1.

Introduction

Optical coherence tomography (OCT) is an emerging technology that can generate high-resolution images of tissues in situ and in real time.¹ OCT is analogous to ultrasound imaging, except that it uses the echo delay of light instead of sound to generate images. By employing broadband optical light sources, axial resolutions of 1 to $2\mathrm{\mu }\mathrm{m}$ can be achieved by OCT,² which is more than an order of magnitude above that obtainable for high-frequency ultrasound.³ Therefore, OCT has the potential of providing high-resolution, noninvasive images of architectural morphology in organs and tissues. This potential has been demonstrated in a wide range of applications such as ophthalmology,^{4, 5, 6} cardiology,^{7, 8} gastroenterology,^{9, 10, 11} dermatology,¹² dentistry,¹³ urology,¹⁴ and gynecology.¹⁵

To date, there have been no OCT investigations attempting to distinguish the uriniferous tubules and glomeruli of the kidney. In this study, we explore the feasibility and evaluate the capability of high-resolution OCT technology to image normal kidneys, as well as kidneys that had been subjected to an ischemic or a toxic insult. OCT imaging may prove especially important for the kidney because excisional biopsies can produce artifacts that are difficult to distinguish from ischemic and other insults.¹⁶ Also, excisional biopsies are invasive and damaging, and they sample only a very small region of the kidney.

Our observations indicate that OCT can provide valuable morphological information about the uriniferous tubules and renal corpuscles. Since histopathological images of the superficial uriniferous tubules can be used to predict the status of donor kidneys prior to their transplantation,¹⁷ we propose that OCT imaging might be a useful, nonexcisional, and quick procedure for evaluating the status of donor kidneys prior to their transplantation.

2.

Materials and Methods

2.1.

Optical Coherence Tomography Imaging

The OCT imaging system used in this study has been described in previous publications.¹⁸ The OCT system was a research prototype based on a commercial OCT system (Lightlab Imaging, Inc., Westford, Massachusetts) that was modified for ultrahigh-resolution performance (Fig. 1 ). Imaging was performed using a special broadband laser source consisting of a compact femtosecond ${\mathrm{Cr}}^{4+}$ :Forsterite laser combined with nonlinear spectral broadening in a dispersion-shifted fiber to generate a 180-nm bandwidth at a center wavelength of $1260\mathrm{nm}$ with 50-mW output power. To achieve ultrahigh-resolution imaging performance, the dispersion between the sample and reference arms of the interferometer needs to be carefully matched. Dispersion compensating glasses (SFL6 and LaKN22) were inserted in the reference arm to compensate for the collimating and focusing optics in the sample arm. An air-gap coupling was used in the sample arm to compensate for the air path in the reference arm from the collimator to the scanning delay mirror. An axial resolution of $4.6\mathrm{\mu }\mathrm{m}$ in air ( $\sim 3.3\mathrm{\mu }\mathrm{m}$ in tissue when scaled by the approximate tissue refractive index of 1.38) was achieved. The average optical power incident on the tissue was $\sim 10\mathrm{mW}$ . The reference arm power was adjusted to maximize the signal-to-noise ratio (SNR), and the system detection sensitivity was measured to be $102\mathrm{dB}$ . The OCT signal was divided into two orthogonal polarization channels by a polarizing beamsplitter, and the two detector outputs were digitally demodulated using a digital-signal-processing (DSP) board. A polarization diversity OCT signal was obtained from the square root of the sum of the squared signal intensities from the two polarization channels. The polarization diversity detection minimizes polarization artifacts due to the specimen or motion of the sample arm fiber during scanning. The system used a high-speed scanning delay line that acquires 3,125 axial scans per second. Imaging was performed using a fiber-based collimator combined with a microlens that produced a transverse spot size of $6\mathrm{\mu }\mathrm{m}$ full width at half maximum (FWHM). The collimator and microlens unit was raster $\left(xy\right)$ scanned by two precisely controlled stages (Physik Instrumente GmbH & Co. KG, Karlsruhe, Germany). Individual cross-sectional OCT images $\left(xz\right)$ were generated at a rate of 2 frames per second with dimensions $3\mathrm{mm}$ in length (600 pixels) and $2.5\mathrm{mm}$ in depth (1,600 pixels). Consecutive OCT images in different planes along the $y$ direction were scanned with $3\text{-}\mathrm{\mu }\mathrm{m}$ separation to generate a three-dimensional (3-D) volume. The 3-D OCT data was processed and visualized using a 3-D visualization and volumetric rendering software package (Amira, Mercury Computer Systems, Inc., Berlin, Germany). The software allowed the generation of 3-D volumetric views and en face $\left(xy\right)$ views.

Fig. 1

Schematic of the high-resolution OCT system. BS: 90/10 fiber-based beamsplitter; PC: polarization controller; AGC: air-gap coupling; DCG: dispersion compensating glass; HS-DS: high-speed delay scanner; PBS: polarization beamsplitter; D: detector; $\Sigma$ : vector summation.

3.

Observations

Using OCT imaging, we were able to obtain cross-sectional images several hundred microns into the kidney parenchyma and to observe cross sections through the uriniferous tubules [Figs. 2a and 2b ]. The uriniferous tubule lumens appeared to be low backscattering (dark region), while the parenchyma appeared to be high backscattering (bright region). An en face image can be reconstructed from consecutive cross-sectional images [Fig. 2b]. Rendered OCT images using 3-D visualization and volumetric rendering software provided 3-D reconstructions that revealed the shapes and contours of the winding proximal convoluted tubules (Fig. 3 ). For comparative purposes, a high-quality, plastic-embedded light microscopic section of this tissue is shown in Fig. 4d .

Fig. 2

(a) Two-dimensional OCT image of normal kidney. The open areas represent different cross-sectional planes through the lumens of the uriniferous tubules (arrows). (b) En face view of the tubular structure. Bar: $250\mathrm{\mu }\mathrm{m}$ .

Fig. 3

Three-dimensional reconstruction of OCT cutaway view revealing the contours as well as the winding nature of the superficial uriniferous tubules. Size: $2.0\mathrm{mm}$ $\left(L\right)\times 1.0\mathrm{mm}$ $\left(W\right)\times 0.7\mathrm{mm}$ $\left(D\right)$ ; cut plane: $150\mathrm{\mu }\mathrm{m}$ below the top surface.

Fig. 4

OCT images of a kidney protected from $1\mathrm{h}$ of ischemic insult due to prior infusion of sucrose. (a) Cross-sectional image (bar: $250\mathrm{\mu }\mathrm{m}$ ). (b) En face image (bar: $250\mathrm{\mu }\mathrm{m}$ ). (c) Three-dimensional view (size: $2.1\mathrm{mm}$ in $\mathrm{length}\times 1.0\mathrm{mm}$ in $\mathrm{width}\times 0.8\mathrm{mm}$ in height). (d) Plastic-embedded light microscopic section image (bar: $75\mathrm{\mu }\mathrm{m}$ ). G: glomerulus.

There was a significant difference in OCT images between those kidney specimens protected from ischemia by prior infusion of sucrose [Fig. 4a, 4b, 4c] and those that received no protection (i.e., no sucrose) prior to the ischemic insult [Fig. 5a, 5b, 5c ]. The latter (i.e., unprotected kidneys) revealed patches where tubules had lumens either entirely or partially filled with cytoplasmic debris. Although evident in cross-sectional images [Fig. 5a], this change was more dramatic when viewed in the en face images [Figs. 5b and 5c]. For comparative purposes, high-quality, plastic-embedded light microscopic sections of normal and ischemic kidneys are seen in Figs. 4d and 5d, respectively. OCT images of kidneys subjected to mercury toxicity revealed regions devoid of tubule lumens due to accumulated cytoplasmic debris and casts. Other tubules exhibit distended lumens due to distal tubule blockage [Fig. 6a, 6b, 6c ]. Nevertheless, it is difficult to distinguish between the distal and proximal convoluted tubules in these images. The foregoing OCT images correlated with light microscopic images of these kidneys, thereby showing debris in selected proximal convoluted tubules, distal tubule casts, and distended lumens following the mercuric chloride insult [Fig. 6d].

Fig. 5

OCT images of a kidney subjected to $1\mathrm{h}$ of ischemia followed by $5\min$ of recovery. (a) Cross-sectional image (bar: $250\mathrm{\mu }\mathrm{m}$ ). (b) En face image (bar: $250\mathrm{\mu }\mathrm{m}$ ). (c) Three-dimensional view (size: $2.0\mathrm{mm}$ in $\mathrm{length}\times 1.7\mathrm{mm}$ in $\mathrm{width}\times 0.8\mathrm{mm}$ in height). (d) Plastic-embedded light microscopic section image (bar: $75\mathrm{\mu }\mathrm{m}$ ). Significantly fewer open tubule lumens are visible due to their occlusion with cytoplasmic debris (arrows). G: glomerulus.

Fig. 6

OCT images of a kidney $48\mathrm{h}$ following infusion of mercuric chloride $(1\mathrm{mg}∕\mathrm{kg})$ . (a) Cross-sectional image (bar: $250\mathrm{\mu }\mathrm{m}$ ). (b) En face image (bar: $250\mathrm{\mu }\mathrm{m}$ ). (c) Three-dimensional view (size: $2.1\mathrm{mm}$ in $\mathrm{length}\times 1.6\mathrm{mm}$ in $\mathrm{width}\times 0.8\mathrm{mm}$ in height). (d) Plastic-embedded light microscopic section image (bar: $100\mathrm{\mu }\mathrm{m}$ ). Although first-segment proximal tubules (1) appear normal, necrotic second-segment proximal tubules (2) and cast-filled distal tubules (D) appear as smudged areas in the OCT images (arrows). G: glomerulus.

Because donor kidneys that are being preserved for possible transplantation are often surrounded by fat and are stored in plastic bags, we evaluated these factors on the ability of the OCT system to image the kidney. Neither translucent plastic wrappings [Fig. 7a ] nor a layer of superficial fatty connective tissue [Fig. 7b] impeded the imaging of several hundred microns into the kidney.

Fig. 7

Two-dimensional OCT axial images of kidneys surrounded by translucent plastic wrap (a) or covered with a layer of connective tissue (b). Note that, despite being within layers of plastic or connective tissue (arrows), the depth and quality of OCT images are unimpaired. Bars: $250\mathrm{\mu }\mathrm{m}$ .

Renal corpuscles containing glomeruli and surrounded by the capsular space of Bowman were also distinguishable in cross-sectional OCT images [Fig. 8a ]. Three-dimensional processing of these images again provided instructive views of glomerular sizes and shapes, and revealed the space of Bowman between the glomerulus and the parietal epithelium [Fig. 8b]. The latter is important because it permits determination of glomerular shrinkage and enlargement, as well as possible attachment to the parietal epithelial capsule. Again, for comparative purposes, a high-quality, plastic-embedded light microscopic section of a superficial renal corpuscle is shown in Fig. 4d.

Fig. 8

(a) Cross-sectional OCT image revealing several glomeruli (arrows) subjacent to the surface of the kidney. Bar: $250\mathrm{\mu }\mathrm{m}$ . (b) Three-dimensional reconstruction of OCT images seen in (a) reveals two glomeruli (arrows) and the surrounding capsular spaces of Bowman. Size: $1.7\mathrm{mm}$ $\left(L\right)\times 0.9\mathrm{mm}$ $\left(W\right)\times 1.0\mathrm{mm}$ $\left(D\right)$ . Bar: $100\mathrm{\mu }\mathrm{m}$ .

4.

Discussion

OCT is a rapidly emerging imaging modality that can function as a type of “optical biopsy,” thus providing cross-sectional images of tissue architectural morphology in situ and in real time.²³ In contrast to standard confocal microscopy, OCT can image with longer working distances and improved penetration depth, without the need for tissue contact. These factors are advantageous for the practical use of this technology in evaluating kidney pathology. While it has been reported that OCT can image up to depths of 1 to $2\mathrm{mm}$ (Ref. 24), the degree of penetration depends on several factors, including the light-scattering properties of the tissue being analyzed and the confocal parameter of the focusing lens. In this study, we used a relatively high transverse resolution focusing $\left(6\mathrm{\mu }\mathrm{m}\right)$ in order to have the fine tubule structures (which are about 30 to $40\mathrm{\mu }\mathrm{m}$ in diam) resolve better; therefore, the depth of field and penetration depth are limited when compared to that of OCT images using 20 to $30\mathrm{\mu }\mathrm{m}$ spot sizes. We can image up to depths of 300 to $400\mathrm{\mu }\mathrm{m}$ , which was deep enough to image several layers of the superficial uriniferous tubules and glomeruli. This was also several times deeper than in our previous tandem scanning confocal microscopy (TSCM) study, wherein only the outermost single layer of uriniferous tubules and no glomeruli could be observed.¹⁷ In addition, OCT can provide 3-D images in arbitrary planes²⁴ and can be performed using a thin flexible endoscope or catheter, ¹⁸ or even with a needle,²⁵ thus enabling ease of use and the possibility of imaging deep within a solid tissue or organ. We should note that OCT had previously been used to image thermal tissue damage to the rat kidney resulting from laser ablation.²⁶ However, this earlier study focused on the thermal insult site and did not attempt to distinguish components of the renal parenchyma (i.e., uriniferous tubules and glomeruli) as reported here.²⁶

Using an ultrahigh-resolution OCT imaging system with a compact broadband ${\mathrm{Cr}}^{4+}$ :Forsterite laser light source, we were able to image the superficial uriniferous tubules and glomeruli of intact, preserved rat kidneys. Using computer rendering software to process these images, we were able to visualize the tubules in different imaging planes and to generate composite 3-D images. The latter proved especially instructive in depicting the shapes, contours, and sizes of the uriniferous tubules and the renal corpuscles. Only scanning electron microscopy of preserved, sectioned, dehydrated, metal-coated samples can provide similar 3-D images.²⁷

The 3-D volume rendering enables the visualization of arbitrary planes and subsurface tissue microstructure. The ability to visualize en face planes at different depths enables the 3-D tubular organization to be assessed. The 3-D OCT data in this preliminary study is from ex vivo samples. This eliminates motion artifacts, since the acquisition speed is limited (2 frames per second). However, with recent advances in OCT imaging technology using spectral/Fourier domain detection, dramatic improvements ( $\sim 100$ -fold) in imaging speed are possible. ^{28, 29, 30, 31} These high imaging speeds enable 3-D OCT in vivo. In addition, novel microscanning devices such as micromechanical mirror or piezoelectric fiber scanners have been recently developed. ^{32, 33, 34} These advances promise 3-D OCT imaging of kidney structures in vivo in the future.

In order to determine the potential usefulness of OCT as a procedure to study kidney pathology, we evaluated kidney samples subjected to ischemia and samples exposed to mercury toxicity. In the ischemic study, some kidneys were protected from the ischemic insult by prior infusion of sucrose. Sucrose is a protective osmotic agent, which prevents cell swelling and rupturing that otherwise would result in damage to the proximal convoluted tubules.^{19, 20} Areas of proximal convoluted tubules with lumens occluded with debris as a result of $60\min$ of ischemia were in sharp contrast to samples protected by prior intravenous infusion of sucrose. It is important to note that the correlative light microscopic images of both normal and pathological kidney samples were obtained from kidneys that were fixed in situ by vascular perfusion procedures. As noted earlier, vascular perfusion procedures will provide images characteristic of in vivo morphology and are not obtainable in excisional biopsies normally taken to analyze histopathology.¹⁶ The latter underscores the need for a nonexcisional procedure to evaluate living kidneys.

The ischemic model of tubular necrosis reflects damage similar to that seen in kidneys being preserved prior to their transplantation. Specifically, cadaver kidneys destined for transplantation will undergo progressive tubular swelling and necrosis, thus resulting in tubular lumens filled with cytoplasmic debris.^{19, 20} Currently, transplant surgeons rely mainly on good preservation techniques and short preservation times to ensure donor kidney viability.³⁵ However, we recently reported that the degree of tubular necrosis seen in superficial proximal convoluted tubules can be used to predict the post-transplant function of kidneys.¹⁷ Therefore, the ability to monitor such damage may be of considerable value in determining the status of donor kidneys. To address the circumstances of renal preservation, we evaluated the ability of OCT to image through translucent plastic bags and layers of fatty connective tissue, both of which can surround potential donor kidneys being preserved in cold conditions prior to their transplantation. We found that several layers of translucent plastic and a thin layer of connective tissue did not interfere with OCT imaging of the superficial uriniferous tubules, at least in the rat kidney. In human kidneys, however, the connective tissue capsule is of variable thickness and, in some cases, may be too thick to permit noninvasive imaging. In those cases, we believe that OCT imaging can be more easily performed using a sterile OCT catheter/needle inserted under the renal capsule. If OCT can be used to determine post-transplant renal function by providing immediate histopathological images, then OCT can be used to eliminate poor donor kidneys, as well as make available more donor kidneys that might otherwise be discarded due to long storage times or because of unknown ischemic damage (i.e., from non-heart-beating cadavers). Other forms of vital microscopy, including tandem scanning confocal microscopy¹⁷ and near-infrared reflectance confocal microscopy,³⁶ have been recently used to image the kidney. However, their limited image penetration, combined with their traditional objectives and stages, make it difficult to evaluate human kidneys. OCT, on the other hand, not only permits deeper image penetration, but it also has the advantages of longer probe working distance and noncontact imaging. Furthermore, OCT can be performed with a small, sterile catheter or needle probe for minimally invasive in vivo imaging in the future.

In addition to studying the uriniferous tubules, we were able to successfully image intact renal corpuscles. Again, 3-D reconstructions of OCT images provided views of the renal corpuscles that are rivaled only by images obtained by scanning electron microscopy.²⁷ These images allowed us to evaluate glomerular size and shape, as well as the space between the glomerulus and the parietal epithelial capsule. Already, good correlations have been shown between glomerular size and such renal diseases as mesangial proliferative glomerulonephritis,³⁷ focal segmental glomerulosclerosis,³⁸ Type I diabetes mellitus,³⁹ and renal ischemia.⁴⁰ In addition to surface glomeruli, OCT can be used with catheters or needles to image glomeruli located deep in the cortex. A small, 27-gauge $\left(410\text{-}\mathrm{\mu }\mathrm{m}\mathrm{diam}\right)$ OCT imaging needle has already been demonstrated.²⁵ Unlike traditional renal needle biopsies that are a blind procedure, OCT imaging needles could potentially provide immediate image information about all kidney glomeruli in the vicinity of the area being probed. Exactly how useful OCT will be in providing images that might help interpret the glomerular pathology awaits future studies.