Spectral analysis of ultrasonic backscatter from human carotid plaque

The predominant causes of stroke and transient ischemic attacks is growth of atherosclerotic plaque in the carotid artery. To aid in diagnosis of these vulnerable plaques, there exists a clinical need for a point-of-care tool for determining plaque composition in carotid arteries. Ultrasound backscatter RF data was collected from 20 human subjects scheduled for a carotid endarterectomy (CEA). The excised plaque was histologically prepared and serially sectioned to permit matching to the pre-surgery ultrasound data. Based on matched histology slides, a square region of interest (ROI) was drawn on the grayscale in vivo image corresponding to homogeneous regions. ROIs for three tissue types were chosen: 32 fibrous or fibro-fatty, 74 hemorrhagic and/or necrotic core with or without micro-calcifications, and 32 calcified. In addition to the CEA subjects, 4 normal volunteers underwent a research ultrasound exam in order to collect data on healthy adventitial sites. The RF data for each ROI was extracted and an average power spectrum was obtained. Effects of attenuation, diffraction, and system transfer functions were compensated and a linear fit was applied to the useable bandwidth. The intercept values show no overlap for the 95% CI between all four tissue types, in contrast to the mid-band fit values which were only separable on a subset of tissue types. Thus spectral analysis of ultrasonic backscatter provides additional ability to distinguish between carotid plaque tissue types as compared to intensity based measures (e.g. mid-band fit, integrated backscatter, and grayscale value).