Point-of-care microscope for real-time acquisition of in vivo volumetric histological images

Abstract

The microscope routes excitation light through a first set of optical components so that the excitation light is projected onto the sample forming a sheet of excitation light at an oblique angle. The position of the sheet varies depending on the orientation of the scanning element. The first set of optical components routes the detection light back to the scanning element, which routes the detection light to a second set of optical components. The second set of optical components forms an intermediate image plane that is imaged onto the detector. In some embodiments, a folding mirror is disposed between the first set of optical components and the second set of optical components. In some embodiments, an optically transparent spacer covers the first objective lens and is configured to press against the tissue being imaged. The spacer sets a working distance for the first objective lens to capture a particular range of depths within the tissue.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of U.S. Provisional Patent Application No. 63 / 125,817, filed December 15, 2020, which is incorporated by reference herein in its entirety.

[0002] STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT This invention was made with government support under grants NS108213, NS09429, NS104649, and CA236554 awarded by the National Institutes of Health, and grants 1644869, 0801530, and 0954796 awarded by the National Science Foundation. The government has certain rights in this invention. [Background technology]

[0003] Since the 19th century, tissue microstructure and cellular organization have been used to distinguish between tissue types, their normal state, and pathological conditions. However, despite major medical advances, histological examination of tissue still requires that the tissue be excised, processed, sliced, stained, and imaged. Biopsy specimens remove valuable tissue, may miss abnormalities, and can take from 20 minutes to several days to process, lengthening procedures, increasing costs, and hindering closed-loop decision-making.

[0004] More than 6 million biopsy procedures are performed in the United States each year. However, the standard practice of excising, fixing, and staining tissue for histopathological evaluation is costly and time-consuming, delaying treatment while being confounded by sampling errors. While intraoperative rapid pathology can provide results within 20 minutes, tissues suffer from freezing artifacts, poor sectioning, swollen cell morphology, and poor staining, especially in fatty tissues such as the brain, which are particularly difficult to freeze and section. The use of physical 2D histology slides also slows the evaluation workflow, as pathologists must track features through multiple sections to gain a better understanding of 3D tissue morphology. Physical slides must be manually viewed through a microscope or digitized before viewing, requiring additional time and resources.

[0005] Most importantly, however, both frozen histology and standard histology require the physical removal of living tissue. For precise tissues such as the eye, heart, or brain, conservative biopsy can result in undersampling and either a misdiagnosis or an incomplete surgical resection. The destructive nature of biopsy also means that it is rarely used for general surgical guidance, such as tissue type identification, or to examine large areas of the body. Ex vivo tissue also rapidly loses characteristics such as perfusion level and metabolic state, which can provide valuable biomarkers of the tissue's health or disease state.

[0006] Confocal endoscopy utilizes confocal scanning through fiber optic conduits to generate 2D images of tissue in situ, which can be achieved through the endoscopic passageway. However, while current commercially available embodiments of confocal endoscopy rely on the systemic injection of bright fluorescent dyes, such as fluorescein, to provide contrast, its ability to capture 2D images over a small field of view has proven difficult to interpret reliably. Specificity can be improved with fluorescent markers that can selectively highlight disease, but regulatory approval of such chemicals has proven prohibitively costly and complex in most cases. Two-photon fluorescence, second-harmonic generation, fluorescence lifetime, and stimulated Raman spectroscopy have been demonstrated for both in vivo imaging and bedside imaging of fresh tissue, revealing impressive endogenous or "label-free" contrast. However, all of these techniques have limited acquisition speeds, which cannot tolerate in vivo motion and prevent real-time, large-area, or 3D imaging, while their reliance on high-cost and / or high-power pulsed laser sources has limited their use in in vivo clinical imaging, except in some exceptional cases. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] U.S. Patent No. 10,061,111 [Patent Document 2] U.S. Patent Application Publication No. 2019 / 0317312 Summary of the Invention [Means for solving the problem]

[0008] One aspect of the present invention is directed to a first imaging device comprising first and second sets of optical components, a scanning element, a folding mirror, a photodetector array, and a third objective lens. The first set of optical components has a proximal end and a distal end, and includes a first objective lens disposed at the distal end of the first set of optical components. The first objective lens has a magnification between 10x and 70x and a numerical aperture between 0.5 and 1.1. The second set of optical components has a proximal end and a distal end, and includes a second objective lens disposed at the proximal end of the second set of optical components. The scanning element is disposed proximal to the proximal end of the first set of optical components and distal to the distal end of the second set of optical components. The scanning element is positioned to route excitation light in a proximal-to-distal direction through the first set of optical components such that the excitation light is projected onto a sample positioned distal to the distal end of the first set of optical components. The excitation light projected onto the sample forms a sheet of excitation light at an oblique angle, the position of which varies depending on the orientation of the scanning element. The first set of optical components routes the detection light from the sample back to the scanning element in a distal-to-proximal direction. The scanning element routes the detection light so that it passes through the second set of optical components in a distal-to-proximal direction, with the second set of optical components forming an intermediate image plane at a location proximal to the proximal end of the second set of optical components. The folding mirror is positioned proximal to the proximal end of the first set of optical components and distal to the distal end of the second set of optical components. The third objective lens is positioned to route light arriving from the intermediate image plane toward the photodetector array.

[0009] In some embodiments of the first imaging device, a folding mirror is positioned between the scanning element and the distal end of the second set of optical components.

[0010] In some embodiments of the first imaging device, the first objective lens has a magnification between 50x and 70x, a numerical aperture between 0.9 and 1.1, and an effective focal length between 2.5mm and 3.5mm, and the second objective lens has a magnification between 40x and 60x, a numerical aperture between 0.65 and 0.85, and an effective focal length between 3mm and 5mm.

[0011] Optionally, in the embodiment described in the preceding paragraph, the first objective lens has a magnification of 60×, a numerical aperture of 1.0, and an effective focal length of 3 mm. Optionally, in the embodiment described in the preceding paragraph, the second objective lens has a magnification of 50×, a numerical aperture of 0.75, and an effective focal length of 4 mm. Optionally, in the embodiment described in the preceding paragraph, the first set of optical components includes at least one Pressl eyepiece. Optionally, in the embodiment described in the preceding paragraph, the first set of optical components includes a 12.7 mm diameter, 38.1 mm EFL achromat, and the Pressl eyepiece includes two 12.7 mm diameter, 50.8 mm EFL achromats. Optionally, in the embodiment described in the preceding paragraph, the second set of optical components includes at least one Pressl eyepiece. Optionally, in the embodiment described in the previous paragraph, the second set of optical components comprises a Pressl eyepiece made from two 1-inch diameter, 101.6 mm EFL achromats and a 1-inch diameter, 76.2 mm EFL achromat. Optionally, in the embodiment described in the previous paragraph, the first set of optical components comprises a telescope with 1.5X magnification, and the second set of optical components comprises a telescope with 1.5X magnification. Optionally, in the embodiment described in the previous paragraph, the third objective lens has a magnification between 15X and 25X and a numerical aperture between 0.65 and 0.85. Optionally, in the embodiment described in the previous paragraph, the third objective lens has a magnification of 20X and a numerical aperture of 0.75.

[0012] Another aspect of the invention is directed to a second imaging device comprising first and second sets of optical components, a scanning element, a photodetector array, a third objective lens, and an optically transparent spacer. The first set of optical components has a proximal end and a distal end, and includes a first objective lens disposed at the distal end of the first set of optical components. The second set of optical components has a proximal end and a distal end, and includes a second objective lens disposed at the proximal end of the second set of optical components. The scanning element is disposed proximal to the proximal end of the first set of optical components and distal to the distal end of the second set of optical components. The scanning element is positioned to route excitation light in a proximal-to-distal direction through the first set of optical components such that the excitation light is projected onto a sample positioned distal to the distal end of the first set of optical components. The excitation light projected onto the sample forms a sheet of excitation light at an oblique angle, the position of the sheet varying depending on the orientation of the scanning element. The first set of optical components paths the detected light from the sample back to the scanning element in a distal-to-proximal direction. The scanning element is further arranged to path the detected light so that it passes through the second set of optical components in a distal-to-proximal direction, the second set of optical components forming an intermediate image plane at a location proximal to the proximal end of the second set of optical components. The third objective lens is arranged to path light arriving from the intermediate image plane toward the photodetector array. An optically transparent spacer is positioned and configured to cover the first objective lens and press against the tissue being imaged.

[0013] In some embodiments of the second imaging device, an optically transparent spacer sets the working distance for the first objective lens to capture a depth range of 50-350 µm into the tissue.

[0014] In some embodiments of the second imaging device, the optically transparent spacer is incorporated into a cap that provides a watertight seal between the optically transparent spacer and the distal end of the first objective lens. Optionally, these embodiments further include a quantity of medium positioned between the optically transparent spacer and the first objective lens, the medium having a refractive index selected to match the immersion medium of the first objective lens, the quantity of medium optically coupling the optically transparent spacer to the first objective lens, and the cap providing the watertight seal.

[0015] In some embodiments of the second imaging device, the optically transparent spacer is formed from a solid medium with a refractive index matching the required immersion medium of the first objective. In some embodiments of the second imaging device, the optically transparent spacer has an outer surface positioned between 25 μm and 250 μm proximal to the primary focal plane of the first objective. In some embodiments of the second imaging device, the optically transparent spacer enables rapid 3D imaging of a sample that is gradually moved across the outer surface of the spacer, enabling stitching of consecutive 3D images of the sample. In some embodiments of the second imaging device, the first objective has a magnification between 10× and 70× and a numerical aperture between 0.5 and 1.1. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 1 is a diagram of an embodiment of a sweeping confocal alignment plane excitation microscope. [Figure 2] FIG. 10 is a diagram of another embodiment of a sweeping confocal alignment plane excitation microscope. [Figure 3A] FIG. 3 is a diagram of the excitation and emission geometry of a single objective light sheet in intact tissue for the embodiment of FIGS. 1 and 2. [Figure 3B] FIG. 10 illustrates how monochromatic or bichromatic yz slices are received along the scan direction (x) to create an oblique volume. [Figure 4A] FIG. 3 shows the theoretical operating range of aberration-free imaging for each of the embodiments of FIGS. 1 and 2. [Figure 4B] FIG. 3 shows the theoretical operating range of aberration-free imaging for each of the embodiments of FIGS. 1 and 2. [Figure 5] 3 shows the optical resolution of the embodiment of FIG. 2 at different focal depths. [Figure 6] FIG. 3 is a diagram of an example of an imaging cap designed to cover the distal end of the first objective lens in both the embodiments of FIGS. 1 and 2. DETAILED DESCRIPTION OF THE INVENTION

[0017] Various embodiments are described in detail below with reference to the accompanying drawings, in which like numerals represent like elements.

[0018] Here, we demonstrate a small-form-factor, sweeping confocal-aligned planar excitation microscope called MediSCAPE, which enables volumetric histological imaging of living tissues in situ, in real time, without the need for tissue resection. MediSCAPE's high-speed 3D imaging performance tolerates in vivo motion, enabling mobile 3D image acquisition. Combined with 3D stitching, it enables sequential analysis of large tissue areas. MediSCAPE's high sensitivity enables real-time, multispectral 3D imaging of clinically relevant tissue structures in intact, in situ living tissues, even for weak endogenous fluorescence, without the need for exogenous staining. MediSCAPE is demonstrated in a variety of in vivo settings, including fresh mouse and human tissues, ensuring robust visualization of tissue architecture, disease markers, in vivo perfusion, and tissue function.

[0019] MediSCAPE is an in vivo imaging method based on swept confocal aligned planar excitation (SCAPE) microscopy that enables rapid, nondestructive, in situ examination of tissue at the microscopic level without the need for resection, processing, and staining. This technique has the ability to provide real-time intraoperative feedback, enabling closed-loop treatment decisions, including assessment of surgical margins and monitoring of large tissue areas to guide biopsy site selection. MediSCAPE's nondestructive nature also makes it valuable for nonpathological applications, such as "tissue classification" or perfusion assessment during robotic or orthopedic surgery. Rapid in situ histopathology could also be transformative for the evaluation of organs donated for human transplantation, specifically kidneys, which are the most commonly transplanted organs and are subject to significant interobserver variability.

[0020] The past decade has seen the emergence of a range of new "bedside" fresh-tissue histopathology techniques that reduce the need for tissue dissection; however, these methods image samples ex vivo and often require additional tissue processing. For example, recent innovations applying light-sheet imaging to ex vivo tissues have proven the value of 3D acquisition and visualization. However, their use in vivo is limited by tissue staining and clearing steps and the need to physically move the tissue to form 3D images. Recent demonstrations of two-photon spectroscopy and stimulated Raman spectroscopy for bedside imaging of unprocessed fresh tissue samples have revealed impressive intrinsic contrast. Nevertheless, their limited speed and reliability in costly high-power pulsed laser sources have made their use for in vivo clinical imaging far more limited, except for some features. Furthermore, imaging excised tissue removes the constraint of utilizing a wide range of selective dyes and labels, making simpler techniques such as MUSE microscopy a viable alternative.

[0021] SCAPE microscopy is a high-speed 3D single-objective light-sheet technique that we originally developed to image cellular-level function and structure in model organisms. However, SCAPE microscopy offers two unique capabilities that make it ideal for imaging human tissues in clinical settings. 1) SCAPE acquires 3D images of intact tissue over more than 10 volumes per second, enabling near-instantaneous capture of 3D multilayer structures in in situ tissue equivalent to a boxful of histology slides. This high speed also provides tolerance for the natural motion unavoidable in human surgical settings, enabling "walk-around" acquisition, where a continuum of volumetric images can be acquired and stitched together into a continuous 3D histopathology spanning large areas of intact tissue. 2) Despite its high speed, SCAPE also possesses high sensitivity, enabling detection of endogenous fluorophores already present in most tissues, eliminating the need for exogenous dyes and the use of high-power pulsed lasers, greatly facilitating clinical interpretation and in situ human use.

[0022] MediSCAPE imaging has been demonstrated in a range of in vivo and freshly excised mouse and human tissues, with structures visualized in representative histology of the same tissues. We highlight that MediSCAPE's volumetric data allows for the examination of tissue microstructure in its natural 3D context, with the ability to scroll through cross sections at any random angle, significantly improving interpretation. We have shown that video-rate volumetric imaging speeds enable 3D stitching of "moving around scans" across large tissue regions and overcome motion artifacts by imaging a beating mouse heart in vivo. We also demonstrated the use of exogenous dyes, including proflavine and sodium fluorescein, to highlight well-known cellular components visible in H&E histology.

[0023] Data sets were acquired using two MediSCAPE implementations: an optimized benchtop system (Figure 1) and a novel, miniaturized version of MediSCAPE (Figure 2), with a form factor suitable for human use during surgery with only minor compromises in performance. This miniaturized design demonstrates MediSCAPE's ability to be used not only for in vivo tissue imaging in accessible orifices, but also during laparoscopic, robotic, and open-field surgery. All images shown were acquired using affordable, visible continuous-wave laser light sources (488 nm and 637 nm) at illumination levels equivalent to those of an FDA-approved confocal endoscope. These demonstrations suggest that MediSCAPE can offer a new paradigm for microscopy-based intraoperative guidance.

[0024] Figure 1 depicts a benchtop embodiment suitable for imaging in vivo rodent models and fresh ex vivo mouse and human tissue samples, such as excised kidneys. This embodiment is similar to the configuration disclosed in U.S. Pat. No. 10,061,111 (incorporated by reference herein in its entirety), but adds 488 nm and 637 nm OBIS lasers for excitation and, if needed, a three-axis motorized stage (Thorlabs DDSM50 and MTS25-Z8) for stage scanning. Dual-color imaging is achieved using a homemade image splitter in front of the camera to receive spectrally resolved emission images in parallel. All imaging with this system is performed in an inverted configuration, with water between the objective lens and the coverslip on which the sample rests.

[0025] Figure 2 depicts the miniaturized MediSCAPE design. This compact "folded" design with an elongated imaging head creates a form factor that can be boom-mounted and hand-guided for clinical use, with the only tradeoff being a slightly reduced field of view. Images in this miniaturized Figure 2 system are captured in an upright configuration, with a coverslip to flatten the tissue if needed. Imaging parameters for all data shown are listed in Table 2.

[0026] In both the FIG. 1 and FIG. 2 embodiments, MediSCAPE uses an oblique light sheet to illuminate the sample and receives the returning emitted fluorescence light through the same single, stationary, high-numerical-aperture (NA) objective lens 12. Galvanometer mirrors 32 in the system sweep the light sheet from side to side (along x), descanning the returning fluorescence and mapping it to a stationary, conjugate oblique image plane that is focused on camera 48 (e.g., an sCMOS camera such as the Andor Zyla 4.2+). A plane corresponding to the oblique yz′ region is acquired by camera 48 as galvanometer mirror 32 sweeps the sheet in x to generate a volumetric image. Because all components of the system remain stationary except for the galvanometer mirror, which sweeps one line per volume, imaging speed is limited only by the camera readout rate. The volume acquisition speed is therefore determined by the number of x-steps covered in each volume and the number of camera rows acquired in the camera (corresponding to the depth in z' imaged); fewer rows allow for faster readouts, e.g., about 2,000 frames per second for 100 rows in a standard sCMOS camera. In all systems, dual-color imaging was achieved using a home-built image splitter in front of the camera, which split the image across the columns and received spectrally resolved emission images in parallel without any speed degradation.

[0027] Images were acquired in one of three imaging paradigms: 1) galvanometer-mirror-based scanning of a light sheet for volumetric imaging of stationary samples (approximately 1x1 mm); 2 1) over an xy field of view; 2) moving-around scanning while the sample is manually moved during continuous mirror-based volumetric imaging; and 3) stage scanning of the sample along x with a stationary light sheet (galvanometer stationary). Larger fields of view have been generated by stitching volumes from either moving-around scanning of in vivo tissue or continuous stage scanning of ex vivo tissue. Stage scanning is well suited for rapid 3D scanning just below the surface (e.g., at depths of 50–350 μm) of large tissue sections or slabs. When stage scanning is performed, the scanning element 32 is held in a fixed position, and the sample is translated relative to the entire microscope (or vice versa). Stage scanning can be performed with the probe in contact with or placed on glass or other flat materials and can be imaged from above or below. In addition to being more easily stained with a range of different dyes and stains, ex vivo tissue can be imaged via autofluorescence. Figure 3A shows the excitation and emission geometry of SCAPE's single-objective light sheet in intact tissue. The oblique light sheet illuminates a single plane along yz' while the fluorescent emission is received through the same sample objective.

[0028] FIG. 3B is a diagram illustrating how monochromatic or bichromatic yz' slices are received along the scan direction (x) to create an oblique volume.

[0029] In the embodiment of Figure 1, 488 mm + 637 mm laser light is passed through a 30° Powell lens 68 (PL) and 50 mm and 75 mm cylindrical lenses 61, 62 (CL) to form a uniform light sheet. This sheet is directed into the primary light path with a dichroic 38, positioned off-center in x to form an oblique light sheet at the sample objective 12 (O1), here a 20x water, 1.0 NA Olympus objective with a 2 mm working distance. The resulting plane of fluorescence is received through the same objective 12 and mapped to a stationary intermediate image plane between two telescopic secondary objectives 26 (O2) (Nikon 20x, 0.75 NA, air) and tertiary objective 42 (O2) (Nikon 10x, 0.45 NA, air), arranged in a 4f configuration. Telescope 1 consists of a 75 mm press-fit eyepiece 16 (SL1) and a 150 mm achromat 14 for 2x magnification, while Telescope 2 consists of a 60 mm achromat 22 (SL2) and a 100 mm achromat 24 (TL2) for 1.67x magnification. A stationary image plane is then imaged onto an sCMOS camera 48, with a homemade image splitter 45 providing spectral separation into the two emission channels when needed. A 70 mm focal length tube lens 46 (TL), providing a final magnification of 4.6x, was used for all benchtop data sets except for data from Movie 2 (described later), which used a variable focal length TL from 70 to 200 mm. During mirror-based scanning, as described in U.S. Patent No. 10,061,111, the volume is imaged using a galvanometer mirror to sweep a light sheet and simultaneously descan the emitted fluorescence onto a stationary image plane.

[0030] Most of the previous SCAPE microscopy systems have been used as benchtop instruments for scientific research and have therefore not required significant miniaturization. However, to translate MediSCAPE into clinical use, some miniaturization and simplification of the design relative to the embodiment of FIG. 1 may be advantageous.

[0031] The embodiment of Figure 2 described herein is a MediSCAPE design that maintains imaging performance while having a more compact form factor suitable for intra-operative use in open surgical fields as well as intra-oral and gynecological examinations. With a smaller, specialized primary objective, this same design can be used in smaller openings, such as the ear, nose, and throat, and for arthroscopy and laparoscopy (especially in combination with robotic surgery).

[0032] In the embodiment of Figure 2, images are acquired through a thin, 15 cm long conduit with a diameter of 2.2 cm (the diameter is limited only by the use of a commercially available 60x objective). After a small bend to accommodate the system's galvanometer mirror, the conduit continues in a straight line for 31 cm with a diameter of less than 3 cm and attaches to a scan head that holds additional optics and the system's camera, which connects to a separate computer. The system's camera and laser source can be positioned at a distance from the imaging head and relayed via fiber optic coupling if necessary.

[0033] The embodiment of Figure 2 is portable and suitable for handheld in vivo imaging while maintaining cellular resolution, a practical depth range, and a lateral field of view. Optionally, this embodiment can be constructed using the components listed in Table 1. The entire unit can be mounted to a surgical microscope frame, allowing for flexible, hand-guided movement of the imaging head within the surgical field with small-scale movement and electromechanically stabilized scanning. A thinner section of the imaging arm can be extended to better accommodate laparoscope insertion, while a rod and Grin lens allow for forward- or side-facing MediSCAPE imaging to be added to intraluminal imaging.

[0034] Some embodiments use a water-immersion primary objective with a 2 mm working distance (WD). For clinical use, a sterile sheath can be fabricated to cover the imaging head, which incorporates an optically clear spacer to provide stabilization of the tissue being imaged while ensuring the objective has an optimal working distance to capture a depth range of 200-300 μm into the tissue. In some embodiments, the depth range is 50-350 μm.

[0035] For example, further miniaturization of MediSCAPE is possible by combining fiber optic bundle-based detection with a distal scan head using MEMs mirrors and GRIN lenses. However, this implementation may sacrifice image quality and field of view, making it primarily suitable for gastrointestinal endoscopy.

[0036] For in vivo clinical applications, the embodiment of FIG. 2 provides a thinner, longer imaging head that allows for maneuvering in the surgical field without obscuring the surgeon's access to the field. As shown in FIG. 2, this feature was achieved by using a fold-widening mirror 34 to widen the normally orthogonal telescopic folds of the system of FIG. 1 and positioning a galvanometer mirror 32 in the primary, elongated beam path. A smaller diameter 60x, 1.0 NA water-immersion primary objective 12 (O1) was selected, while the 2-inch diameter lens in the system of FIG. 1 was replaced with a 12 mm diameter optic. To provide more mechanical stability for alignment, the laser illumination is introduced via a single-mode fiber and then directed to the objective 26 (O2), simplifying the imaging head while allowing the image rotation objective and laser emission to all be rigidly mounted on the distal plate of the imaging head.

[0037] The embodiment of Figure 2 illuminates the tissue with an oblique light sheet incident from primary objective 12 (O1). Fluorescence excited by this sheet is received through the same objective. Lens telescopes 14, 16 (TL1 and SL1) map the light between O1 (12) and galvanometer mirror 32, sweeping the excitation sheet from side to side across the sample, and redirect the light returning to second imaging telescopes 24, 22 (TL1 and SL2) to secondary objective 26 (O2). This lens creates an intermediate image of the sample, which remains stationary relative to the scanning light sheet thanks to the descanning function of galvanometer mirror 32. The oblique image of the light sheet in this intermediate image space is relayed to camera 48 by a third, obliquely aligned objective 42 (O3). In the embodiment depicted in Figure 2, this intermediate image space is also used to introduce excitation light.

[0038] The primary objective lens 12 (O1) should provide as high an NA and long a working distance (WD) as possible. We replaced the 20x, 1.0 NA Olympus XLUMPLFLN20XW used in the system of Figure 1, which has a diameter of 30 mm, with a much smaller 60x Olympus water immersion objective (LUMPLFLN 60XW) that is 22 mm in diameter, has a 3 mm effective focal length (EFL) in water, a full NA of 1.0, and a 2 mm WD. The primary impact of moving to this higher magnification objective is a reduction in the system's usable field of view (FOV) from 1.0 mm to 0.4 mm. In an alternative embodiment, the primary objective lens has a magnification between 50x and 70x, a numerical aperture between 0.9 and 1.1, and an effective focal length between 2.5 mm and 3.5 mm.

[0039] To isotropically reproduce the 3D sample volume imaged by O1 in the intermediate 3D image formed by O2, the half-aperture acceptance angle of O2 must be equal to or greater than that of O1. Therefore, we selected a 50x Mitutoyo planar apochromatic objective (#58-237, Edmund) with a 0.75 NA (in air) and an EFL of 4 mm as O2. This objective features a WD of 5.2 mm, which allows sufficient space and flexibility for both reimaging the stationary intermediate image through O3 and launching the excitation sheet. In an alternative embodiment, O2 has a magnification between 40x and 60x, a numerical aperture between 0.65 and 0.85, and an effective focal length between 3 mm and 5 mm.

[0040] The embodiment of FIG. 2 has a first set of optical components 10 having a proximal end and a distal end. The first set of optical components 10 includes a first objective lens 12 disposed at the distal end of the first set of optical components. The first objective lens 12 has a magnification between 10× and 70× and a numerical aperture between 0.5 and 1.1. This embodiment also has a second set of optical components 20 having a proximal end and a distal end. The second set of optical components 20 includes a second objective lens 26 disposed at the proximal end of the second set of optical components 20. In some (but not all) variations of the embodiment of FIG. 2, the first objective lens 12 has a magnification between 50× and 70×, a numerical aperture between 0.9 and 1.1, and an effective focal length between 2.5 mm and 3.5 mm, and the second objective lens 26 has a magnification between 40× and 60×, a numerical aperture between 0.65 and 0.85, and an effective focal length between 3 mm and 5 mm.

[0041] A scanning element 32 is positioned proximal to the proximal end of the first set of optical components 10 and distal to the distal end of the second set of optical components 20. The scanning element 32 is positioned to route excitation light through the first set of optical components 10 in a proximal-to-distal direction such that the excitation light is projected onto a sample positioned distal to the distal end of the first set of optical components 10. The excitation light projected onto the sample forms a sheet of excitation light at an oblique angle, the position of which varies depending on the orientation of the scanning element 32. The first set of optical components 10 routes detected light from the sample back to the scanning element 32 in a distal-to-proximal direction. The scanning element routes the detected light such that the detected light passes through the second set of optical components 20 in a distal-to-proximal direction, with the second set of optical components 20 forming an intermediate image plane at a location proximal to the proximal end of the second set of optical components 20.

[0042] A folding mirror 34 is disposed proximally relative to the proximal end of the first set of optical components 10 and distally relative to the distal end of the second set of optical components 20. In the embodiment shown in FIG. 2, folding mirror 34 is positioned between scanning element 32 and the distal end of second set of optical components 20. However, in an alternative embodiment (not shown), the positions of scanning element 32 and folding mirror 34 are swapped, in which case folding mirror 34 would be positioned between scanning element 32 and the proximal end of first set of optical components 10.

[0043] A third objective lens 42 is positioned to direct light arriving from the intermediate image plane towards a photodetector array 48 .

[0044] Experimental and theoretical characterization reveals equivalent or even better resolution and optical efficiency for the system in Figure 2 compared to that of Figure 1, with the only trade-off being a near-beam-waist resolution of 0.811 ± 0.123 μm (y), 1.07 ± 0.115 μm (x), and 2.10 ± 0.479 μm (z) and a minor reduction in field of view (approximately 1 mm × 1 mm xy for Systems A and B versus approximately 0.4 × 0.6 mm xy for the system in Figure 2).

[0045] Figures 4A and 4B show the theoretical operating range of aberration-free imaging provided by the design of Figure 1 and this combination of O1-O2 of Figure 2, respectively. The Strehl ratio and defocus coefficient of an on-axis point source at varying defocus distances were calculated for both embodiments. As shown, the O1-O2 combination can accommodate a defocus range of approximately ±80 μm (at a Strehl ratio greater than 0.9 and excess defocus less than 5 μm), providing an axial range of approximately 160 μm, which is sufficient for optical imaging of biological tissue. The O1-O2 combination selected for the embodiment of Figure 2 offers a good tradeoff between operating range and compactness.

[0046] In the embodiment of Figure 2, the scan and tube lenses 16, 14 (SL1 and TL1) for the relay telescope between O1 and the galvo mirrors were selected to meet the following criteria: 1) the outer diameter should be as compact as possible, 2) the entire 4f system should form a handheld section long enough for easy maneuverability, 3) the focal length of TL1 should be long enough to reach the back focal plane of O1 (12), which is located 19.1 mm inside the objective but not so long that there is cropping of peripheral rays from the edge of O1's FOV (approximately 400 μm in diameter), and 4) the 6 mm diameter back pupil of O1 should be reduced at the galvo mirror without loss of aperture. Factoring these considerations into the selection of a 12.7 mm diameter, 38.1 mm EFL achromat for TL1 and a press-fit eyepiece with two 12.7 mm diameter, 50.8 mm EFL achromats for SL1. The second relay telescopes (SL2 and TL2) were mounted further away from the sample volume, thus relaxing the constraint on their physical diameter to 1 inch. A press-type eyepiece made from two 1-inch-diameter, 101.6 mm EFL achromats was selected for SL2, and a 1-inch-diameter, 76.2 mm EFL achromat was selected for TL2. For all press-type assemblies, the separation between the component achromats was first modeled in OpticStudio 16.5 (Zemax LLC) and Solidworks 2016 (Dassault Systèmes) and then practically controlled to sub-mm accuracy by stacking 0.4 mm or 1.0 mm thick optical spacers (SM1S01, SM05S1M, and SM1S1M, Thorlabs).

[0047] Although both of these telescopes have a magnification of 1.5x, the ratio of EFLs for O1 (labeled 12) (3 mm) and O2 (labeled 26) (4 mm) produces an effective magnification of 1.33 from the sample to the intermediate image, satisfying the "full 3D imaging condition" for reimaging based on the use of water immersion (n = 1.33) and air (n = 1) objectives for O1 and O2, respectively. The second telescope is folded near the galvo mirror with a 90-degree silver mirror 34 to form a linear configuration as shown in Figure 2.

[0048] To help compact this design of Figure 2, the location where the excitation light is launched into the system has been moved from between SL2 (22) and galvanometer mirror 32 (as in the benchtop design shown in Figure 1) to instead an entrance at O2 (26). This approach effectively creates a light sheet at the intermediate image and relays the light sheet to the sample in the same way that a return image is relayed from the sample to the intermediate image, as described in U.S. Patent Application Publication No. 2019 / 0317312, which is incorporated by reference herein in its entirety.

[0049] To form the sheet, laser light (488 nm) from a single-mode fiber 60 (SM450, Thorlabs) with a mode field diameter of 2.8 μm was filtered through a 15 mm EFL aspheric lens at 1 / e 2 The sheet is collimated into a Gaussian beam approximately 3.33 mm at the waist. This Gaussian beam is expanded approximately 3.3× by cylindrical 4f systems 61, 62 (CL1 and CL2) and then focused by a 50 mm EFL cylindrical lens 66 (CL3), all along the x-direction, to generate an elliptical Gaussian beam. The sheet beam waist is carefully aligned to coincide with the focal plane of O2 (26) and launched at an oblique angle of approximately 39° into O2, corresponding to a 2.5 mm lateral beam offset at the back aperture of O2 (26).

[0050] A Nikon Plan Apo λ 20× 0.75 NA objective was selected as the detection objective O3 (42). This lens was paired with a cylindrical lens 46 (TL3) of appropriate focal length (e.g., 35 mm EFL for tissue imaging or 135 mm EFL for resolution calibration) to magnify the intermediate image onto an sCMOS camera 48 (Andor Zyla 4.2+). Because O3 was calibrated for a 170 μm thick coverslip, a coverslip mount was fabricated using 3D printing and placed in front of O3 to minimize spherical aberration. In an alternative embodiment, O3 has a magnification between 15× and 25× and a numerical aperture between 0.65 and 0.85.

[0051] Resolution was characterized by imaging 200 nm diameter fluorescent beads embedded in 1% agarose gel. A 135 mm EFL cylindrical lens (SAL135F18Z, Sony) was used as TL3 to provide approximately 18x overall magnification from the sample to the camera (Zyla 4.2+, 6.5 μm pixel size). Sampling density was confirmed by manually translating the sample by 100 μm along the x, y, or z direction and quantifying the bead displacement, yielding Δx = 0.337 μm, Δy = 0.371 μm, and Δz = 0.286 μm, respectively. The sheet angle at the sample was calibrated to 39.5°. After deskewing the MediSCAPE data, the FWHM resolution of the system was estimated to be 0.811 ± 0.123 μm (y), 1.07 ± 0.115 μm (x), and 2.10 ± 0.479 μm (z) near the seat waist. The x and y resolutions do not change substantially with depth, but the z resolution decreases with distance from the beam waist, as expected.

[0052] Figure 5 shows the optical resolution of the MediSCAPE embodiment of Figure 2 at different focal depths. Approximately 6,300 beads were extracted from the skew-corrected 3D data, and their FWHM magnitudes along all three directions were estimated. The beads were then grouped according to their depth into 5 μm thickness intervals, whereby the average FWHM and standard deviation were calculated for each depth interval.

[0053] [Table 1]

[0054] FIG. 6 depicts an example of an imaging cap 82 that is fabricated to cover the imaging head (or, more specifically, the distal end of the first objective lens 12 of the imaging head). The cap 82 incorporates an optically clear spacer to provide stabilization of the tissue being imaged. The cap 82 can be used with either the FIG. 1 or FIG. 2 embodiments described herein. In some embodiments, the imaging cap 82 provides the necessary water immersion as well as precise spacing of the primary objective lens (O1) from the tissue being imaged. The imaging cap 82 also advantageously stills and stabilizes the tissue being imaged.

[0055] An example of the cap 82 was fabricated using 3D printing to fit over a standard objective lens 12. A circular glass coverslip 88 was glued to the front surface using cyanoacrylate adhesive, providing a watertight seal. Once placed over the objective lens 12, water 85 was injected into the gap between the lens and cap 82, bonding the objective lens 12 to the coverslip 88. Once the cap's position was adjusted to the correct distance and alignment with the imaging plane, a set screw (not shown) on the body of the cap 82 allowed for fixation. The outer surface of the coverslip 88 was typically positioned 50-150 microns from the primary focal plane of the objective lens 12 (e.g., 1.85 mm from the front surface of a 2 mm working distance objective). In this way, if the focal plane is centered at a depth of 150 microns, tissue pressed against the outer surface of the glass 88 can be imaged over a depth range of 300 microns. (This distance can be selected based on the expected penetration depth into the tissue being imaged.) Indeed, the cap has proven extremely beneficial for unconstrained imaging of human tissue in vivo due to its ability to press against tissue (e.g., oral tissue) to stabilize the tissue being imaged, and its ability to glide across the tissue while maintaining the tissue at a desired working distance. In some embodiments, the cap may be sterilizable and / or disposable for the protection of the patient.

[0056] The cap designed for use with the embodiment of Figure 1 was shaped and sized for attachment to a standard 60x, 1.0 NA, 2 mm working distance, cover-glass corrected, water-immersion, commercially available objective (Olympus). This approach can be applied to any type of objective, including miniaturized and specialized lenses. For example, the lens can be fabricated with alignment grooves or other guides for precise placement / attachment of the cap at the correct distance. The distance may be adjustable by mechanical, electrical, pneumatic, or hydraulic mechanisms. While the use of a glass front surface 88 is ideal when the objective is cover-glass corrected, alternative front surface media can be used if required, including PTFE, due to its refractive index matching to water, or other materials such as PDMS or refractive index-specific polymers. In the case of alternative front surface media, the entire spacer can be solid, or a small droplet of water or a refractive index-matching medium can be bonded to the objective. If the spacer is sufficiently rigid, the cap component that extends back over the objective lens can be more flexible, for example, a thin plastic sheath with the spacer attached on top. In other embodiments, the spacer can be designed into the objective lens itself, providing a lens with a working distance of approximately 150 microns. A thin, index-matched sheath can provide a disposable cover, or the lens can be chemically or thermally stable. Optical components in the optical path (including electrically adjustable lenses) can be used to adjust the effective working distance of the objective lens, allowing the imaging depth range to be adjusted without having to reposition the front surface.

[0057] Optionally, the cap may incorporate or accommodate a method of interacting with tissue, such as to identify injection of a marker dye into the imaging location, or even to obtain a sample at the imaging location.

[0058] In vivo label-free human imaging of the oral cavity with Medi-SCAPE The cap configuration depicted in Figure 6 was successfully used to acquire in vivo human data in the oral cavity of a healthy adult volunteer using both the embodiments of Figures 2 and 1. The cap is configured to ensure an optimal working distance for each objective lens to capture a depth range of 200-300 µm into the tissue while preserving a water immersion interface for the lenses. In some embodiments, the depth range is 50-350 µm.

[0059] Label-free moving-around scans of the tongue, inner lip, and outer lip were acquired by having an adult subject position the appropriate tissue in the imaging cap depicted in Figure 6 and slowly move its position during continuous volumetric imaging for up to 120 seconds at 3-5 VPS. These moving-around scans were stitched into a large, continuous 3D volume. Data from both the Figure 1 and Figure 2 embodiments consistently revealed characteristics of oral tissue layers, including different types of lingual papillae and transitions between different tissue types that recapitulate standard features of oral mucosal histopathology. Bright fluorescence is visible in the filiform papillae of the tongue, likely from keratin and bacteria, while the epithelium of the fungiform papillae is transparent, allowing unobstructed viewing of the bright green internal branching structures that are well-matched to the structure of capillaries. Interestingly, one of the main sources of image contrast in vivo is known to be blood vessels, as seen in both the green autofluorescence of the vessel walls and the red cue corresponding to the blood itself. In the lips, a variety of different vascularized interpapillary structures was seen, progressing from finely pointed at the inner lip to thicker and more stubby at the inner-to-outer lip transition. The lip-to-skin transition captured the hallmarks of hair follicles surrounded by microvasculature.

[0060] The ability of MediSCAPE to image the regularity of these protrusions and continuity patterns of the basement membrane beneath the surface epithelium, as well as the vascular pattern within the lamina propria, suggests that MediSCAPE can feasibly detect a range of oral mucosal diseases, from ulcers and scar tissue to squamous cell carcinoma. Importantly, large areas of tissue of interest, up to 13 mm in length, can be stitched for demonstration and interrogated for early detection of suspicious lesions as well as noninvasive follow-up and monitoring. While the oral cavity was chosen for this first in vivo human demonstration because it is easily accessible in healthy volunteers, the data provide valuable evidence that MediSCAPE is broadly applicable to imaging in situ human tissues in a wide range of clinical settings, including dentistry, otolaryngology, ophthalmology, gynecology, and a variety of open and laparoscopic surgeries and procedures.

[0061] Real-time label-free volumetric imaging of the kidney and heart in vivo Although typically considered a hindrance in fluorescence imaging, autofluorescence in biological tissues can enable visualization of morphological features routinely used for histological evaluation of tissues. Examples of endogenous fluorescence sources in biological tissues include elastin fibers, lipopigments (e.g., lipofuscin and ceroid), phospholipids, and flavins (e.g., flavin adenine dinucleotide, riboflavin, and flavin mononucleotide). These fluorophores can be visible under 488 nm excitation in MediSCAPE. The approximately 525 nm emission channel can capture elastin, flavin, lipofuscin, ceroid, phospholipids, bilirubin, and hyaline, while the approximately 618 nm channel captures relatively larger signals from lipofuscin, ceroid, and porphyrin.

[0062] Furthermore, the distribution and concentration of endogenous fluorophores such as elastin and FAD provide a rich range of molecular information and can indicate changes in tissue health even before structural changes become visible. Label-free imaging in humans is particularly valuable because the use of dyes in vivo is limited by safety constraints, the complexity and cost of obtaining FDA approval, limited penetration depth, extraneous staining, and the time sensitivity of dye administration in clinical settings.

[0063] To demonstrate MediSCAPE's ability to capture autofluorescence contrast in vivo at high speed, mirror-based scanning was used to image the exposed kidney and heart of heavily anesthetized wild-type mice.

[0064] MediSCAPE Benefits and Applications MediSCAPE enables real-time volumetric imaging of intact, in vivo, fresh tissue without the need for exogenous dyes, which can enable simple yet comprehensive assessment of tissue in clinical settings. MediSCAPE's unique advantage over conventional confocal endoscopy is its ultrafast 3D imaging speed combined with much higher sensitivity. These features allow for high-quality in vivo imaging of cellular features and 3D morphology using only autofluorescence contrast, while tolerating in vivo motion and enabling dynamic monitoring of large areas of tissue in real time. MediSCAPE can image a range of different exogenous fluorophores, expanding its utility for a wider range of clinical applications.

[0065] We anticipate that MediSCAPE's primary clinical application will be surgical guidance for lesion resection and biopsy site selection. The form factor of the embodiment in Figure 2 is currently compatible with open surgical fields, including brain, cardiac, orthopedic, and abdominal surgery, as well as tissues within accessible orifices such as the mouth and neck, and potentially laparoscopic and robotic surgery. Results in a mouse model of pancreatic cancer suggest that MediSCAPE can provide valuable guidance during complex Whipple maneuvers. A smaller form factor system of MediSCAPE, or a GRIN lens-based extension, could enable "probe"-style imaging that can guide or be incorporated into needle biopsy procedures.

[0066] MediSCAPE's ability to nondestructively image intact tissues can enable assessment of tissue health, tissue classification, nerve localization, microvasculature mapping, and reperfusion assessment using intravascular dyes for both clinical and veterinary applications. Furthermore, MediSCAPE's sensitivity to autofluorescence can be exploited to reveal metabolic changes as novel disease biomarkers. MediSCAPE may also prove extremely valuable in combination with wide-field imaging of targeted "molecular probes" to visualize cellular-level uptake and disambiguate labeling, particularly during early clinical evaluation. As demonstrated by the successful comprehensive imaging of freshly excised tissues, MediSCAPE microscopy also has considerable potential for rapid 3D assessment of bedside biopsies and excised tissues, with or without exogenous contrast agents.

[0067] Although MediSCAPE's penetration depth is limited by the scattering properties of the tissue being imaged, the high-speed 3D data it generates is equivalent to 10–3 orders of magnitude of thin, serial histological sections. In many situations, this 3D information provides valuable additional information about tissue structure while also enabling motion-around and stitching not possible with 2D planar imaging. While this penetration depth limitation precludes noninvasive imaging of deep tissue structures, the ability to repeatedly image during resection allows for flexible in-situ investigation and the space left behind when overlying tissue is removed. Penetration depth can also be improved through light-sheet optimization or the use of red or near-infrared illumination, particularly in conjunction with red-shifted contrast agents.

[0068] MediSCAPE also advantageously facilitates rapid, nondestructive testing of donor organs prior to transplantation. Many donor kidneys are discarded due to the difficulty of assessing kidney health in the short time interval between donation and transplant. MediSCAPE's ability to visualize key diagnostic features in intact human kidneys supports this potential application, which could be extended to in situ evaluation and biopsy guidance in other transplant organs, such as the liver and heart.

[0069] As demonstrated by comprehensive imaging of freshly excised tissue via stage-scanning acquisition, MediSCAPE microscopy also has considerable potential for rapid 3D evaluation of biopsies and excised tissue at the bedside. MediSCAPE far exceeds the 3D imaging speed limitations of point-scanning confocal, two-photon, and Raman microscopy while avoiding the need for costly dedicated lasers that can be difficult to position at the bedside. Furthermore, because imaging excised tissue eliminates the constraints of utilizing a wide range of selective dyes and labels compatible with fresh tissue, MediSCAPE results on stained fresh tissue demonstrate that the bedside form of MediSCAPE can provide a more comprehensive evaluation of biopsy tissue as a complementary / cross-validation to its in vivo use. Ex vivo tissue can also be chemically cleaned to provide a more comprehensive 3D visualization. Although the step of cleaning the tissue can take excessive time, the cleaned tissue can be imaged using the embodiment of FIG. 1, which offers advantages over a two-objective light sheet system, including the simplicity of the single-objective light sheet geometry and the ability to image to the full depth of the working distance of the primary objective.

[0070] Source of contrast The majority of images described herein were acquired with a single 488 nm laser for fluorescence excitation. However, a wider range of excitation wavelengths can be easily incorporated into MediSCAPE, including 405 nm, 561 nm, and the near-infrared range. Additional wavelengths can be utilized for autofluorescent molecules such as NADH, collagen, or retinol, as well as exogenous dyes that extend into the near-infrared, such as indocyanine green.

[0071] Although autofluorescence imaging has been compared with traditional histological contrast as a representative, autofluorescence has the potential to reveal additional valuable information beyond that seen in histology. For example, autofluorescence detected with 488 nm excitation in human kidneys was particularly intense in the elastic lamina of arterial walls, cytoplasmic lipofuscin deposits, and urinary cast material. Also clearly visible were cytoplasmic granular structures within the epithelial cells of pseudohypertrophied proximal tubules and focal arterioles with intense punctate perinuclear autofluorescence suggestive of lysosomal signals. Nearly all of these tissue features are largely unremarkable in routine histology, suggesting the potential for MediSCAPE to glean additional information beyond what traditional histology can provide. Novel diagnostic features have significant clinical significance, especially for limited or precious human tissue, such as small needle-core biopsies.

[0072] Visualization, display, and automated analysis A key factor in the clinical adoption of MediSCAPE is how the data can be visualized and interpreted in real time by both the acquiring surgeon and the examining pathologist. While all analysis and rendering of MediSCAPE images described herein was performed offline, real-time stitching and visualization of depth and cross-sections should be feasible using field-programmable gate array (FPGA) technology, which performs well for real-time visualization and rendering of ultrasound and OCT data. Furthermore, the digital nature of MediSCAPE data allows for online review of datasets by remote pathologists (as is common in radiology), who can easily select their preferred viewing and color scheme. MediSCAPE's rich volumetric data would ideally also be amenable to machine-learning-based analysis, which can automatically classify normal and suspicious regions and extract important tissue features. The results of online analysis could be projected onto a visualized surgical field using augmented reality. When available, MediSCAPE data can be spatially registered to stereotactic coordinates and other imaging modalities, such as MRI, and fully archived as part of the patient's electronic health record.

[0073] Technical Developments and Form Factors While most of the results described herein utilized the MediSCAPE embodiment of Figure 1, the embodiment of Figure 2 demonstrated nearly equivalent performance, and its form factor is compatible with being mounted on a surgical microscope frame and manually guided through the surgical field. Further miniaturization using MEMs mirrors, optical fibers or rod and GRIN lenses, and specially constructed small diameter, high NA objectives can all further reduce the system's form factor to enable laparoscopic and even endoscopic use.

[0074] For routine clinical use, the system optionally uses an optically clear spacer, incorporated into a disposable or sterilizable sheath, at the tip of the primary objective to press against the tissue at an optimal working distance. Features such as microscale stabilization and automated scanning over a fixed distance can improve ease of use, while the ability to mark, capture, or even laser ablate identified areas in conjunction with imaging can provide considerable benefits for microscale resection. MediSCAPE's ability to dynamically zoom to features of interest is also beneficial, providing a compromise between covering larger areas through movement and capturing key features of disease in the tissue of interest.

[0075] In short, MediSCAPE is a powerful new approach to in situ histopathology that harnesses the unique benefits of light-sheet scanning to enable fast, 3D, label-free imaging of a wide range of tissues. Beyond replacing biopsies and traditional histopathology, MediSCAPE has the potential to open new doors to the nondestructive evaluation of a wide range of valuable tissue features in situ. These new capabilities could significantly improve the standard of care while also reducing the time and cost of a wide range of surgical procedures.

[0076] Data Processing MediSCAPE data processing consisted of background subtraction, deskewing the data, and fusing the dichroic images with a custom-written MATLAB graphical user interface (GUI). A pseudo-flatfield correction was applied to the dichroic images along the x- and y-axes by dividing the volume with a Gaussian blur mean intensity z-projection. For better visualization of details, the MediSCAPE data shown in Image Set 3 were processed with unsharp masking (radius 1, amount 0.3) and CLAHE histogram equalization (block size 75, slope 2).

[0077] For H&E pseudocoloring of MediSCAPE datasets acquired with proflavine staining, the virtual H&E algorithm developed by Giacomelli et al. was used to create brightfield H&E color channels from fluorescence data based on the Beer-Lambert law. Autofluorescence emission received through a 618 / 45 nm bandpass filter at 488 nm excitation was used to indicate gross nonnuclear background structures on a logarithmic scale (eosin), while proflavine fluorescence excited at 488 nm was used to indicate nuclear structures (hematoxylin).

[0078] Data Stitching A key feature of MediSCAPE is its extremely fast 3D imaging speed, even when imaging weak autofluorescence. This speed can be exploited to enable exploration of large areas of tissue by "moving around" or continuously moving the tissue relative to the system's 3D field of view. MediSCAPE's speed can tolerate this transformation without significant artifacts in each individual volume, and because each volume ultimately has some spatial overlap, a series of volumes can be stitched to generate a fully contiguous 3D strip of data spanning several millimeters or more. This feature, which does not require continuous or motorized movement and can tolerate inevitable in vivo movements such as respiration, makes it ideal for assessing transitions between tissue types or for exploring exotic regions for spatial patterns of multiple scales at the cellular and mesoscopic levels.

[0079] To stitch consecutively acquired overlapping volumes from moving scans, a special ImageJ macro was written using the existing Pairwise Stitching plugin in Fiji to operate on volumes already saved as background-subtracted two-color TIFF stacks in MATLAB. Volumes were stitched pairwise to simulate real-time stitching that can be implemented using an FPGA. Because volume velocity is generally much higher than the rate at which tissue is translated during acquisition, every nth acquired volume (where n = 2–5) was used for stitching to shorten overall processing time and reduce stitching errors. The stitched volumes shown in Image Set 1, Image Set 2, Image Set 7, and Movies 1, 3, and 10 (described below) were created by fusing approximately every fourth volume acquired consecutively in a pairwise manner. During each stitching step, the volumes were downsampled 2x along the Y and Z axes and roughly aligned, given the aligned position found in the previous successful stitching step. If the alignment r value exceeded a given threshold (approximately 0.8), a fine alignment of the raw volumes was performed using the initial coarse alignment value, and the raw volumes were fused using linear blending with a 10% overlap. If the coarse alignment r value fell below the threshold due to excessive motion, the next successive volume was loaded and aligned until the volumes were accurately aligned. The data were deskewed in MATLAB after stitching. To produce the "real-time stitching" movies described below for Movie 1 and Movie 3, each stitching step was deskewed and positioned on a blank canvas of the same 3D size as the final fully stitched volume.

[0080] The Bigstitcher plugin in Fiji was used to stitch the stage-scanned data. A specialized MATLAB and ImageJ pipeline was implemented to automatically save background-subtracted and deskewed two-color TIFF stacks in MATLAB, convert the data in HDF5 format for loading into BigStitcher, pre-align the volumes with stage coordinates, and stitch the data using the default stitching wizard presets and linear blending with fine ICP alignment.

[0081] Fourteen sets of sample images (referred to herein as Image Set 1 through Image Set 14) and ten movies (referred to herein as Movie 1 through Movie 10) were captured / created to demonstrate the capabilities of the hardware described herein. More specifically, they are as follows:

[0082] Image set 1 is a mirror-based scan of 1x1.4x1.1 μm in 0.78 seconds 3 / voxel sampling density, 802x861x275μm 3 Figure 1 shows a volume rendering and individual planes of an in vivo mouse kidney captured as a two-color xyz volume. Autofluorescence was excited with 488 nm light, and the two emission channels were captured through 525 / 45 nm and 618 / 45 nm bandpass filters, using blue and "yellow hot" color maps that allowed for better visualization of overlapping channels.

[0083] Renal tubules exhibited robust autofluorescence in both emission channels, with proximal tubules exhibiting higher emission at approximately 525 nm (yellow-hot) than distal tubules (blue / purple). Autofluorescence in this range may be due to flavins in metabolically active proximal tubule cells. Nuclei are distinguishable along the tubule wall as punctate dark areas. H&E histology processed from a similar region of mouse kidney cortex showed normal tubular structure. While structural information between both types of images was similar, MediSCAPE provided additional molecular contrast based on the spectrally resolved emission of endogenous fluorophores, including flavins, elastin, porphyrins, and lipofuscin.

[0084] A key feature of MediSCAPE is that this autofluorescence contrast can be captured in real time, allowing for easier exploration of large 3D fields of view in tissue. To capture a "moving scan," an anesthetized mouse was manually translated along three dimensions to mimic how the MediSCAPE imaging probe moves across tissue in intact living tissue: 358 x 798 x 165 μm in x, y, and z. 3 The bichromatic volume is 2.5x1.4x1.1μm 3The scan was acquired at 9.3 VPS while moving continuously across the intact kidney cortical surface with a sampling density of 1 / voxel. In addition to providing a continuous sequence of high-quality volumetric images across a 1 x 3 mm strip of living kidney, this moving data was stitched to generate a contiguous volume. To generate this larger field of view, overlapping 3D volumes were stitched using the pairwise stitching plugin in ImageJ, similar to how volumes are stitched in real time on a field-programmable gate array (FPGA). In the more detailed image, nuclei appear as negative space along the tubule wall, with clear differences between proximal and distal tubules based on both structure and spectral emission. These features are similar to the previous single volume scan, despite a coarser x-step of 2.5 µm across the reduced x-range and depth range, allowing for real-time speed. Video 1 (described later) shows a real-time replay of the moving scan with cross-sections and depth sections from each acquired volume, and the stitching of the larger field of view as overlapping volumes are acquired.

[0085] While cellular features are visible at this sampling density, MediSCAPE has the ability to trade off resolution for field of view to "zoom in" on features of interest. Video 2 (described later) shows mouse kidney data acquired using a variable 70-200 mm focal length tube lens, switching between normal (4.6x) and high magnification (11.4x), revealing a more detailed visualization of tubular structures. This "zoom-in" feature can be easily automated and stitched together with coarser images of a larger field of view acquired at lower magnification.

[0086] Image Set 2 demonstrates MediSCAPE's tolerance to inherent in vivo motion. The same in vivo specimen was used to image a beating, intact, in vivo mouse heart. The data are from 12.9 VPS (2.5 x 1.4 x 1.1 μm3 305x798x138μm with a sampling density of / voxel 3 The image was acquired by moving across the exposed cardiac surface while continuously acquiring two-color volumes in a galvanometer scan (for a volume size of 1000 mm). Image Set 2 shows xy slices in a 3D stitched field of view created from 15.6 seconds of data acquired while using a three-axis stage to manually move across the cardiac tissue. Striated cardiomyocytes in the myocardium were clearly visualized. While veins and arteries appear as negative space, arteries can be distinguished by the highly autofluorescent elastin along their walls. Elastic fibers are also seen on the myocardial surface. Granular autofluorescence along muscle fibers may be lipofuscin, a lipopigment that accumulates in highly active cells over time. The periodic cardiac pulsation that occurred during this acquisition appears as abrupt lateral movements along the y-axis in the kymograph, with peaks of maximum intensity in x and y over the 15.6 seconds of imaging. The system was able to successfully acquire volumes that could be stitched together with minimal visible motion artifacts or blurring. Video 3 (described below) shows a real-time playback of cross-sections of a beating heart, as well as the stitching of these volumes as they were acquired. Stitching three-dimensional tissue volumes compensates for tissue motion in all three dimensions, allowing for movement along the lateral and depth axes. Compared to stitching traditional 2D views, volume stitching inherently reconstructs 3D tissue structure more reliably, correcting for out-of-plane motion that is unavoidable in vivo.

[0087] Image Set 3 demonstrates the characterization of tissue structures visible solely by autofluorescence in various freshly resected mouse tissues imaged with MediSCAPE. Image Set 3 shows xy transverse slices at various tissue depths where H&E histology shows the same or adjacent regions in the mouse tissue. Movies 3 and 4 (described later) show full-depth fly-through movies of each 3D volume. The freshly resected tissues included the following: myocardial fibers in the ventricles; the cerebellum in a sagittal section of the brain; the alveoli and visceral pleura in the lungs; typical hepatocyte cord formations and capsules in the liver lobule; the red pulp and surrounding capsule in the spleen; the superficial layer in the urinary bladder mucosa, with pixel intensities displayed on a logarithmic scale for better visualization; visible muscle fibers deep within the thigh muscle; and the Lieberkühn pits in the colonic mucosa.

[0088] Through intrinsic contrast alone, micron-scale structures were visible in all fresh tissues studied, corresponding well to those visible in H&E histology. For example, the shape and diameter of Lieberkühn's pits can be clearly distinguished in colonic mucosa. Alveoli in lung tissue, clearly delineated by highly fluorescent elastin, appear intact, but histology often reveals significant distortion due to dissection of delicate, air-filled tissue. Layers within tissues, such as the urinary bladder mucosa, are clearly visible and can be evaluated in 3D, allowing for a more comprehensive assessment than 2D histological sections and single-plane confocal microscopy. The maximum depth at which histological-level resolution is possible is tissue-dependent and also varies with excitation wavelength. For many tissues, resolution begins to decline after 50 μm with 488 nm excitation. For example, in skeletal muscle, cellular-level contrast was observed 121 μm into the tissue. The images were taken at 100 fps with a laser power of 5-7 mW at each sample, with a resolution of 1 x 1.4 x 1.1 μm. 3 801x1065x275~330μm in xyz at a sampling density of / pixel 3 The volume size was obtained.

[0089] Detection of histological features associated with disease states in human tissues To test MediSCAPE's ability to capture features associated with disease in human tissue, fresh human kidney tissue was obtained from surgical nephrectomy specimens and imaging results were compared with conventional periodic acid-Schiff (PAS) and H&E histology on the same samples.

[0090] Image set 4 shows autofluorescence imaged by MediSCAPE in a nephrectomy specimen from a patient with underlying chronic kidney disease (CKD). A 13.3x10.6x0.3mm 3D image was used to map the area in the SCAPE image with the area in the histology image. 3 To acquire and stitch a volume of 2.1 x 1.6 mm, a motorized stage scan was used to image the entire planar surface of the fresh specimen. The imaging data was acquired over a period of 196 seconds. Movie 7 (described later) shows a depth fly-through movie showing a cross section through the fully stitched volume. From the fully stitched volume, a 2.1 x 1.6 mm 2The xy ROIs were obtained. Examples of important diagnostic features identified by MediSCAPE include arteriosclerosis and arteriolar hyalinosis. Arterial identification is aided by the strong autofluorescence of the internal elastic lamina of the arterial wall, which becomes even more prominent in MediSCAPE imaging in the setting of hypertensive arteriosclerosis, where the lumen is narrowed by intimal thickening with double elastic lamina. We were able to clearly identify glomeruli and distinguish glomeruli showing total sclerosis. We were also able to distinguish subglomerular structural elements, including the glomerular capillary tuft, Bowman's space, and Bowman's capsule, especially when Bowman's capsule had partial sclerosis (Image Set 4d, note arrow). We were also able to identify several glomerular features associated with CKD, including segmental glomerulosclerosis, focal hyalinosis, and nodular mesangial sclerosis (data not shown). Characteristic chronic changes in the tubulointerstitial compartment, including tubular atrophy and interstitial fibrosis, known to have the strongest correlation with renal outcome, were clearly evident in MediSCAPE images. We were able to distinguish atrophy from non-atrophic tubules and identify pseudohypertrophy of proximal tubules and tubular casts.

[0091] Image Set 5 highlights the unique value of MediSCAPE's isotropic 3D imaging of intact, fresh tissue. Image Set 5 shows an example of a clinically relevant lesion that may be unclear to identify from 2D thin sections. In several planar images, a single image revealed small cystic structures that could be either large, dilated atrophic tubules or simple renal cysts. However, the 3D data revealed a residual, compressed, and sclerotic capillary plexus pressed against the inner wall of the cystic space, distinguishing this structure as an atubular glomerulus (or "glomerular microcyst") rather than some kind of tubular-generated element. When examining perirenal fat from normal human kidney tissue, we also found that MediSCAPE could capture the 3D arrangement of elastic fibers and adipocytes based on intrinsic autofluorescence. Evaluating these features in volumetric space could enable a more accurate assessment of fiber structure, density, and identity (e.g., collagen versus elasticity) in different tissue compartments, in addition to fat content.

[0092] Volumetric imaging of localized dyes in fresh human tissue Image Set 6 demonstrates that MediSCAPE can image clinically relevant features using a wide range of fluorescent contrast agents, when available. Image Set 6 shows example MediSCAPE data received from a fresh normal human kidney specimen stained with proflavine, a topical nuclear dye commonly used in clinical imaging studies. Proflavine and red autofluorescence emissions were acquired at 488 nm excitation using stage scanning to create a 7500 x 918 x 164 μm volume in 5.6 seconds. Proflavine staining revealed nuclear size, shape, and distribution, while autofluorescence provided complementary structural information. Following recent practice for visualizing tissue fluorescence in histopathology, we also generated dual-color MediSCAPE data using a pseudocolor H&E color scale using proflavine as a hematoxylin analog (purple) and "eosin" (pink), represented by the autofluorescence signal received with a 618 / 45 nm bandpass filter. Pseudocolored MediSCAPE images closely resemble traditional brightfield H&E histology and can allow for easier assessment of nuclear detail when needed. Video 8 (described below) shows a fly-through of the top 30 μm of depth in a pseudocolored MediSCAPE volume.

[0093] In vivo volumetric imaging of perfusion around the clock using intravascular fluorophores Image Set 7 demonstrates MediSCAPE's ability to perform real-time 3D in vivo imaging of microvascular perfusion. We imaged the brain of a live, head-fixed mouse through a glass cranial window following intravenous injection of dextran-conjugated fluorescein. Moving scans were acquired at 9 VPS, generating 3D stitched volumes and multiview maximum intensity projections of a single volume. Video 10 (described later) shows real-time moving data, demonstrating its clear ability to observe dynamic flow within an organ while also capturing clean details of 3D microvascular architecture without motion artifacts. In addition to neurological surgical applications, this technique may be valuable for assessing the microvasculature at tumor borders or after tumor embolization, arteriovenous malformations, and organ reperfusion. MediSCAPE can utilize commonly used intravascular fluorophores, such as fluorescein and the near-infrared fluorophore indocyanine green for deeper tissue penetration.

[0094] Image Set 8 demonstrates MediSCAPE images of 200-nm fluorescent beads embedded in gel. Maximum intensity projections over a 60 μm range were taken along all three axes from the skew-corrected raw data. Each cross section measures 390x742x145 μm. 3 The image was scaled to give an isotropic μm / pixel across the xyz field of view.

[0095] To compare the imaging performance of the system in Figure 2 with that of Figure 1, we imaged fresh, unstained mouse tissue. Using 488 nm excitation and approximately 4.6 mW of incident power on the sample, we imaged an area of ​​1.0 × 1.4 × 1.08 μm in x, y, and z. 3Galvanometer scanning was used to capture two-color volumes measuring 400 x 700 x 162 μm in x, y, and z, acquired at a sampling density of 100 μm / voxel. Images were acquired at 300 Hz (0.75 VPS) to compare tissue structures with those acquired with the system in Figure 1. Image set 10 shows cross sections displaying tubules in the kidney cortex, fibrous capsules and underlying cords of hepatocytes and sinusoids in the liver, myocardium on the cardiac surface, and Lieberkühn's gland pits in the colonic mucosa. These volumes demonstrate significantly closer penetration depth and tissue structure resolution compared to the system in Figure 1, despite the smaller, rounded field of view (caused by using a smaller form factor 60x objective as O1).

[0096] As a further demonstration of imaging performance in fresh tissue, we performed imaging of a 250x700x136 μm area at 11.2 VPS while moving across the colonic mucosa. 3 xyz volumes were acquired sequentially. A stitched field of view from a 16-second walk was created by stitching the received volumes together, approximately pairwise. Distinct crypt structures were visible along each dimension, even with a sampling density of 2 μm along x and frames received at 1400 fps.

[0097] During these scans, equivalent signal levels were detected on the Andor Zyla 4.2+ camera. The well-matched signal-to-noise and resolution of both systems are consistent with models predicting that the MediSCAPE FIG. 2 configuration actually receives a larger NA of emitted light and is more light-efficient than the FIG. 1 system. The primary tradeoff for the smaller form factor is the size of the field of view resulting from the use of a more compact 60x primary objective in the FIG. 2 embodiment. This demonstration supports the feasibility of implementing MediSCAPE through a smaller primary objective and a longer, thinner telescope or relay lens to allow for easy steering and precise positioning of the MediSCAPE imaging head into the surgical field.

[0098] Image set 10 shows label-free imaging of fresh mouse tissues with the embodiment of FIG. 2. Xy (top) and yz (bottom) cross sections of various fresh mouse tissues were acquired with the system of FIG. 2 with 488 nm excitation and dual-color emission channels. The cross sections showed tubules in the kidney cortex, hepatocyte capsules and underlying cords in the liver, myocardium in the heart, and Lieberkuhn's gland pits in the colonic mucosa. Image quality is similar to the design of FIG. 1, with nuclei visible in the kidney tubules, Lieberkuhn's gland pits in the colonic mucosa, and individual elastin fibers in the liver capsule. The main difference is the reduced field of view, which can be mitigated by stitching a larger field of view by moving around the tissue.

[0099] Image set 10 demonstrates two-color autofluorescence visualization. An xy image plane was acquired with MediSCAPE of fresh mouse brain cortex, including prominent blood vessels. The contrast corresponded to autofluorescence excited by 488 nm light. Two-color emission images were acquired simultaneously using an image splitter in front of the camera, positioning each color channel next to each other (along the y axis) at the camera tip. Two images in this set show the grayscale raw emission channels acquired with 525 / 45 nm and 618 / 45 nm bandpass filters, respectively. These channels were converted to "yellow-hot" and blue color maps and then fused to the other images in the set.

[0100] Image set 11 demonstrates autofluorescence in kidney tissue from a diabetic individual imaged with MediSCAPE. The autofluorescence captured by MediSCAPE revealed features more than those seen in routine histology. One image in this set was a PAS histology image of renal cortical tissue from an elderly diabetic patient with features of mild diabetic nephropathy. The other image in this set was a MediSCAPE xy slice from a stage-scanned volume of the same tissue section (while fresh) showing autofluorescence excited at 488 nm. Renal capsules and urinary casts were visible in both the MediSCAPE and PAS images. Another image in this set showed focal subcapsular collections of renal tubules with autofluorescent cytoplasmic granules. Urinary cast material was also evident in the xy plane and was further revealed by characteristically strong autofluorescence in the yx plane. Another image in this set showed renal tubules with accentuated peritubular autofluorescence. Another image in this set showed glomeruli with focal autofluorescent granules.

[0101] Image set 12 demonstrates MediSCAPE's label-free imaging of elastic fibers and adipocytes in human perirenal fat. One image in this set was a 3D rendering (ImageJ 3D Viewer) of a section of normal human perirenal fat showing highly fluorescent elastic fibers and adipocytes. The other image in this set was a yz-section from the indicated plane, showing a layer of fibers over adipocytes, which are distinguishable as round yellow droplets. The other image in this set was a cross-section in which adipocytes and crossing blood vessels were visible.

[0102] Image set 12 also demonstrates stained human kidney tissue imaged with MediSCAPE. Fresh human kidney tissue showing features of arteriosclerosis was stained with either methylene blue or proflavine nuclear dyes and imaged with MediSCAPE. The same tissue block faces were then processed for histology stained with PAS and / or H&E. Three images in this set demonstrate how the four major kidney tissue elements that must be routinely evaluated in both PAS and H&E histology appear in PAS histology, in xy slices of a MediSCAPE volume stained with methylene blue, and in H&E histology. These images revealed glomeruli, arteries, tubules, and interstitium. Similar to H&E, methylene blue in MediSCAPE images defined the cytoplasmic, nuclear, and extracellular compartments of cells, but, similar to PAS histology sections, better highlighted the arterial elastic lamina, tubules, and interstitial compartments.

[0103] A second biopsy from the same patient showed scarred tubulointerstitium in focal areas of fibrosis in both the MediSCAPE image and the corresponding H&E histology image. The other image in this set was a 3D rendering (Imaris) of a larger stage-scanned volume acquired in MediSCAPE. This image showed the 3D structure of fibrosis, arteries, and glomeruli. The origin of the xz depth zone is visible. Two more images in this set showed non-sclerotic glomeruli in greater detail over a 20 μm depth range.

[0104] All four renal histologic compartments, generally assessed through a combination of H&E and PAS histology, are clearly distinguishable in MediSCAPE images, particularly with methylene blue staining.

[0105] Image set 14 demonstrates a comparison of topical dyes applied to fresh mouse colonic mucosa. Single xy and yz slices of a sample of fresh mouse colonic mucosa were imaged with MediSCAPE. In three images from this set, contrast was derived from a) 0.01% proflavine, a nuclear dye (excitation 488 nm, emission 525 / 45 nm), b) 1% methylene blue, a clinically used nuclear dye (excitation 637 nm, emission >685 nm), and c) sodium fluorescein, an FDA-approved topical IV dye (excitation 488 nm, emission 525 / 45 nm). Corresponding cross sections and nuclear locations were visible. Depth penetration of the topically applied dyes, as seen in yz depth sections, was both stain- and tissue-dependent. Lieberkühn's gland pits and goblet cells were visible. These results demonstrate MediSCAPE's ability to capture a variety of exogenous contrasts with high signal-to-noise and also highlight the difficulty of ensuring dye penetration compared to utilizing endogenous contrast.

[0106] Video 1 demonstrates a label-free in vivo mouse kidney imaged with MediSCAPE at 9.3 VPS. Dimensions: 358x798x165μm 3 Xy and yz sections from the two-color volume were acquired at 9.3 VPS while moving around the in vivo mouse kidney. 3D renderings (ImageJ 3D Viewer) of larger fields of view and cross sections were stitched from overlapping volumes as they were acquired. Reconstruction was real-time with volume stitching performed after processing. Imaging parameters are listed below in Table 2.

[0107] Video 2 demonstrates a fresh mouse kidney imaged at low and high magnification in MediSCAPE. Autofluorescence in the same region of fresh kidney tissue was imaged at different magnifications. A variable focal length tube lens allows for dual-color imaging at a range of magnifications that is adjustable based on the required resolution, field of view, and volumetric speed. One set of data was received with the variable tube lens set at f = 70 mm for 4.6x magnification, and the other set of data was received at the same region immediately after setting the variable tube lens to f = 170 mm for 11.4x magnification. Autofluorescence emission in the range of approximately 525 nm is shown in yellow, and emission at approximately 618 nm is shown in blue. Imaging parameters are as shown below in Table 2.

[0108] Video 3 demonstrates level-free MediSCAPE imaging of an in vivo mouse heart imaged at 12.9 VPS. 3 X-y and y-z sections from a two-color volume of were acquired at 12.9 VPS while moving around across an intact, beating mouse heart. The beating of the heart was periodically visible in the maximum intensity projection of the data over time in each volume. Transverse (x-y) sections at different z-depths from the larger 3D field of view were stitched from overlapping volumes as they were acquired. Reconstruction was real-time, with post-processing stitching of the volumes. Autofluorescence emission in the range of approximately 525 nm is shown in yellow, and emission at approximately 618 nm is shown in blue. Imaging parameters were as shown below in Table 2.

[0109] Video 4 demonstrates autofluorescence in fresh mouse heart, brain, lung, and liver imaged with MediSCAPE. Depth fly-throughs of endogenous contrast were imaged 50 μm from the surface of intact, freshly excised mouse heart, sagittally cut cerebellum, intact lung, and intact liver tissue. Autofluorescence emission in the range of approximately 525 nm is shown in yellow, and emission at approximately 618 nm is shown in blue. Imaging parameters are as shown below in Table 2.

[0110] Movie 5 demonstrates autofluorescence in fresh mouse spleen, bladder, muscle, and colon imaged with MediSCAPE. A depth fly-through of endogenous contrast was imaged 100 μm from the surface of the intact, freshly excised spleen, bladder mucosa, thigh muscle, and colon mucosa. Autofluorescence emission at approximately 525 nm is shown in yellow, and emission at approximately 618 nm is shown in blue. Imaging parameters are as shown below in Table 2.

[0111] Movie 6 demonstrates the kidney, liver, heart, and colon of a fresh mouse imaged with the system in Figure 2. Depth fly-throughs of endogenous contrast were imaged 50 μm from the surface of the kidney, liver, heart, and colon mucosa of four fresh, intact mice. Autofluorescence emission in the range of approximately 525 nm is shown in yellow, and emission at approximately 618 nm is shown in blue. Imaging parameters are as shown below in Table 2.

[0112] Video 7 demonstrates a MediSCAPE autofluorescence image of a fresh human kidney biopsy with chronic kidney disease. A 13.3x10.6x0.3mm xyz field of view was stitched from 12 stage-scanned two-color volumes acquired in a total of 196 seconds. Autofluorescence emission in the range of approximately 525nm is shown in yellow, and emission at approximately 618nm is shown in blue. Imaging parameters are as shown below in Table 2.

[0113] Video 8 demonstrates a depth fly-through of H&E pseudocolor MediSCAPE images of fresh normal human kidney tissue stained with proflavine. Proflavine fluorescence was coded as hematoxylin (purple), and red autofluorescence emission was coded as eosin (pink) (488 nm excitation). The complete 7500x918x164 μm xyz volume was acquired by stage scanning in 5.6 seconds.

[0114] Video 9 demonstrates a 3D rendering and fly-through of proflavine-stained human kidney tissue imaged with MediSCAPE. The 2732 x 921 x 273 μm xyz-stage scanned volume showed signs of arteriosclerosis. Localized areas of scarred cortex due to tubular atrophy and interstitial fibrosis were evident near the center. Glomeruli and arteries were clearly visible due to autofluorescence and the proflavine signal excited at 488 nm, and their structures were more easily appreciated by scrolling through both cross- and depth sections.

[0115] Video 10 demonstrates mobile MediSCAPE imaging of the mouse brain vasculature in vivo with IV FITC-dextran. Volumes were acquired through a glass cranial window at 9 VPS while moving around the glass cranial window. 3D rendering of the real-time data was performed during the move. A larger 3D field of view was constructed by stitching overlapping volumes as they were received. Playback was real-time with volume stitching performed after processing.

[0116] [Table 2A] [Table 2B]

[0117] The following notes apply to Table 2: (a) Unless otherwise noted, samples were label-free, fresh, in vitro tissue. (b) For dual-color acquisitions, the y-dimension is given as the final cropped y-dimension of one color channel. The original y-dimension at the camera is 2x greater than the final cropped y-dimension because the color images were acquired simultaneously, side-by-side along the y-axis at the camera. The x-dimension is given as the unskewed x-dimension of the acquired volume (# x-steps, *x-step magnitude). (c) Unless otherwise noted, scans were acquired with a 70-mm focal length tube lens, giving an effective magnification of 4.66x. (d) If the scan type was a mirror-based moving-stage scan, the volume velocity was reported. For mirror-based stationary stage scans, the total acquisition time is given in seconds. (e) Laser power is typically for 488 nm laser excitation.

[0118] The MediSCAPE embodiment described herein was compared with confocal and two-photon microscopy. More specifically, MediSCAPE with 488 nm excitation was compared with confocal microscopy with 488 nm and 561 nm excitation and two-photon microscopy with 800 nm excitation. To compare autofluorescence contrast, fresh mouse colon mucosa and kidney samples were imaged with all three techniques. While cellular and tissue-level features were largely similar across all three techniques, point scanning required significantly longer acquisition times for weak endogenous fluorescence.

[0119] Key advantages of MediSCAPE are its real-time 3D speed, which facilitates imaging of large areas in vivo, and its sensitivity, which enables detection of weak endogenous contrast. Here, we demonstrate why MediSCAPE's speed and sensitivity are orders of magnitude better than point-scanning confocal and two-photon microscopy, the traditional techniques of choice for optically cropped fluorescence imaging. We also demonstrate that MediSCAPE autofluorescence images of histology in fresh tissue are qualitatively similar to images acquired by confocal and two-photon microscopy at similar excitation and emission wavelengths.

[0120] Feasibility of high-speed 3D scanning with MediSCAPE and point-scanning confocal and two-photon MediSCAPE's use of light-sheet excitation offers considerable improvements in sensitivity due to parallelized excitation and emission detection across planes in the tissue volume. This parallelization allows for longer integration times and gentler laser excitation power, which reduces photobleaching and phototoxicity in the tissue. Confocal endoscopy and bedside two-photon systems, on the other hand, use point scanning, in which each individual pixel in the tissue volume is excited and captured sequentially. Point scanning significantly reduces the available integration time per pixel but requires high galvanometer scanning speeds. Table 3 below shows the substantial differences in galvanometer line scanning speed and integration time per pixel for roughly equivalent volume imaging speeds between MediSCAPE and point-scanning microscopes. The imaging parameters listed for MediSCAPE are from the first two data sets shown in Image Set 1 as an example.

[0121] [Table 3]

[0122] The following notes apply to Table 3: (1) Scanning parameters are based on static mirror-based scanning of in vivo mouse kidneys. (2) Scanning parameters are based on moving mirror-based scanning of in vivo mouse kidneys. (3) FPS is calculated over yz frames for MediSCAPE and over xy frames for point scanning.

[0123] For a single high-resolution scan of a mouse kidney, MediSCAPE takes 0.79 seconds to acquire an 802 x 615 x 250 xyz pixel volume of two-color autofluorescence contrast. To acquire an equivalent monochromatic volume at the same speed, a point-scanning microscope would need to scan the galvanometer at a linear rate of 195 kHz, a speed unattainable even using a resonant scanner. Furthermore, the integration time per pixel is 6.3 ns, which is close to the fluorescence lifetime of many fluorophores. In contrast, MediSCAPE's scanning mirror only needs to move at 1.27 Hz, resulting in an integration time per pixel of 0.98 ms. This longer integration time of 153,379 times highlights why weak intrinsic contrast can be imaged much more easily with MediSCAPE than with confocal and two-photon imaging, while maintaining reasonable laser power levels and acquisition times (see Hillman et al. for additional supporting models). For moving scans taken at 9.3 VPS, the same degree of difference is seen between the MediSCAPE and point-scan systems in the required galvanometer line scan speed and the resulting integration time per pixel.

[0124] Furthermore, point-scanning microscopes designed for in vivo and bedside use require mirror scanning or physical movement of the probe or tissue to acquire z-stacks. This can be mechanically challenging to implement and prone to motion artifacts in the presence of moving tissue in vivo. MediSCAPE's lateral scanning and simultaneous recording from all depths at once eliminate the need to change the microscope's focal depth during 3D imaging.

[0125] Comparison of autofluorescence contrast between MediSCAPE, confocal, and two-photon To compare the autofluorescence features captured by MediSCAPE with other microscopy techniques, freshly resected sections of mouse colonic mucosa and kidney cortex were imaged sequentially with confocal, two-photon, and MediSCAPE microscopes. Confocal imaging was performed on a Nikon A1R inverted confocal. Two-photon imaging was performed using a galvanometer mirror and a Mai-Tai HP laser. Tissues were kept on ice and hydrated with saline between imaging sessions and imaged within 3 hours of resection.

[0126] Representative images from each tissue were captured, and the imaging parameters used to acquire each volume shown are provided in Table 4. In fresh mouse colonic mucosa, each imaging technique revealed strong punctate green autofluorescence in the epithelial cells lining the Lieberkühn gland crypts and red emission in the lamina propria between the crypt structures. While both MediSCAPE and confocal images were nearly identical, two-photon excitation revealed strong blue emission in the epithelial cells as well as in the fibers surrounding the crypts. In fresh mouse kidney cortex, all three imaging techniques were able to clearly visualize the tubules through autofluorescence, with proximal tubules showing greater emission in the green channel and distal tubules showing greater emission in the red channel. Two-photon imaging also revealed blue autofluorescence in the tubules, with significant overlap with the green channel. Note that the tubules here are imaged deeper within the kidney cortex and appear morphologically different from those closer to the cortical surface, as shown in the MediSCAPE and confocal images. Also note that sample drift was a major issue in confocal and two-photon imaging over the course of the volume acquisition in both tissues.

[0127] Although these data sets were acquired solely to compare the contrast acquired in MediSCAPE with that acquired in confocal and two-photon, the imaging parameters shown in Table 4 indicate that confocal and two-photon imaging require almost twice as long to acquire a monochromatic volume of similar quality as an image acquired with MediSCAPE on the same tissue.

[0128] [Table 4]

[0129] *Voxel rate was calculated as the # of color voxels acquired per second of total imaging time.

[0130] method Preparation and imaging of mouse tissue in vivo In vivo mouse imaging was performed according to a protocol reviewed and approved by Columbia University's Institutional Animal Care and Use Committee. Prior to imaging, wild-type mice were deeply anesthetized using isoflurane, and their snouts were placed in a mouse mask. Body temperature was maintained with a heating pad positioned over the mouse, and breathing was continuously monitored. The abdominal organs were first exposed, and the mouse was placed in a 60-mm diameter glass-bottom dish mounted on a three-axis stage. The organs were positioned against the surface of a glass coverslip for imaging from below. For mobile imaging, the mouse's position was translated between successive imaging sessions. Warm saline was used to periodically flush the tissues to minimize desiccation and maintain body temperature. After imaging the organs, the chest cavity was then opened, and the heart was quickly positioned for imaging before euthanasia.

[0131] To image the brain microvasculature shown in Image Set 7 and Movie 10 (described previously), mice were anesthetized with urethane, and sealed bilateral glass cranial windows were implanted over the somatosensory cortex as previously described. A metal head plate was glued to the skull to allow fixation of the head under the MediSCAPE objective in an upright configuration. Imaging was performed following a tail vein injection of approximately 0.1 ml of 5% w / v 70,000 MW fluorescein isothiocyanate-dextran. The xyz position of the mouse was manually translated by a 3-axis stage during mobile imaging.

[0132] Preparation and imaging of fresh mouse tissue Fresh mouse tissue was excised from wild-type mice according to a protocol reviewed and approved by Columbia University's Institutional Animal Care and Use Committee. Mice were deeply anesthetized and then euthanized using cervical dislocation. The excised tissue was kept on ice until imaging or at least 30 minutes before staining. All data were acquired within 3 hours of excision. Tissues were imaged from below in a 30 mm diameter glass-bottom dish with a MediSCAPE objective in an inverted configuration. The tissue was kept hydrated with saline and gently pressed down with a coverslip when needed to create a flatter imaging surface. In the system shown in Figure 2, tissues were placed in a Petri dish and imaged from above (in an upright configuration). The tissue was kept hydrated with saline and pressed down with a coverslip when needed. For datasets showing stained fresh tissue, tissues were topically stained for 1–3 minutes at room temperature, rinsed with saline, and immediately imaged as previously described.

[0133] Receiving and Imaging a Human Kidney Biopsy Deidentified fresh human kidney tissue was obtained through the tissue bank at the Columbia University Medical Center Department of Pathology under an IRB-approved protocol. Tissues were imaged within 24 hours of resection, stored at 4°C in a Petri dish with a saline-soaked cloth, and kept on ice before imaging. Tissues were imaged from below in a 30 mm diameter glass-bottom dish with saline to maintain moisture.

[0134] Staining of fresh tissue Where indicated, tissues were topically stained with 0.01% proflavine (Sigma, 131105), 1% methylene blue (Ricca, 485016), and / or 0.01% sodium fluorescein in saline. The dye was gently applied to the tissue with a cotton swab for 1-3 minutes at room temperature and then rinsed three times with saline. The stained area was immediately imaged.

[0135] histology After imaging, all fresh tissues were marked with a tissue marker to clearly indicate the surface to be imaged in MediSCAPE and placed in the histology cassette, with the imaged surface lying flat on the biopsy paper. Tissues were fixed in 10% formalin for at least 24 hours at 4°C. Subsequent histological embedding, sectioning, staining, and mounting were performed by Molecular Pathology Histology Services at CUMC Herbert Irving Cancer Center. All mouse tissues were cut horizontally into several 5-μm flat sections extending 50–100 μm from the surface to be imaged and stained for H&E. Kidney biopsy tissues were cut into 2-μm flat sections extending 50–100 μm from the surface to be imaged and stained for H&E and PAS. Histology slides were digitally scanned using a Nikon AZ100 slide scanner. Regions of interest were matched by manual comparison of architectural features visible in the MediSCAPE images and digital histology data.

[0136] conclusion While the present invention has been disclosed with reference to particular embodiments, numerous improvements, modifications, and variations to the described embodiments are possible without departing from the sphere and scope of the invention as defined in the appended claims. It is therefore intended that the invention not be limited to the described embodiments, but that the invention have the full scope defined by the words of the appended claims and their equivalents. [Explanation of symbols]

[0137] 1 telescope 2 Telescope, video 10 First set of optical components 12 Primary objective lens, first objective lens 14 Achromat, Lens Telescope 16 Pressle eyepiece, lens telescope 20 Second set of optical components 22 Achromat 24 Achromat, the second imaging telescope 26 Secondary objective, second imaging telescope, second objective 32 Galvanometer mirror, scanning element 34 Folding mirror, silver mirror 38 Dichroic 42 tertiary objective lens, third objective lens, detection objective lens 45 Homemade Image Splitter 46 Tube Lens 48 sCMOS camera, photodetector array 60 Single-mode fiber 61, 62 cylindrical lenses, cylindrical 4f system 66 Cylindrical Lens 68 Powell Lens 82 Imaging Cap 85 water 88 Coverslips, Cover Glasses, Glass Fronts

Claims

1. a first set of optical components (10) having a proximal end and a distal end, the first set of optical components including a first objective lens (12) disposed at the distal end of the first set of optical components, the first objective lens having a magnification between 10x and 70x and a numerical aperture between 0.5 and 1.1; a second set of optical components (20) having a proximal end and a distal end, the second set of optical components including a second objective lens (26) disposed at the proximal end of the second set of optical components; a scanning element (32) disposed proximally relative to the proximal ends of the first set of optical components and distally relative to the distal ends of the second set of optical components, the scanning element is positioned to route the excitation light in a proximal-to-distal direction through the first set of optical components such that the excitation light is projected onto a sample positioned distal to the distal ends of the first set of optical components; the excitation light projected onto the sample forms a sheet of excitation light at an oblique angle, the position of the sheet varying depending on the orientation of the scanning element; the first set of optical components path the detected light from the sample back to the scanning element in a distal-to-proximal direction; the scanning element is further arranged to path the detection light such that the detection light passes through the second set of optical components in a distal-to-proximal direction, the second set of optical components forming an intermediate image plane at a location proximal to the proximal ends of the second set of optical components. a scanning element; a folding mirror (34) positioned proximally relative to the proximal ends of the first set of optical components and distally relative to the distal ends of the second set of optical components; a photodetector array (48); a third objective lens (42) positioned to direct light arriving from the intermediate image plane towards the photodetector array; Equipped with the folding mirror is a fixed mirror; the folding mirror is disposed between the scanning element and the distal end of the second set of optical components, and the folding mirror is configured to reflect the excitation light to the scanning element such that a path of the excitation light through the first set of optical components is parallel to a path of the excitation light through the second set of optical components; or an imaging device, wherein the folding mirror is disposed between the scanning element and the proximal end of the first set of optical components, and the folding mirror is configured to reflect the detection light to the scanning element such that a path of the detection light through the first set of optical components is parallel to a path of the detection light through the second set of optical components.

2. The apparatus of claim 1 , wherein the folding mirror is positioned between the scanning element and the distal end of the second set of optical components.

3. the first objective lens has a magnification between 50x and 70x, a numerical aperture between 0.9 and 1.1, and an effective focal length between 2.5mm and 3.5mm; 10. The apparatus of claim 1, wherein the second objective lens has a magnification between 40x and 60x, a numerical aperture between 0.65 and 0.85, and an effective focal length between 3 mm and 5 mm.

4. 4. The apparatus of claim 3, wherein the first objective lens has a magnification of 60x, a numerical aperture of 1.0, and an effective focal length of 3 mm.

5. 4. The apparatus of claim 3, wherein the second objective lens has a magnification of 50x, a numerical aperture of 0.75, and an effective focal length of 4 mm.

6. The apparatus of claim 3 , wherein the first set of optical components includes at least one pressl-type eyepiece.

7. 4. The apparatus of claim 3, wherein the first set of optical components comprises a 12.7 mm diameter, 38.1 mm EFL achromat and the Pressl eyepiece comprises two 12.7 mm diameter, 50.8 mm EFL achromats.

8. The apparatus of claim 3 , wherein the second set of optical components includes at least one pressl-type eyepiece.

9. 4. The apparatus of claim 3, wherein the second set of optical components comprises a press-type eyepiece made from two 1-inch diameter, 101.6 mm EFL achromats and a 1-inch diameter, 76.2 mm EFL achromat.

10. 4. The apparatus of claim 3, wherein the first set of optical components comprises a telescope with a 1.5X magnification and the second set of optical components comprises a telescope with a 1.5X magnification.

11. 4. The apparatus of claim 3, wherein the third objective lens has a magnification between 15x and 25x and a numerical aperture between 0.65 and 0.

85.

12. 4. The apparatus of claim 3, wherein the third objective lens has a magnification of 20x and a numerical aperture of 0.

75.

13. The device of claim 1, further comprising an optically transparent spacer positioned and configured to cover the first objective lens and press against the tissue to be imaged.

14. The device described in claim 13, wherein the optically transparent spacer is incorporated into a cap that forms a watertight seal between the optically transparent spacer and the distal end of the first objective lens.

15. The apparatus described in claim 14, further comprising a predetermined amount of medium positioned between the optically transparent spacer and the first objective lens, the medium having a refractive index selected to match the immersion medium of the first objective lens, the predetermined amount of medium optically coupling the optically transparent spacer to the first objective lens, and the cap forming a watertight seal.

Images