Use this URL to cite or link to this record in EThOS:
Title: Tissue characterisation from intravascular ultrasound using texture analysis
Author: Nailon, William H.
ISNI:       0000 0004 2728 6475
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 1997
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Intravascular ultrasound has, over the past decade, significantly changed the clinical diagnosis and therapeutic strategy of coronary and vascular disease assessment, as it not only allows visualisation of the vessel lumen, but gives a unique view of the pathophysiologic structure of the artery wall. This information is currently unavailable from the universally accepted instrument for artery assessment, angiography, which has on several occasions had its diagnostic accuracy questioned. With intravascular ultrasound, there is the potential to categorise diseased arterial tissue belonging to distinct pathological groups which can ultimately aid in the understanding of individual lesions as well as making a significant contribution to treatment choice and management of cardiac patients. The high resolution image information offered by intravascular ultrasound provides excellent cross-sectional views of coronary artery disease at the level of the disease process itself. This information can be used by the clinician to characterise atherosclerotic plaque composition and vessel wall morphology, both of which are important, in determining the clinical response to the disease condition. However, this visual diagnosis is in general highly subjective due to inter- and intra-observer error. To overcome the short comings inherent in the visual assessment of intravascular ultrasound images, texture analysis was used to assess plaque in regions of interest identified by a clinician. In the two dimensional images produced by intravascular ultrasound, texture is perceived as homogeneous visual patterns representing the surface composition being imaged. Since every tissue sub-group has its own texture, verified from histological analysis, it can be used as a means of characterising it. In this thesis, the findings of applying texture analysis techniques to 30 MHz intravascular ultrasound data, gathered in vitro, to assess its potential in quantitative coronary plaque characterisation are presented. Histo-pathological analysis was used to form a gold standard based upon clot composition, from which the results were verified. The ultimate aim of the work was to determine a reliable protocol based upon textural analysis for assessing plaque composition in vivo. Textural properties, in the form of features, were calculated for regions of interest using first-, second- and higher-order statistics. These were found to be computationally expensive and in certain instances produced duplicate, and hence redundant, information. Feature selection was used to increase the computational efficiency of the algorithm by optimising the feature set. In a further attempt to overcome the weaknesses of the aforementioned techniques, fractal texture analysis was used to obtain textural information on regions of interest. Fractals proved useful in describing the texture of these areas by a single measure. This measure, the fractal dimension, described the degree of irregularity in the surface texture. A new method is proposed for classifying arterial plaque which relies on a combination of the two powerful techniques previously mentioned, statistical and fractal texture analysis. The results presented show the ability of the texture analysis techniques used to discriminate certain tissue sub-groups. Limited success was achieved for the analysis on the atherosclerotic plaque groups studied, however, the approach adopted significantly discriminated the three types of clot composition studied: plasma; white thrombus; and red thrombus.
Supervisor: McLaughlin, Stephen. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available