Necrotic tissue distribution analysis: Preliminary investigation for reducing necrosis overestimation in intravascular ultrasound virtual histology images

• Published in: Computing in Cardiology 2014 pp. 321 - 324
• Main Authors: Sales, Fernando J. R., Falcao, Breno A. A., Falcao, Joao L. A. A., Furuie, Sergio S., Lemos, Pedro A.
• Format: Conference Proceeding
• Language: English
• Published: CCAL 01.09.2014
• Subjects: Abstracts; Classification algorithms; Histograms; Programmable logic arrays; Ultrasonic imaging
• ISBN: 9781479943463; 1479943460
• ISSN: 0276-6574; 2325-8853

Background: The intravascular ultrasound (IVUS) Virtual Histology (VH) classifies the atherosclerotic plaque components, according to four classes - fibrous (FT), fibrofatty (FF), dense calcium (DC), and necrotic tissue (NC). Past works have proven a possible overestimation of necrotic content in VH images, particularly when NC is confluent to DC. Additionally, NC has been used as a severity marker for atherosclerotic plaque. Aiming to observe a possible correction rule for overestimation problem, an experiment was designed. Eight patients have been submitted to stent implantation (PCI). VH examinations have been performed before (PRE) and after (POS) PCI, totalizing more than 300 frames. NC content was divided into two classes: RW and RP, which corresponds, respectively, to NC confluent to DC and not confluent. Grayscale IVUS images were processed to extract normalized intensity histograms according to NC classification. Differences between intensity distributions RW and RP necrotic tissue were found, which may be useful to be considered to minimize overestimation of necrosis in VH images. These preliminary results encourage further investigation with larger sample sizes for proposing correction-imaging methods.