Imaging the Subcellular Structure of Human Coronary Atherosclerosis Using 1-µm Resolution

Transcription

1 Imaging the Subcellular Structure of Human Coronary Atherosclerosis Using 1-µm Resolution Optical Coherence Tomography (µoct) Linbo Liu, Joseph A. Gardecki, Seemantini K. Nadkarni, Jimmy D. Toussaint, Yukako Yagi, Brett E. Bouma, and Guillermo J. Tearney Author to whom correspondence should be addressed: G.J.T. Supplementary Information 1. Supplementary Figures Supplementary Figure 1. µoct instrumentation. ao: analog output board; bs: beam splitter; imaq: image acquisition board; cl: camera lens; lsc: line scan camera; smf: single mode fiber; pc: personal computer; clc: Camera Link cables.

2 Supplementary Figure 2. µoct images of endothelium. (a) Cross-sectional µoct image of endothelial cells in culture, demonstrating raised features that correspond to cell bodies (arrows). (b) Reformatted, en face image of the cultured endothelial cells showing cell outlines. (c) Native swine coronary artery cross-section with an irregular luminal surface and a bright underlying structure that may represents the internal elastic lamina. Scale bars, 30 µm.

3

4 Supplementary Figure 3. µoct images of calcifications. Human cadaver specimens. Microcalcifications are seen as bright densities within the µoct image of the fibrous cap (a, red arrow) and purple densities on the corresponding histology (b, red arrow and inset). (c) Large calcium nodule, demonstrating disrupted intima (red dotted box). (d) Expanded view of the region enclosed by the red box in c shows microscopic tissue strands, consistent with fibrin (yellow arrow), adjoining the unprotected calcium (white arrow) to the opposing detached intima. (e) Corresponding histology showing fibrin (f, black arrows) and a denuded calcific surface (red arrow). Scale bars for primary images, 100 µm. Scale bar for inset in b is 10 µm.

5 Supplementary Figure 4. µoct data obtained from stents. (a) µoct images of stent struts with/without polymer/drug (Taxus Liberte platform, Boston Scientific). For polymer-coated struts, polymer reflection (pr), strut reflection (sr) and multiple reflections (mr1, mr2) can be seen. (b) Box-whisker plot of the µoct signal from strut polymers with and without drug demonstrates that the µoct polymer signal intensity is greater when drug is present (n = 20 measurements).

6 2. Supplementary Methods µoct system and image acquisition. The µoct system was composed of a SD-OCT imaging console and the common-path benchtop probe (Supplementary Fig. 1). A supercontinuum light source (SC450, Fianium Ltd.) was used to illuminate a 50/50 beam splitter. Half of the source light was transmitted from the splitter to the benchtop probe. Light returning from the probe was relayed back to the spectrometer. The spectrometer was composed of a 600 lines/mm volume phase holographic transmission grating (1007-1, Wasatch Photonics Inc.), a multi-element camera lens, and a line scan camera (AVIIVA M4 CL, Atmel). We used 1,600 camera pixels to detect a total spectral range of 800 ± 200 nm with a full width at half maximum 800 ± 150 nm, so that the coherence length was 1 µm and the ranging depth was 0.5 mm in tissue. We digitized the detected signals at 10- bit resolution and transferred them to the personal computer through the camera link cable and image acquisition board (PCIe-1429, National Instruments) at 16,000 lines (spectra) per second. The maximum camera exposure time at this line rate was 60 µs. In the benchtop probe, the light beam was split into two wavefronts by a 45 º rod mirror (NT54-092, Edmund Optics Inc.) The central circular wavefront went to the reference arm and the annular wavefront went to the sample arm. The optical power on the sample was 10 mw. The reference arm was equipped with optics identical to those of the sample arm, so that dispersion was balanced. Light backreflected from the reference arm and backscattered from the sample arm were recombined through the rod mirror and guided by the single mode fiber (SM600, Thorlabs Inc.) back to the console. Transverse (x,y) scanning was performed using a pair of galvanometer scanners (Cambridge Technology) driven by an analog output board. The three-dimensional data size was voxels (x, y, z) and the corresponding image size was 2 mm 3 mm 0.5 mm (x, y, z). SDOCT image processing. All raw spectral data were firstly subtracted with a reference spectrum acquired with the sample beam blocked to obtain background-free spectra. Subsequently, background-free spectra were converted from wavelength space (λ) to the wave-vector space (k = 2π/λ) and corrected for spectrometer nonlinearity through linear interpolation. 1 The interpolated spectral data were zero padded followed by Fourier transforms to obtain complex depth-resolved backscattering profiles with an axial dimension sampling of 0.5 µm in tissue. We produced cross-sectional images by squaring the complex depth depth-resolved backscattering

7 profiles. All SDOCT cross-sectional images in this paper are displayed using a logarithmic gray scale look up table. 1. Park, B., et al. Real-time fiber-based multi-functional spectral-domain optical coherence tomography at 1.3 µm. Opt. Express 13, (2005).