2.1.

OCT Imaging

The use of human hearts for research was approved by the Institutional Review Boards at both Washington University and the University of Minnesota.

Explanted failing human hearts $(n=12)$ were obtained at the time of cardiac transplantation (Barnes-Jewish Hospital, St. Louis, Missouri) and immediately perfused with cardioplegic solution $(4°\mathrm{C})$ . Tissue preparations from the right atria (RA), left ventricle (LV), infarcted right ventricle (RV), and atrioventricular junction (AVJ) were isolated from the rest of the heart. An additional sinoatrial nodal (SAN) specimen was obtained through the Upper Midwest Organ Procurement Organization (Life Source, St. Paul, Minnesota) and the University of Minnesota and shipped frozen to St. Louis. The tissues were perfused and fixed with 3.7% formaldehyde overnight and then transferred to 20% sucrose for $2\text{days}$ . The tissues were photographed and clearly marked with pins to precisely correlate the OCT images with histology. OCT was performed $3\text{to}4\text{days}$ later. Table 1 shows the clinical data regarding each specimen.

Table 1

Clinical data. Relevant clinical information regarding each specimen. The mean age of the patients was 50±10years .

OCT, based on the principle of low-coherence interferometry, measures the backscatter of light. The coherence of the backscattered light from the sample arm (or tissue) and reference arm result in the generation of multiple A-scans (axial scans), which are then converted to pixel intensity and combined to form a tomogram (B-scan), or slice of tissue²⁶ (Fig. 1 ). The basics of our OCT system have been described previously.^{16, 27} We imaged all cardiac tissues in $1\times 1$ , $2\times 2$ , or $4.5\times 4.5\mathrm{mm}$ overlapping segments from the epicardial, endocardial, or transmural surfaces, with final axial and lateral resolutions of $10\mu \mathrm{m}$ (in air).

Fig. 1

Principles of OCT. (a) The setup for our benchtop OCT system includes a light source feeding directly into a $50∕50$ coupler splitting light between the reference and sample arms. The reflected light is then sensed by the detector, filtered, and amplified before advancing to the frame grabber. (b) An example A-scan describing the intensity of the reflected light at various depths in the tissue. (c) Multiple A-scans create the final image slice, or 2-D tomogram (B-scan).

2.2.

Histology

After imaging, the tissue specimens were embedded and frozen in Tissue-Tek OCT compound (Electron Microscopy Sciences, Hatfield, Pennsylvania) and stored at $-80°\mathrm{C}$ . The tissues were then cryosectioned at $16\mu \mathrm{m}$ , mounted on Superfrost Plus glass slides (Fisher Scientific). Sections were then stained using Masson Trichrome (IMEB, San Marcos, California) and visually compared with the corresponding OCT images.

2.3.

Computational Analysis and Statistics

Images were processed using MATLAB (Mathworks) and Volocity (Improvision). Initially, each 2-D slice was acquired six times, and the six frames were then averaged together to reduce background noise. Subsequently, during post acquisition data analysis, median filtering was performed with a $3\times 3$ smoothing kernel to reduce speckle noise.²⁸ Three-dimensional (3-D) tissue reconstructions were created using both Volocity and custom-written MATLAB programs. Additional analysis programs were developed in MATLAB for optical tissue slicing, to determine fiber orientation in relation to the tissue surface,¹⁶ and for determination of the depth of OCT penetration. Depth penetration of each $4.5\times 4.5\mathrm{mm}$ field of view acquired across all tissue preparations was assessed via a global thresholding approach.²⁶ Quantitative data are expressed as $\text{mean}\pm \text{standard}$ deviation. A Student’s $t$ -test was used to determine statistical significance. A $p$ value of $<0.05$ was considered statistically significant.

3.1.

OCT of the Atria and Ventricles

Atrial $(n=6)$ and ventricular $(n=8)$ preparations were imaged from both the endocardial and epicardial surfaces. Atrial depth penetration $(0.94\pm 0.14\mathrm{mm})$ significantly differed $(p=1.74\times {10}^{-8})$ from that of the ventricles $(1.29\pm 0.46\mathrm{mm})$ .

In atrial tissues, we created 3-D reconstructions of the trabeculations and distinguished fiber structures up to $1\mathrm{mm}$ below the endocardial surface (Fig. 2 ). In the ventricles, we were able to distinguish changes in fiber orientation in areas located within $4.5\mathrm{mm}$ of each other and $0.4\mathrm{mm}$ below the epicardial tissue surface [Figs. 3a, 3b, 3c ]. Although our current means of imaging does not achieve the depth required to reconstruct fiber architecture throughout the entire thickness of the human myocardium, we do have evidence that this can be accomplished in animal models, in which our current OCT system is able to penetrate through a larger percentage of the myocardium.¹⁶

Fig. 2

OCT of the atria. (a) Atrial trabeculations of the RA appendage as viewed from the endocardium. (b) 3-D OCT rendering of the trabeculations. (c) Fiber architecture located $0.44\mathrm{mm}$ below the surface.

Fig. 3

OCT of the ventricles. (a) Photograph of an LV preparation as viewed from the epicardium. (b) Fiber architecture in the area indicated by the red box, $0.4\mathrm{mm}$ below the epicardial surface. (c) Fiber architecture $0.4\mathrm{mm}$ below the epicardial surface located $4.5\mathrm{mm}$ to the right of (b) as indicated by the blue box. The average angle of the fibers shifted by $30\mathrm{deg}$ . (d) Histologic cross section of the LV, affording a view of a coronary vein. (e) 2-D OCT image corresponding to (d). (Color online only.)

Epicardial OCT can also afford us a view of the coronary vessels. Figure 3d shows a histological cross section through an intact LV coronary vein, with the corresponding OCT cross section shown in Fig. 3e. Figure 4 shows the resultant 2-D and 3-D OCT images from a coronary artery, isolated from the RV epicardial surface, with a main branch diameter of $2\mathrm{mm}$ . In particular, note that in Fig. 4c, we are able to distinguish arterial branches less than $1\mathrm{mm}$ in diameter, as indicated by the arrowhead.

Fig. 4

OCT of an isolated coronary artery. (a) Photograph of an isolated coronary artery. (b) 2-D cross section of the artery. (c) 3-D OCT rendering of the artery. The red arrowhead indicates an arterial branch less than $1\mathrm{mm}$ in diameter. (Color online only.)

In addition, we imaged RV infarcts $(n=3)$ epicardially, endocardially, and transmurally. Although we were able to detect significant $(p=3.16\times {10}^{-7})$ changes in depth penetration as compared with adjacent noninfarcted ventricles, as measured from the epicardium and endocardium, we could not visually identify the fibroses from these surfaces with our OCT system. In contrast, when we imaged from the dissected transmural surface, as shown in Fig. 5, the fibrotic tissue of the infarct is clearly visible using OCT and correlates well with histology.

Fig. 5

Transmural OCT of the infarcted RV. (a) Histology of an epicardial RV infarct and its underlying tissue. (b) Corresponding OCT cross section of the infarct. The dotted line represents the approximate border between the infarct and viable myocardium.

3.2.

OCT of the SAN

OCT of the SAN was initially attempted from the endocardial surface, the primary point of access during intracardiac procedures. The SAN specimen was dissected to include the RA appendage, crista terminalis (CT), and approximately $2\mathrm{cm}$ of atrium beyond the CT [Fig. 6a ]. The specimen was then imaged in overlapping $4.5\times 4.5\mathrm{mm}$ segments, beginning along the CT and extending to the border of dissection. Figure 6b shows 3-D reconstructions of the tissue re-created from scans along the CT/RA border, as indicated by the dashed line in Fig. 6a. Compared with the 2-D tomogram in Fig. 6c, Fig. 6d exhibits an increased number of tissue layers as the transitional area closer to the SAN is approached. As the OCT field of view begins to incorporate the tissue layers above the SAN [Figs. 6e and 6f], it is readily apparent that the primary pacemaker of the heart, defined as a subepicardial structure,²⁹ is located too deep within the myocardium (at $1.75\mathrm{mm}$ ) for imaging from the endocardial surface. The average depth penetration for this specimen from the endocardial surface was $1.06\pm 0.21\mathrm{mm}$ . Imaging from the epicardial surface was hindered by a layer of fat along the sulcus terminalis. Although fat intrinsically has a lower backscattered reflectance than normal cardiac tissue,³⁰ the loss in depth penetration was such that we could not visualize the SAN. In addition, the thin layer of epicardial fibrous tissue indicated by the arrow in Fig. 7a may contribute to the loss of depth when the SAN is imaged from the epicardium.

Fig. 6

Endocardial OCT of the SAN. (a) Dissected SAN preparation. (b) 3-D reconstruction of the tissue area as indicated by the red box. (c) and (d) 2-D tomograms from (b) as indicated by the green and purple lines. (e) Histology with the SAN indicated by the solid line in the lower-right corner. (f) OCT cross section corresponding to (e) showing a strong relationship between the transitional area as defined histologically and via OCT; however, the SAN is unable to be imaged endocardially using this method. Abbreviations—RAA: RA appendage; SVC: superior vena cava; IVC: inferior vena cava; CT: crista terminalis; IAS: interatrial septum; SAN: sinoatrial node. (Color online only.)

Fig. 7

Transmural OCT of the SAN. (a) Histologic section of the SAN and surrounding tissues. (b) Transmural OCT of the same area. The asterisk (*) identifies the SAN, and the black arrow in (a) points to a thin layer of epicardial fibrous tissue that may have prevented OCT imaging of the SAN from the epicardium. Abbreviations—CT: crista terminalis; SAN: sinoatrial node.

We also imaged the SAN specimen transmurally (i.e., in dissected tissue), which allowed us to consistently locate the SAN throughout its entire length and its proximity to the SAN artery. In this way, we were also able to compare the OCT image to a sister histology section. Due to the prominence of fibroblasts and connective tissue, which serve to isolate the SAN myocytes from the surrounding atrial tissue,³¹ the SAN is readily identifiable as a unique structure with transmural OCT (Fig. 7), without the need for histological staining.

3.3.

OCT of the AVJ

OCT of the AVJ was achieved endocardially in five of nine preparations (56%). In the five preparations whose AVJ was imaged, those junctional elements, including the His bundle and AVN, were located within $2\mathrm{mm}$ of the endocardial surface. As the average depth penetration in this region across all specimens was $0.96\pm 0.25\mathrm{mm}$ , those elements located deeper than $2\mathrm{mm}$ , as confirmed via histology, were unable to be imaged in this manner.

All AVJ specimens were dissected for imaging of the entire Triangle of Koch, including a $1\text{to}2\text{-}\mathrm{cm}$ border incorporating additional atrial and ventricular tissues [Fig. 8a ]. The specimens were then optically mapped in overlapping $4.5\times 4.5\mathrm{mm}$ segments, with higher resolution images acquired in regions of interest, particularly the His bundle and atrioventricular node (AVN). In Fig. 8, 3-D reconstructions revealed discrete structures located approximately $1\mathrm{mm}$ below the tissue surface [Fig. 8b], corresponding to the locations of the His bundle and AVN as identified with traditional histology. In addition, 2-D tomograms showed this discrete structure encased in fibrous tissue, which manifests as a black envelope surrounding the brighter tissue of the bundle itself [Figs. 8c and 8d]. As our field of view progresses proximal to the His bundle, approaching the rightward extension (RE), also known as the inferior nodal extension (INE), we have a clear view of the AVN/RE [Figs. 8e and 8f]. The clarity is such that we can resolve the blood vessels (ranging from $320\text{to}750\mu \mathrm{m}$ in diameter) located within the AVN itself. The blood vessels are identified due to their consistent locations as identified with both histology and OCT.

Fig. 8

Endocardial OCT of the AVJ. (a) Dissected AVJ preparation. (b) 3-D reconstruction of the tissue area as indicated by the red box. (c) and (d) 2-D tomograms from (b) as indicated by the purple and yellow lines. (e) Histology of the AVN as indicated by the green line on (a). (f) Corresponding 2-D OCT image of (e) showing a strong relationship between the AV node as defined histologically and via OCT. Red arrowheads indicate blood vessels within the node, visible with both histology and OCT. Abbreviations—IAS: interatrial septum; TT: tendon of Todaro; CS: coronary sinus; LE: leftward extension; CN: compact node; LNB: lower nodal bundle; CFB: central fibrous body; RE: rightward extension; INE: inferior nodal extension; VS: ventricular septum; TV: tricuspid valve; AVN: atrioventricular node. (Color online only.)