In vivo non-invasive identification of cell composition of intimal lesions. A combined ultrasound- and immunocytochemistry-based approach

Purpose: We investigated whether differences in cellular composition of the shoulder region of carotid plaque, a cell-rich, debris-free area, can be revealed with computer-driven analysis of ultrasound scans. Methods: In 26 patients referred for carotid endarterectomy, the shoulder region of plaque eligible for surgical removal was identified with ultrasound scanning. Digital images were obtained and evaluated with a specially developed computer-driven system (Medical Image Processing [MIP]). The gray level distribution of the region of interest (ROI), along with some statistical parameters exploring the spatial distribution of pixels, such as entropy and second angular moment, were analyzed. In the specimen retrieved at surgery, the area corresponding to the ROI was selected. Cryosections were tested at immunocytochemistry with monoclonal antibodies specific to smooth muscle cells (SMCs), macrophages), and lymphocytes. Computerized image analysis was performed to quantify each cellular component of the lesion. Results: Mean gray levels were related positively to the content of SMCs (r 0.576, P .002) and negatively to the content of macrophages (r 0.555, P .003). Lymphocytes did not show any correlation. Prevalence of SMCs, expressed as the ratio SMC/(SMC macrophages), was related positively with entropy (r 0.517, P .007) and negatively with the second angular moment (r0.422, P .032). The quartiles of gray level were useful for detecting significant differences in terms of cellular composition. Conclusions: Some cellular features of the shoulder region of plaque are associated with specific videodensitometric patterns evaluated with MIP. This approach enables in vivo noninvasive prediction and monitoring of cell composition of the shoulder region, and could be extended to study of the thickened intima