تشخيص پلاك در تصاوير صوت درون عروقي

چكيده انگليسي :

100 Plaque detection in IntraVascular UltraSound IVUS images Behnam Nasr Behnam nasr@ec iut ac ir Date of Submission 2012 03 07 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language FarsiSupervisors Rasoul Amirfattahi Fattahi@cc iut ac ir Mohammadreza Yazdchi yazdchi@eng ui ac ir AbstractHeart attack and stroke are the major causes of human death in the world and industrial countries because ofcardiovascular diseases This kind of death is more frequent than all forms of cancer in my country One ofthe most commonly used diagnostic coronary artery disease tools is intravascular ultrasound IVUS technique IVUS imaging is more accurate than coronary angiography and it is becoming a well knownimaging technique for direct visualization of coronary arteries However visual evaluation andcharacterization of plaque require integration of complex information and suffer from substantial variabilitydepending on the observer This fact explains the difficulties of manual segmentation prone to highsubjectivity in final results Therefor automatic segmentation will save time and provides objective vesselmeasurements Moreover IVUS image segmentation is the main step of measurement analysis So accuracyof IVUS segmentation i e segmentation of lumen intima plaque and wall borders is the most importantfactor in quantitative analysis and it is a prerequisite for quantitative analysis Segmentation of arterial wallboundaries from intravascular images is an important problem for many applications in the study of plaquecharacteristics mechanical properties of the arterial wall its plaque shape reconstruction and itsmeasurements such as lumen size lumen radius and wall radius Detection of the vessel wall in IVUSimages has been approached in several recent works In Chun Yang and et al work a combining IVUSimaging computational modeling angiography and technical testing is proposed to perform mechanicalanalysis for human coronary atherosclerotic plaques for potential more accurate plaque vulnerabilityassessment Gozde Unal et al present a shape driven approach to segmentation of the arterial wall fromIVUS images in the rectangular domain M H Roy Cardinal and Meunier present a segmentation methodbased on the fast marching method which uses gray level probability density functions PDFs of the vesselwall structures Although the previous methods in IVUS image segmentation that explained are usuallyhampered by noise and artifacts on the IVUS images and in the dark shadow and soft plaque region havesome inaccuracy but our method is robust against this problem In this thesis we use a novel method basedon texture method which is a texture feature extractor based on gray level co occurrence matrices patterns Since it is a fundamental property of texture it can segment IVUS images in a high level accuracy Wepresent a texture approach for detection of plaque wall from intravascular ultrasound IVUS images In aproperly built shape of plaque using texture feature extraction we constrain the internal lumen and externallumen borders so intimae zone can reconstructed We utilized data coming from IVUS probes at both 20 and40 MHz from which our database is provided This database is input for a post processing stage which cancalculate measurements such as lumen and intimae size plaque shape and hard plaque size The proposedmethod contains some stages such as polar transform gray level co occurrence matrix GLCM fuzzy C means FCM clustering morphological processing and curve fitting function In this thesis we detect softand hard plaque separately Keywords Intravascular Ultrasound Plaque Detection Feature extraction GLCM

نصر، بهنام

تشخيص پلاك در تصاوير صوت درون عروقي

تصوير برداري عروق كرونر , IVus , تحليل بافت , استخراج ويژگي , ماتريس هم رخداد , ناحيه بندي تصوير و تشخيص پلاك