1Department of Pathology, Harvard Medical School, 2Massachusetts General Hospital, 3Wellman Center for Photomedicine, Harvard Medical School, 4Pulmonary and Critical Care Unit, Massachusetts General Hospital, 5Pulmonary and Critical Care Unit, Harvard Medical School
Hariri, L. P., Applegate, M. B., Mino-Kenudson, M., Mark, E. J., Bouma, B. E., Tearney, G. J., et al. Optical Frequency Domain Imaging of Ex vivo Pulmonary Resection Specimens: Obtaining One to One Image to Histopathology Correlation. J. Vis. Exp. (71), e3855, doi:10.3791/3855 (2013).
Lung cancer is the leading cause of cancer-related deaths1. Squamous cell and small cell cancers typically arise in association with the conducting airways, whereas adenocarcinomas are typically more peripheral in location. Lung malignancy detection early in the disease process may be difficult due to several limitations: radiological resolution, bronchoscopic limitations in evaluating tissue underlying the airway mucosa and identifying early pathologic changes, and small sample size and/or incomplete sampling in histology biopsies. High resolution imaging modalities, such as optical frequency domain imaging (OFDI), provide non-destructive, large area 3-dimensional views of tissue microstructure to depths approaching 2 mm in real time (Figure 1)2-6. OFDI has been utilized in a variety of applications, including evaluation of coronary artery atherosclerosis6,7 and esophageal intestinal metaplasia and dysplasia6,8-10.
Bronchoscopic OCT/OFDI has been demonstrated as a safe in vivo imaging tool for evaluating the pulmonary airways11-23 (Animation). OCT has been assessed in pulmonary airways16,23 and parenchyma17,22 of animal models and in vivo human airway14,15. OCT imaging of normal airway has demonstrated visualization of airway layering and alveolar attachments, and evaluation of dysplastic lesions has been found useful in distinguishing grades of dysplasia in the bronchial mucosa11,12,20,21. OFDI imaging of bronchial mucosa has been demonstrated in a short bronchial segment (0.8 cm)18. Additionally, volumetric OFDI spanning multiple airway generations in swine and human pulmonary airways in vivo has been described19. Endobronchial OCT/OFDI is typically performed using thin, flexible catheters, which are compatible with standard bronchoscopic access ports. Additionally, OCT and OFDI needle-based probes have recently been developed, which may be used to image regions of the lung beyond the airway wall or pleural surface17.
While OCT/OFDI has been utilized and demonstrated as feasible for in vivo pulmonary imaging, no studies with precisely matched one-to-one OFDI:histology have been performed. Therefore, specific imaging criteria for various pulmonary pathologies have yet to be developed. Histopathological counterparts obtained in vivo consist of only small biopsy fragments, which are difficult to correlate with large OFDI datasets. Additionally, they do not provide the comprehensive histology needed for registration with large volume OFDI. As a result, specific imaging features of pulmonary pathology cannot be developed in the in vivo setting. Precisely matched, one-to-one OFDI and histology correlation is vital to accurately evaluate features seen in OFDI against histology as a gold standard in order to derive specific image interpretation criteria for pulmonary neoplasms and other pulmonary pathologies. Once specific imaging criteria have been developed and validated ex vivo with matched one-to-one histology, the criteria may then be applied to in vivo imaging studies. Here, we present a method for precise, one to one correlation between high resolution optical imaging and histology in ex vivo lung resection specimens. Throughout this manuscript, we describe the techniques used to match OFDI images to histology. However, this method is not specific to OFDI and can be used to obtain histology-registered images for any optical imaging technique. We performed airway centered OFDI with a specialized custom built bronchoscopic 2.4 French (0.8 mm diameter) catheter. Tissue samples were marked with tissue dye, visible in both OFDI and histology. Careful orientation procedures were used to precisely correlate imaging and histological sampling locations. The techniques outlined in this manuscript were used to conduct the first demonstration of volumetric OFDI with precise correlation to tissue-based diagnosis for evaluating pulmonary pathology24. This straightforward, effective technique may be extended to other tissue types to provide precise imaging to histology correlation needed to determine fine imaging features of both normal and diseased tissues.
1. Imaging System
The technical details of OFDI have been described previously4-6. Circumferential OFDI was conducted at imaging speeds between 25 and 100 frames per second and between 512 and 2,048 axial depth profiles per circular cross-sectional image. Custom 2.4 Fr (0.8 mm diameter) helical scanning catheters used in this study were designed to operate through the access port of standard bronchoscopes. The catheters consisted of an inner optical core to focus the light onto the bronchial wall and a single-use outer sheath. The catheter body remained stationary during imaging while the inner core was rotated at a rate between 25 and 100 Hz and translated at a pullback speed of between 1.25 and 5 mm/sec. The axial resolution of the system was 6 mm in tissue and provided an image ranging depth of 7.3 mm4-6. Catheter-based OFDI was performed in this study to replicate in vivo bronchoscopic OFDI (Figure 1). However, this protocol may also be applied to imaging with a bench-top optical system (Figure 3 and 4).
2. Imaging System Set-up
3. Tissue Preparation
4. Tissue Marking
5. Imaging Tissue
6. Collecting and Processing Tissue
7. Image Processing
If images were acquired with a benchtop scanner, or other scanning technique where both ink marks were visible in a single cross-sectional image, then the image can be directly correlated with corresponding histology. If volumetric datasets were acquired with a helical scanning catheter, the images will need to be re-interpolated so that a single 2D image bisects both ink marks for correlation with histology. This can be accomplished using ImageJ or other image processing software. In some instances, the ink may not be readily visible in which case adjacent sections/slides should be examined.
The black ink marks should be between 1 - 1.5 cm apart to indicate the imaging region of interest. The green ink mark should be placed at the beginning of the imaging scan, before the first black ink mark to orient the specimen (Figure 2 and Figure 3a). Tissue ink marks should be visible on both OFDI imaging and histology (Figure 3 and 4). In normal swine (Figure 3) and human airway (Figure 4), typical airway layering should be visible. The epithelium (E) is visible as a thin, moderately signal dense, homogeneous layer at the luminal aspect of the airway. The lamina propria consists of organized signal-intense to signal-poor tissue, corresponding to various components of the lamina propria (LP) such as signal intense connective tissues including elastin and collagen (EL), and signal poor salivary-type glandular tissue (G). There are occasionally visible signal poor ducts (D) traversing the respiratory epithelium to connect with the bronchial lumen. Smooth muscle appears as discontinuous, interspersed smooth muscle fascicles and is thus not identifiable in OFDI. On the H&E and trichrome stains, airway layering can be visualized (Figure 3c, 3d, 3f, 3g, 4b, 4c, 4e, and 4f), where on trichrome the superficial dense elastic and collagenous tissues appear deep blue and the underlying smooth muscle stains red (SM). Cartilage rings (C) appear as signal poor crescent-shaped structures with well-defined boundaries, which overlap in the swine airway and do not overlap in the human airway. The perichondrium surrounding the cartilage rings appears as a thin layer of signal intense tissue encompassing the signal poor cartilage rings. In the peripheral human airways (Figure 4g and 4h), alveolar attachments (A) are visible as thin, signal intense lattice-like alveolar walls with signal void alveolar spaces. Vascular spaces within the lamina propria are visible as signal void linear or circular structures with mild underlying shadowing artifact (arrows).
Figure 1. OFDI of swine airway. In vivo images obtained from a swine airway under mechanical ventilation. (a) ODFI cross-section of proximal airway. (b) OFDI cross-section of distal airway. (c) ODFI longitudinal section of proximal airway, higher magnification image of panel e in red highlighted region. (d) OFDI longitudinal section of distal airway, higher magnification image of panel e in green highlighted region. (e) ODFI longitudinal section of airway from proximal to distal (left to right). Catheter diameter is 0.8 mm and tick marks represent 0.5 mm increments. Although different layers of the airway wall and alveolar attachments are discernible in the OFDI images, it is difficult to precisely interpret the anatomic correlate of the OFDI signals without directly registered histology. e: epithelium, lp: lamina propria, sm: submucosa, c: cartilage, a: alveolar attachments.
Figure 2. Tissue marking of swine airway. (a) Opened airway with two black ink marks on the luminal surface placed parallel to the longitudinal aspect of the airway, 1.5 cm apart. (b) OFDI catheter placed over two black ink marks to include both marks within the OFDI pullback. (c) Airway with additional green ink mark to orient the beginning of the imaging scan on the specimen.
Figure 3. OFDI and histology of swine airway demonstrating precise correlation using tissue marking. (a) Opened airway with two black ink marks on the luminal surface placed parallel to the circumferential aspect of the airway. Pins are used to further open the airway (arrows). (b) OFDI of swine airway with both ink marks visible (asterisks) with (c) precisely correlated histology stained with H&E (asterisks: black ink marks visible on respiratory epithelium) and (d) correlated trichrome stain. Scale bar: 2 mm. (e) Higher magnification view of OFDI image with (f) corresponding histology stained with H&E and (g) correlated trichrome stain. E: respiratory epithelium, EL: dense collagen and elastic tissues, SM: smooth muscle, C: cartilage rings (histological artifact has resulted in artificial separation of the cartilage rings), G: salivary gland tissue, D: salivary duct entering epithelium. Scale bar: 250 μm. Click here to view larger figure.
Figure 4. OFDI and histology of human airway demonstrating precise correlation using tissue marking. (a) OFDI of human proximal airway with both ink marks visible (asterisks). (b) Precisely correlated histology stained with H&E with black ink marks visible on respiratory epithelium (asterisks) and (c) correlated trichrome stain. Scale bar: 2 mm. (d) Higher magnification view of OFDI image and (e) corresponding histology stained with H&E and (f) trichrome. Scale bar: 250 μm. E: respiratory epithelium, LP: lamina propria, G: salivary gland tissue, C: cartilage rings, PC: perichondrium. In the human airway, typical layering is visible. Within the loose connective tissue, there are interspersed fascicles of red-staining smooth muscle (SM, panels c and f), which do not form a continuous band and thus are not visible as a distinct layer in OFDI. (g) OFDI of human distal airway and (h) precisely correlated H&E histology with black ink marks visible on respiratory epithelium (asterisks). Scale bar: 2 mm. Alveolar attachments (A) are visible as signal intense lattice-like alveolar walls with signal void alveolar spaces. Vascular spaces within the lamina propria are also visible as signal-void structures with underlying mild shadowing (arrows).
Assessment of early lung malignancies can be extremely challenging due to lack of symptoms and the inability to visualize early neoplastic changes radiologically or bronchoscopically. OFDI provides near histologic resolution, large area 3-dimensional views of tissue microstructure in real time2-6. Endobronchial OFDI has been demonstrated in patients as a safe technique that can be used to obtain high-resolution volumetric datasets over long airway segments in the pulmonary airways11-13 (Animation). However, only small biopsies are obtained as histopathological counterparts in the in vivo setting, which do not provide adequate correlates to OFDI for the development of imaging criteria for pulmonary pathology. In order to accurately assess OFDI features seen in pulmonary imaging, it is essential to obtain precisely matched image to histology correlations. We present a simple and effective method for precise, one to one correlation between OFDI and histology applied to airway imaging of ex vivo pulmonary resection specimens, which is applicable to nearly any ex vivo tissue type. Once imaging criteria have been established ex vivo with matched one-to-one histology, these criteria can then be applied to in vivo imaging.
The tissue dye used to mark the imaging region of interest is clearly visible in both OFDI and histology. By using straightforward techniques to orient the tissue, ink marks may be correlated in both imaging and histology to allow one to one comparisons of OFDI features and histology findings to determine the identifiable imaging characteristics of the tissue pathology. The technique is inexpensive and practical, thus making it useful in many optical imaging applications.
In the in vivo setting, methods such as laser marking may be used for tissue orientation25. However, the small size of the bronchial biopsy is still a limiting factor in using in vivo studies to develop specific imaging criteria for pulmonary pathology. Although ex vivo studies serve as an adequate alternate to in vivo imaging, there are some limitations. Ex vivo lung specimens are uninflated and often display surgically-induced atelectasis, which alters the appearance of normal alveolar structures. Inflating surgically resected lung tissue with tissue marking for histology correlation is technically challenging as most surgical lung specimens are received after pathology frozen section evaluation during which the pleural surface is disrupted, interfering with specimen inflation. Non-pathologic atelectasis is not an artifact seen in the in vivo setting, thus this limitation would not be pertinent to in vivo pulmonary imaging. Additionally, lack of blood within vessels in ex vivo specimens could make it difficult to distinguish vascular structures from other signal void structures. In the in vivo setting, the addition of Doppler OCT/OFDI26-28 to structural OCT/OFDI would aid in the identification of vessels.
Motion artifacts may be seen in vivo where they are not present ex vivo. This could be potentially problematic in standard OCT systems with slower acquisition rates. However, the rapid frame rates of OFDI systems are currently > 200 fps29-31. Thus, it is not expected that motion artifact will be a significant issue. Previous in vivo OCT and OFDI imaging studies have demonstrated successful visualization of fine imaging features14,15,18,19.
In this study we have demonstrated volumetric OFDI with precise correlation to tissue-based diagnosis for evaluating pulmonary pathology. The procedure described is intended to provide precisely matched histology to be used as gold standard for the OFDI image interpretation.
Once specific imaging criteria for pulmonary pathology have been developed and validated ex vivo with matched one-to-one histology, the criteria may then be applied to subsequent in vivo imaging studies with the use of a bronchial biopsy as a gold standard assessment of the imaging features seen. This technique is presented as an application to pulmonary resection specimens, but can be applied to almost any tissue type to provide the precise imaging to histology correlation needed to determine fine imaging features of both normal and pathological tissues.
Production and Free Access to this article is sponsored by NinePoint Medical Inc.
The authors would like to thank Mr. Sven Holder and Mr. Stephen Conley for their invaluable assistance in this study. This work was funded in part by the National Institute of Heath [Grant number R00CA134920] and the American Lung Association [Grant number RG-194681-N]. NinePoint Medical Inc. sponsored the publication costs associated with this manuscript.
|Tissue marking dye||Triangle Biomedical||TMD-BK, TMD-G|