ELSEVIER @Original Contribution Ultrasound in Med. & Biot., Vol. 23, No. 2, pp. 191-203, 1997 Copyright 0 1997 World Federation for Uitrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/97 $17.00 + .OO PI1 SO301-5629(%)00199-S CHARACTERISATION OF ATHEROSCLEROTIC PLAQUE BY SPECTRAL ANALYSIS OF INTRAVASCULAR ULTRASOUND: AN IN VITRO METHODOLOGY TIMOTHY SPENCER,* M. PAULIINA RAMO,+ DONALD M. SALTER,~ TOM ANDERSON,* PETER P. KEARNEY,~ GEORGE R. SUTHERLAND,? KEITH A. A. Fox’ and W. NORMAN MCDICKEN* Departments of *Medical Physics and +Cardiology, Edinburgh University, Western General Hospital, Edinburgh, Scotland; and ‘Department of Pathology, University of Edinburgh, Edinburgh, Scotland, U.K. (Received 21 March 1996; in final form 29 March 1996) Abstract-Raw 30-MHz intravascular ultrasound data have been captured from postmortem coronary arteries (n = 4) to develop radio frequency analysis techniques for the characterisation of atherosclerotic plaque. Digitised data acquired from positions (n = 8) within diseased sections of artery were compared with the corresponding histology and radiology. Scan-converted images were used to locate regions of interest (ROI = 33) within areas of known tissue composition: loose fibrotic tissue (LFT), dense fibrotic tissue (DFT) and calcium (CA). A range of parameters was extracted from the normal&d power spectrum of each ROI within the bandwidth 17-42 MHz. Significant discrimination between LFT/DFT and between LFTKA was provided by maximum power and spectral slope (dBMHz-‘). However, the greatest discrimina- tive power was given by the y-axis (0 Hz) intercept of the spectral slope: LFT/DFT (p = 0.001); LFT/CA (JJ = 0.0001); and DFT/CA (p = 0.089). 0 1997 World Federation for Ultrasound in Medicine & Biology. Key Words: Tissue characterisation, Atherosclerosis, Computer-assisted signal processing, Interventional ultrasonography, Intravascular ultrasound, Histology, Radio frequency, Spectral analysis, Coronary disease, Plaque. INTRODUCTION Currently, intravascular ultrasound (IVUS) is unique in providing tomographic imaging of coronary arteries in viva. For this reason, the technique is being used on an increasing number of patients to clarify lesions with angiographically dubious appearances (Ge et al. 1993, 1994; Kearney et al. 1994; Zamorano et al. 1994), to select and guide coronary interventions (Goldberg et al. 1994; Mintz et al. 1994; Smucker et al. 1993) and to assess the mechanisms and efficacy of revascularisation techniques (Kovach et al. 1993; Matar et al. 1994; Potkin et al. 1992). In addition to its ability to provide information on plaque geometry and morphology, the role of IVUS in the cardiac catheterisation laboratory would be further enhanced if it could give accurate identification of Address correspondence to: Dr. Timothy Spencer, Department of Medical Physics (O.P.D.), Western General Hospital, Crewe Road, Edinburgh, EH4 2XU, Scotland, UK. E-mail: timsp@tattoo. ed.ac.uk. plaque constituents within the diseased vessel. This would make a significant contribution to both the choice of interventional technique (e.g., balloon angio- plasty, rotablation) and the indication for subsequent pharmaceutical administration (e.g., thrombolytics, lipid-lowering drugs). The use of IVUS to visualise vascular anatomy and to identify basic plaque types has been validated extensively, both in vitro and in viva. In vitro, images obtained from elastic, muscular and transitional arter- ies have been assessed visually with respect to histo- logically prepared sections taken at the corresponding sites. At best, three plaque types have been able to be discriminated, subjectively, from normal tissue: soft, mixed, calcified (Nishimura et al. 1989, 1990); fibrous, calcific and lipid (Potkin et al. 1990); stable atheroma, disrupted atheroma and thrombus (Siegel et al. 1991); fibrotic, calcific, lipid/necrotic fluid (Tobis et al. 1991); and fibrous intimal thickening, calcium deposition, lipid deposition (Di Mario et al. 1992). Similarly, vi- 191