Histology Of Tissues Pdf

[Yoshi Heffernan]

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Histological images are critical in the diagnosis and treatment of cancers. Unfortunately, current methods for capturing these microscopy images require resource intensive tissue preparation that may delay diagnosis for days or weeks. To streamline this process, clinicians are limited to assessing small macroscopically representative subsets of tissues. Here, a combined photoacoustic remote sensing (PARS) microscope and swept source optical coherence tomography system designed to circumvent these diagnostic limitations is presented. The proposed multimodal microscope provides label-free three-dimensional depth resolved virtual histology visualizations, capturing nuclear and extranuclear tissue morphology directly on thick unprocessed specimens. The capabilities of the proposed method are demonstrated directly in unprocessed formalin fixed resected tissues. The first images of nuclear contrast in resected human tissues, and the first three-dimensional visualization of subsurface nuclear morphology in resected Rattus tissues, captured with a non-contact photoacoustic system are presented here. Moreover, the proposed system captures the first co-registered OCT and PARS images enabling direct histological assessment of unprocessed tissues. This work represents a vital step towards the development of a rapid histological imaging modality to circumvent the limitations of current histopathology techniques.

Ideally, three-dimensional (3D) nuclear and extranuclear-tissue morphology could be visualized label-free directly on unprocessed resected samples. This would avoid the delays of tissue preparation, enabling near real-time histological analysis. Moreover, direct imaging of resected tissues would facilitate more thorough analysis of excised specimens, rather than a restricted subset of tissues. This would permit the most highly informative tissues to be submitted, where required, for further immunohistochemical analysis. A technology capable of imaging in this fashion could revolutionize the workflow of anatomic pathology.

Although several novel histological imaging techniques have emerged, no single technology provides label-free 3D histological imaging of bulk unprocessed resected tissue samples. Most technologies in this histological imaging space require exogenous dyes or tissue clearing to image within specimens. These techniques include microscopy with ultraviolet surface excitation (MUSE)5, fluorescence microscopy6,7, non-linear microscopy8,9, and light sheet microscopy10,11. Objectively, there are few techniques which have achieved measurable success providing label-free histology like contrast in resected tissues. The most notable are stimulated Raman scattering microscopy12,13, photoacoustic (PA) microscopy14,15,16,17,18,19 and optical coherence tomography (OCT)20,21,22,23,24,25. Stimulated Raman scattering microscopy has recently provided histological visualizations in resected tissues12,13. However, this was applied in a transmission mode architecture only compatible with thin translucent tissue sections12,13. Therefore, PA microscopy and OCT are the most promising means to achieve label-free imaging of thick tissue samples.

PA microscopy modalities offer distinct advantages as they capture optical absorption contrast. PA offers chromophore specific visualizations by leveraging the unique optical absorption spectra of biomolecules. This has been applied to selectively image DNA/RNA, hemoglobin, melanin, lipids, collagen, and more14,17,27,28,29,30,31. These unique chromophore specific visualizations position PA as a potentially powerful tool for label-free histological imaging14,15. The main challenge hampering the clinical adoption of this technique is that PA microscopy is a hybrid opto-acoustic modality. Conventional PA systems require physical contact with samples to perform imaging. However, a revolutionary new non-contact PA modality, Photoacoustic Remote Sensing (PARS) microscopy, has emerged as a frontrunner in label-free imaging15,16,18,19,30,32. Unlike traditional photoacoustic systems, PARS uses an all-optical pump-probe architecture. In PARS, the pump generates photoacoustic signals, which the probe captures as back-reflected intensity modulations30. Previously this technique was successfully applied to histology-like imaging in unprocessed resected tissue specimens, thin tissue sections, and paraffin embedded tissue blocks15,16,18,19,33. In these works, PARS provided histological visualizations analogous to hematoxylin staining of cell nuclei by leveraging the UV absorption contrast of DNA15,16,18,19,33. More recently, PARS hematoxylin-like imaging has been extended to full H&E emulation15,19,34. These methods leveraged endogenous absorption15,19 and scattering contrast34 to capture extra-nuclear morphology to accompany the PARS nuclear visualizations. However, these methods cannot match the single acquisition volumetric imaging of extra-nuclear tissue structures provided by OCT.

This paper presents a conjoined PARS microscope and swept source OCT (SS-OCT) system for rapid 3D virtual histology in bulk unprocessed resected tissue specimens. The UV excitation PARS microscope is used to provide chromophore specific recovery of nuclear morphology. Concurrently, the SS-OCT is used to capture 3D volumetric images of tissue morphology. The co-registered PARS and OCT data are then merged to provide three-dimensional, histological, and structural tissue visualizations. While a combined PARS and OCT system has previously been proposed for ocular and vascular imaging35,36, this is the first such system for histopathology. Though, since the submission of this article, Martell et al. have presented a PARS-OCT histology system37. However, their technique does not provide co-registered PARS and OCT imaging or three-dimensional histology37. Applying the SS-OCT in resected Rattus tissues volumetric images containing adipocytes, ducts, fascia layers and other tissue features are recovered. Applying the PARS system in unprocessed resected human and Rattus tissues for the first time presents novel high spatial resolution imaging of predominately nuclear structures. Within the PARS visualizations tissue boundaries, and nuclear atypia, arrangement, organization, and density may be assessed. Moreover, the optical sectioning capabilities of the PARS microscope are leveraged to capture the first 3D volumetric images of subsurface nuclear morphology in unprocessed tissues. Together this system enables label-free non-contact assessment of nuclear structures directly on 3D visualizations of resected tissue specimens. These visualizations provide qualitatively similar diagnostic features to conventional H&E preparations. If employed in a clinical setting, the proposed technique could allow direct histology-like imaging of unprocessed resected tissue samples. Thus, the PARS-OCT system holds the potential to dramatically reduce current limitations in the histopathological processing of tissues. This could provide near-real time histological analysis, circumvent current diagnostic limitations, and greatly improve patient outcomes.

For the PARS system (Fig. 1), UV excitation was provided by a 266 nm 400 ps 50 kHz pulsed laser (WEDGE XF, Bright Solutions). The 266 nm excitation was separated from residual 532 nm output using a CAF2 prism. Once isolated, the UV excitation beam was expanded and combined with the probe beam. The 1310 nm probe was supplied by a continuous wave super-luminescent diode laser from Thorlabs (S5FC1018P, Thorlabs). The horizontally polarized probe beam was passed through a polarizing beam splitter and quarter wave plate into the imaging system. The combined excitation and probe beam were then focused onto the sample using a 0.5 NA reflective objective (LMM-15X-UVV, Thorlabs). The back-reflected probe beam containing the PARS signals returned to the quarter wave plate and polarizing beam splitter by the same path as forward propagation. Upon passing through the quarter wave plate a second time, the returning probe beam became horizontally polarized. The horizontally polarized probe beam was redirected down the detection pathway by the polarizing beam splitter. Here, the probe beam was isolated through spectral filtering, then focused onto a photodiode (PDB425C-AC, Thorlabs).

Combined PARS and OCT visualizations were generated using the final PARS images, and processed OCT volumetric data. Prior to processing, the PARS images were overlayed over the max amplitude projection of the OCT volume to manually tune the co-registration, as there were some minor motion artifacts. Since the PARS images were substantially higher resolution than the OCT frames, the datasets had to be rescaled prior to combination. For the large area scans, the PARS image was compressed to match the OCT volume. Conversely, for the high-resolution frames, the OCT volume was interpolated to match the PARS image dimensions. Merging the PARS data into the OCT volume was performed one B-scan at a time. In each B-scan, the tissue surface was extracted based on the regional maximum of the OCT signal. The corresponding stripe of PARS contrast was then merged into the OCT B-scan along the extracted surface. The entire volumetric dataset was processed in this fashion. Volumetric and enface images were generated from the resulting 3D data sets with ImageJ38.

Tissue samples were collected under protocols approved by the Research Ethics Board of Alberta (Protocol ID: HREBA.CC-18-0277) and the University of Waterloo Health Research Ethics Committee (Humans: #40275, and Animals: #41543). During this study, informed consent was waived by the Research Ethics Board of Alberta and the University of Waterloo Health Research Ethics Committee as samples were archival tissue not required for diagnostic purposes, and no patient identifiers were provided to the researchers.

c80f0f1006