sections were analyzed with VideoTesT 4.0 Morphology software and the area and perimeters of intima and adventitia were estimated.4. Statistical analysis: all analysis was performed with the SPSS software. Student t test and q test were performed on the comparison of two and three means respectively. Linear regression was also used to analyze the correlation between two variables.Results1. The custom-developed elastography software: the main working interface of the custom-developed software included 3 regions - images region, control region and diagram region. The lumen and vessel borders could be detected automatically with a surface model containing 36 vertices. Elastogram and palpogram could be acquired throughout al the study objects. The regions with different stiffness could be demonstrated via the software. The reconstructed three-dimensional IVUS images could be rotated at any angle and a virtual endoscopic perspective was realized in the reconstructed dataset. Three-dimensional IVUSE was capable of illustrating the intima strain information within the whole vessel length. The spatial distribution patterns of plaque strain were ready to be observed.2. The validation of accuracy of the software: (1) the accuracy of border detection: the intima and adventitia areas of the phantoms calculated by the software correlated well with the real values (r=0.96). The area calculation estimates of elastography, manual tracing and pathology correlated with each other very well. Compared with commercial available border detection software, the elastography demonstrated moreaccuracy on the border detection. From the analysis of the estimated perimeters, the same good correlativity as observed in area estimates comparison could be proved. (2) the accuracy of three-dimensional images reconstruction: no significant difference could be found between the length of the vessel measured from the software and that obtained from anatomic measurement. However, the estimates obtained from the software underestimated the anatomic measurement. Conclusions1. Two-dimensional elastography has been developed in this study in step with international researches. The three-dimensional IVUSE with virtual endoscopic perspectives has been firstly initiated in this study world widely.2. The elastic characteristics can be demonstrated via two-dimenstional IVUSE. Three-dimensional IVUSE with endoscopic perspective illustrates the spatial distribution patterns of varied vessel and plaque locations as well as the spatial geometric shape and localization of plaque.3. High accuracy of border detection and three-dimensional IVUS reconstruction is achieved with the both in vitro and in vivo validation . |
PDF Full Text Request