Open Access
Add to BibSonomy
Abstract
Tissue fibrosis is a progressive and destructive disease process that can occur in many different organs including the liver, kidney, skin, and lungs. Fibrosis is typically initiated by inflammation as a result of chronic insults such as infection, chemicals and autoimmune diseases. Current approaches to examine organ fibrosis are limited to radiological and histological analyses. Infrared spectroscopic imaging offers a potential alternative approach to gain insight into biochemical changes associated with fibrosis progression. In this study, we demonstrate that IR imaging of a mouse model of pulmonary fibrosis can identify biochemical changes observed with fibrosis progression and the beginning of resolution using K-means analysis, spectral ratios and multivariate data analysis. This study demonstrates that IR imaging may be a useful approach to understand the biochemical events associated with fibrosis initiation, progression and resolution for both the clinical setting and for assessing novel anti-fibrotic drugs in a model system.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Fibrosis-related diseases are believed to contribute to about 45% of deaths worldwide [1,2]. Fibrosis is characterized by the scarring and stiffening of tissue which is due to abnormal deposition of extracellular matrix. This excessive scarring leads to tissue destruction and ultimately organ dysfunction [3]. Fibrosis can be triggered as a result of both acute and chronic inflammation within many tissues, such as, lung, kidney, liver, skin and heart. The inflammation can be triggered by a variety of stimuli including persistent infections, autoimmune reactions, allergic responses, chemical insults, radiation, and other types of tissue injury [4].
One of the main challenges in the management of fibrotic diseases is the early and accurate diagnosis of the pathologic processes involved. Unfortunately, the current diagnostic approaches for fibrosis are complex and suboptimal and require both radiologic assessment and histologic evaluation, and therefore depends on a multidisciplinary discussion between a physician, radiologist, and pathologist [5]. Radiologic assessment offers the initial detection of fibrosis based on several radiological features [6–10], but the diagnosis must be validated in cases where the clinical and imaging information are insufficient to exclude other causes of disease in that particular organ. This validation is frequently accomplished by a biopsy followed by histopathological analysis to demonstrate characteristic features and exclude alternative diagnoses [11,12]. But this technique, which consists of chemical staining and labeling to enhance contrast in light microscopy, requires the subjective evaluation by a pathologist to determine the percentage and grade of fibrosis [13]. In addition, staining is limited to morphological information and specific epitopes based on the stain used and does not provide comprehensive deep biochemical information about the changes that are occurring within tissues. Furthermore, the currently available methodologies for assessing fibrosis do not provide any prognostic information about the potential for further fibrosis development. There is also a growing need to develop rapid imaging approaches to assess fibrosis in model systems to test the effects of novel anti-fibrotic treatments. Label-free spectroscopic imaging approaches are emerging as a quantitative, automated means of assessing the chemical and morphological changes of tissue samples.
Mid-infrared spectroscopic imaging (IR) is an approach that assesses the chemical status of a tissue without the need for exogenous labels [14] and is non-destructive to the tissue. This method combines the morphological information gleaned from microscopy with a biochemical profile derived from the sample’s infrared absorption spectrum, and offers a means to selectively analyze the chemical compositions of different features of diseased tissues [15]. It is based on the principle that chemical bonds (such as C-H, N-H, O-H, P-O) absorb different frequencies of mid-infrared light that allows for the measurement of key cellular and tissue biomarkers within tissues such as proteins, lipids, collagen, or glycosylation. The alterations of these biomolecules are often a key part of disease processes and thus can be probed and characterized using IR imaging.
In IR imaging, a spectrum is acquired at every pixel within a scan; therefore, while visible images are typically comprised of only three channels (red, green and blue), in IR each pixel is composed of hundreds of biochemical channels. These information-rich images which give the “biochemical signature” of the tissue can then be used towards cell type classification, disease diagnosis or understanding disease progression. There have been significant technological and computational advances over the past few years that have enabled mid-IR imaging technologies to be more amenable for the assessment of fibrosis such as: significant increases in imaging speed due to multi-element detectors; improved signal to noise images due to new quantum cascade laser based technologies; improved optics allowing for high-resolution diffraction-limited measurements; and new computational tools for data mining and visualization of large hyperspectral data sets [16–21].
IR spectroscopy in both point and imaging format has been widely investigated for its uses in tissue imaging, biofluids, exfoliative cytology, and tissue engineering [16–18,22–25]. Lung tissue has been studied using spectroscopic approaches, for example, developing Raman and FT-IR methodology on normal lung tissues [26,27], Raman spectroscopy for lung cancer detection [28,29], FT-IR spectroscopy for lung cancer detection [30–34], FT-IR of sputum for lung cancer diagnosis [35], and studying lung cancer cell invasion in a model system [36]. In the cancer research community, the paradigm has shifted from focusing on the cancer cells/epithelial cells themselves to the adjacent stromal and fibrotic regions to probe for diagnostic and prognostic biomarkers [37]. As such, there has been a renewed focus on the application of IR imaging to interrogate diseases of connective tissue and on fibrotic processes in a variety of organ systems with a recent review discussing applications of IR imaging in fibrosis [38]. There has also been interest in whether prognostic biomarkers from fibrotic areas could be identified [39,40] and using model tissue engineered systems to identify fibrotic changes [41–43]. These studies indicate how IR imaging could be an indispensable tool in the diagnosis and therapeutic interventions of fibrosis by having the ability to clearly distinguish the changes in pathophysiology across different areas in the same tissue; predict the rate progression and possible cause of the fibrosis. Furthermore, there is a growing need for the rapid interrogation of fibrosis model systems to allow for quantification of fibrosis and derive additional biochemical information for the development of anti-fibrotic drug testing.
One of the leading causes of fibrosis-related deaths are due to pulmonary fibrosis. This study investigates using IR imaging to detect biochemical alterations occurring in lung fibrosis in preclinical models of pulmonary fibrosis. We employed a well-accepted murine model of bleomycin-induced pulmonary fibrosis [44], creating a disease course with rapid onset of alveolar damage (7 days) and progressive fibrosis (14 days to 21 days) with the beginning of resolution within one month (28 days) [45]. IR data from lung sections at weekly timepoints from induction to the start of resolution were compared to conventional histopathological evaluation and investigated with ratiometric and multivariate analysis techniques to search for biochemical markers of disease onset and progression.
2. Materials and methods
2.1 Bleomycin induced pulmonary fibrosis in mice
All animal experiments were carried in accordance with the procedures approved by the Institutional Animal Care and Use committee at the University of Illinois at Chicago, called the Animal Care Committee (Protocol #15-240). The bleomycin-induced rodent model for pulmonary fibrosis was developed as per the review of various animal models by Moore et al., [24]. Mice were anesthetized with a mixture of ketamine (100 mg/kg) and xylazine (5 mg/kg) for bleomycin instillation. The rodent was placed at 45° angle on a platform hanging by its incisors. The tongue was pulled out gently and held to the side using forceps, and 1.5 U/kg of body weight of bleomycin sulfate (Hospira Inc., Lake Forest, IL, USA) was given to the mouse by intratracheal injection with a maximum volume of 50 μL. The mouse was then taken out of the platform and placed under a heating lamp for a few minutes, before returning it to the cage. Based on the time point of harvest post-bleomycin challenge, the animals (N=3 in each group) were grouped as 1. Day 0 (control); 2. Day 7 (inflammation); 3. Day 14 (onset of fibrosis); 4. Day 21 (fibrosis); and 5. Day 28 (recovery phase) [46]. At each time point, the sacrificed animals were euthanized and the lungs were harvested, fixed in formalin, and embedded in paraffin.
Formalin-fixed paraffin embedded (FFPE) lung tissues were sectioned by a microtome at 4 μm thickness onto low-emissivity glass slides (Kevley Technologies, Chesterland, Ohio), suitable for transflectance-mode infrared imaging. FFPE tissues were used for this study, as while there is loss of some biomolecular constituents (such as lipids) during sample preparation, FFPE samples are the most commonly used format in anatomical pathology. In addition, FFPE tissues typically have less sectioning artefacts resulting from sample preparation. Slides were then dewaxed prior to IR imaging using serial washes with xylene [19]. Adjacent tissues sections were cut onto regular glass slides and stained with hematoxylin and eosin (H&E) and Masson’s trichrome by the University of Illinois at Chicago Research Histology and Tissue Imaging Core (RHTIC). The stained sections were later scanned using the Leica Aperio Scanscope CS (Buffalo Grove, IL) and were examined by a pathologist (A.K.B) to identify the key areas in the lung samples and in particular the regions of collagen deposition, based on trichrome staining.
2.2 Infrared spectral imaging
Infrared hyperspectral images of the unstained lung tissues on IR reflective slides were acquired in transflectance mode using a Cary 600 Series FT-IR spectrometer coupled with a Cary 600 Series microscope with a 128 × 128 focal plane array detector and 15X objectives (Agilent Technologies, Santa Clara, California). Images of an entire mouse lung took approximately two hours to acquire. Scanning was performed in the spectral range of 900 cm-1 to 4000 cm-1, at a spectral resolution of 4 cm-1, and taking the average of 8 co-scans to acquire the spectrum with a high signal-to-noise ratio.
2.3 Spectral data preprocessing
The FT-IR scan data was processed using the ENVI + IDL software package (L3Harris Geospatial, Broomfield, CO). Multipoint linear baseline correction was performed on all of the data, and a threshold filter based on absorbance at the 1650 cm-1 (Amide I) peak was applied to exclude non-tissue pixels from analysis. The spectra were then normalized to their absorbance at 1650 cm-1 (Amide I) to correct for differences in the thickness of the tissue and concentration across all the samples.
2.4 K-means cluster analysis
Using MATLAB (MathWorks, Natick, MA), the IR images were subjected to K-means cluster analysis using the Euclidean distance metric to search for natural divisions or segmentation of the tissue on the basis of the spectral data. The K-means analysis was attempted with different values of k (or numbers of clusters/partitions), and the results were compared to the light microscopy images of the adjacent H&E and trichrome tissue sections to see if the division appeared to correspond to any anatomical or pathological features of the lung. K-means cluster analysis was conducted on the pooled dataset of normalized, baseline-corrected IR images of all samples simultaneously. Six clusters were used, and all images in this paper are from this K-means analysis.
2.5 Spectral analysis of lung fibrotic regions
To compare spectral changes over time associated with fibrosis development, regions of interest (ROIs) were manually drawn in the fibrotic areas of all of the lung samples, guided by the adjacent H&E and trichrome sections. As control samples at day 0 did not have active fibrosis, areas of lung interstitium were used to select ROIs. Ten ROIs were drawn for each sample so as to obtain a representation of the intrasample heterogeneity present in the tissues during analysis. Averaged IR spectra in the fingerprint region (900–1800 cm−1) were then extracted for each of the ROIs (about 10,000 pixels). These extracted, pre-processed spectra were then subjected to spectral ratio analysis and multivariate analysis.
Multiple peak absorbance ratios previously described in the literature were computed from the processed spectra. These included ratios associated with collagen (1232 cm-1/1336 cm-1) [25], glycosylation (1030 cm-1/1080 cm-1) [26], and protein conformational changes (1654 cm-1/1554 cm-1) [27]. Principal component analysis (PCA) and linear discriminant analysis (LDA) were performed using SPSS-24 (SPSS Inc., Chicago, Illinois) and the associated plots were made in Origin (OriginLab Corporation, Northampton, MA).
3. Results and discussion
3.1 K-means analysis
IR images of lung tissues were acquired from tissues harvested at days 0, 7, 14, 21 and 28 days after bleomycin treatment. Serial tissue sections from each sample were acquired and stained with H&E and trichrome stains. H&E is the most common stain used by pathologists due to its ability to visualize the major cell types and components of tissues. The trichrome stain is particularly useful for examining fibrosis in tissues due to it staining keratin a dark red color (a major component of epithelial cells) and collagen a bright blue color (a major component of fibrosis) (Fig. 1).
Fig. 1. Array of representative tissue sections examined by infrared microscopy (a)-(e) Brightfield images of the parallel sections stained with H&E (f)-(j) Brightfield images of parallel tissue sections stained using Masson’s trichrome. (k)-(o) Infrared observation of the parallel section from the tissue based on K-means clustering image (6 clusters) calculated from a full band (900–1800cm−1) infrared absorbance dataset.).
IR images were subjected to K-means analysis (Fig. 2), and spectra obtained from regions of fibrosis were compared (Fig. 3), had spectral metrics computed (Fig. 4), and were subjected to multivariate data analysis (Fig. 5). The data-rich hyperspectral data cubes obtained from IR imaging allows for probing of the biochemical changes in lung tissue in response to fibrosis development. This involved examining how biochemical spatial patterns change (K-means analysis) and how the biochemistry of regions of interest change (spectral comparisons, spectral ratios, and multivariate analysis) at different disease timepoints.
Fig. 2. K-means cluster analysis depicting the histopathological regions of each lung section across different groups (a) K-means clustering image (6 clusters) of the lung tissues from each group calculated from a full band (900–1800cm−1) infrared absorbance dataset. Each image is from a different mouse (c) Mean absorption spectra calculated from each cluster, which is quantified as seen in (b).
Fig. 3. Spectral data were extracted from the fibrotic regions of the IR image scans. The average spectra for different stages of fibrosis in the fibrotic regions of the lung tissue showing a significant variation in certain points of the IR spectra.
Fig. 4. Ratiometric analysis of the IR spectra by time point. The ratio of intensities at (a) 1232/1336 gives the collagen map in the tissue for the fibrotic regions. (b) Glycosylation patterns across the lung tissue are interpreted using the 1080/1030 spectral ratio in the fibrotic regions. (c) The spectral ratio of 1654/1554 in fibrotic areas indicates not only any changes in the structural rearrangements of the existing proteins, but also the expression of a new proteins with varied structural characteristics.
Fig. 5. PCA-LDA analysis could classify the different groups with varying stages of fibrosis based on the time points from bleomycin treatment with control (black), day 7 (red), day 14 (green), day 21 (blue), and day 28 (magenta). Different symbols are used to identify the three mice per time-point and each ROI is an individual symbol.
K-means clustering is a multivariate data analysis approach that can be applied to hyperspectral data. This technique assesses how the pixels within an IR image cluster together based on spectral and thus biochemical similarity, by selecting number of clusters to be identified which can adjust the coarseness of the clustered image. This approach allows for the recognition of biospatial patterns within the data that may hold additional diagnostic and prognostic information that would not be available by interpretation of the extracted IR spectra alone [18,47]. In addition, this approach can identify biochemical spatial patterns within the data that may not be identifiable using conventional staining approaches.
K-means analysis ultimately separated the pixels from all of the samples into clusters or groups on the basis of their spectral information, representative of their biochemical state. Six clusters were chosen as it allowed for the complexity of the tissue to be visualized while demonstrating patterns across the data. Light microscopy examination of the H&E and trichrome sections from each time point suggested that this spectral segmentation corresponded to features of normal lung parenchyma, inflammation, and fibrosis from the samples across the five timepoints (Fig. 1).
Figures 2(a) and 2(b) illustrate how the different clusters were distributed across all samples from all timepoints. One of the clusters (cluster 2, green) predominated in the control samples, and was present in the background of other samples, especially at the Day 28 timepoint, at which point recovery from the insult is expected; this particular cluster could represent areas with the chemical composition of normal or healthy non-fibrotic lung tissue. By the day 7 timepoint, this original cluster diminished in prominence, and was replaced by other clusters. This change may be taken to correspond to the inflammatory changes observed at this stage of the bleomycin model, namely hypercellularity of interstitium due to increased type I pneumocytes and infiltration by polymorphonuclear leukocytes.
In the Day 14 samples, the original cluster 2 (potentially healthy tissue) is further diminished, and replaced largely by an increase in another cluster (cluster 1, red). Histologically, this stage shows an increased presence of macrophages and chronic inflammatory cells. By Day 21, there is a substantial increase in a new cluster (cluster 5, cyan), which may be taken to be a spectroscopic correlate to the increased collagen deposition evident by histology. Finally, by Day 28, the new cluster 5 falls off and there is a resurgence of the original cluster 2, while there is some persistence of the clusters that predominated during the early phases of damage (cluster 1, red; cluster 4, yellow). This may indicate the recovery phase of the model, in which fibrosis is healed, but there is still diffuse lymphocytic clustering in the interstitium and macrophages encompassing alveolar cells.
While a perfect assignment of each cluster to a particular tissue feature or cell type was not expected, because they are measures of different biological changes, the results of the K-means analysis suggest that the spectroscopic data collected through IR detect the pathological changes occurring at different phases of the bleomycin damage model that manifests as changes in the biochemical patterning. Comparison of the K-means maps to the H&E and trichrome stained adjacent sections indicated a good agreement between these techniques as to the presence and distribution of changes across the lungs over time.
In summary, these results offer evidence that IR imaging is sensitive to the pathologic processes at work in this model; Fig. 2(c) shows that each of the clusters has a different spectral profile, which in turn indicates a different biochemical composition stemming from different cellular components and metabolic processes taking place.
3.2 Spectral analysis of fibrotic regions of lung tissues
The IR spectra extracted from ROIs in the fibrotic regions of all of the samples for each time point (total of 10 fibrotic regions per lung, 3 mice per timepoint) are presented in Fig. 3. Variations in absorbance at different bands were seen from fibrotic regions between the different timepoints (Fig. 3).
Ratiometric analysis on spectra for each mouse lung (3 mice per timepoint) were performed for the collagen (1232 cm-1/1336 cm-1) [48], glycosylation (1030 cm-1/1080 cm-1) [49], and protein conformation (1654 cm-1/1554 cm-1) [50]. The ratio of spectral bands 1232 cm-1 and 1336 cm-1 (amide III to the collagen CH2 side chain vibrations) is thought to represent changes in collagen in tissue [48]. In fibrotic areas, this ratio decreased at Day 7 but then appeared to peak at Day 21, when the model demonstrated the most fibrotic changes, and a similar trend was seen in non-fibrotic tissues (Fig. 4(a)). These changes may relate to deposition of abnormal collagen, difference in structure and composition from the collagen normally present in this tissue. This would be reasonable to expect in the areas of fibrosis, and it may be true to a lesser extent in the tissue areas which are not grossly fibrotic, reflecting the dysregulated response of fibroblasts to cytokines mediating the disease process.
Glycosylation is another important biochemical process in tissue, which can manifest spectroscopically through changes of the 1030 cm-1/1080 cm-1 absorbance ratio as shown in Fig. 4(b) [49]. The absorbance at 1030 cm-1 represents the glycogen COH deformation, whereas 1080 cm-1 represents the glycogen C-C stretch, and their ratio in fibrotic regions gradually decreased from control to Day 21 and increased on Day 28 (Fig. 4(b)). While the exact product of glycosylation in human tissues this spectral ratio detects is unclear, it has been clearly associated with disease processes such as in diabetes [49].
Finally, a band ratio examining the spectral changes in the protein regions was also processed. The absorbance at 1654 cm-1 represents the protein C = O stretching of the structural protein (amide I), whereas 1545 cm-1 represents the N-H bending and C-N stretching of the polypeptides and protein background (amide II). Changes in the ratio 1654 cm-1/1554 cm-1 may correspond to changes in the structural rearrangements of existing proteins or the expression of new proteins with varied structural characteristics. In both fibrotic and nonfibrotic regions, this ratio decreased at Day 7 and then increased on the following days (Fig. 4(c)). These changes could be due to a shift from the neutrophilic activity in Day 7 to the deposition of ECM proteins like elastin, fibronectin, and laminin that predominate in later timepoints with the development of fibrosis [51]. The potential benefit of using simple spectral ratios is that advances in new discrete frequency IR imaging approaches such as QCL imaging may offer substantial speed benefits in the clinic over conventional FT-IR approaches when only a few spectral frequencies will suffice for analysis.
3.3 Multivariate data analysis
IR spectra extracted from ROIs can be subjected to multivariate analysis approaches such as PCA which allows for the identification of spectral variance between the different groups. This permits visualization of spectra as single points on a graph and can determine whether different groups have underlying biochemical similarities or differences. PCA converts the large dataset (232 variables per spectrum) to a smaller set of representative variables known as principal components through PCA, so that the new dataset would reflect as much of the variability in the original data as possible, while minimizing redundancy. PCA may be coupled with the supervised approach LDA which increases discrimination between pre-selected classes. LDA maps the PCA data onto a set of axes designed to maximize interclass variability and minimize intraclass variability, to determine whether the samples from each timepoint were in fact more different from each other than from themselves (i.e. both samples from the same timepoints and the other ROIs within the same sample).
PCA-LDA analysis of the spectra extracted from fibrotic regions was able to distinguish the different timepoints of the bleomycin challenge, with the areas from the samples of the same timepoint clustering together (Fig. 5). The separation of the observations by timepoint suggests a difference in biochemical composition (as measured by the spectral signature) over time. Interestingly, the values of the first LD for both fibrotic and nonfibrotic areas varied with the time course, with increasing timepoints moving further from control, with the exception of Day 28, which most closely resembled the control samples.
While a precise interpretation cannot be assigned, the course would suggest that the spectral signature from fibrotic tissue changes with the pathological changes being developed in the model. Inflammation predominates in Day 7, with an influx of inflammatory cells and profibrotic cytokines that transform the healthy control lung. There is parallel development of fibrosis, due to activation of cells in lungs which lead to the deposition of extracellular matrix components, further perturbing the biochemical composition and spectral characteristics of the tissue. By Day 28, there is the beginning of recovery, and the biochemical composition begins to return to the baseline at control demonstrated histologically by partial recovery.
4. Conclusion
This study investigates using IR imaging for the detection of biochemical changes in a murine model of bleomycin-induced pulmonary fibrosis. Based on K-means clustering analysis, the IR dataset reveals biochemical changes across the spatial features of lung tissue that correspond to the tissue variations during fibrosis initiation, consolidation and repair phases. This suggests that the underlying chemical processes that give rise to structural changes are accessible to this imaging modality.
Furthermore, the spectral changes in fibrotic regions throughout the time course of the model, as evidenced by both the spectral ratios and PCA-LDA analysis, suggest that IR imaging can assess how the biochemical status, and therefore the disease state, of the lung tissue is perturbed by the process of fibrosis, and how it begins to return to a healthy baseline.
These two features of the technique, the ability to detect chemical changes across the morphology and biochemical composition of the tissue and to monitor the disease state of the tissue over time, are the first steps in developing this modality as a tool for the study of fibrosis. The capacity to locate areas of lung tissue that are undergoing biochemical change can help assess other models of the disease, while sensitivity to disease progression and recovery offers utility in therapeutic investigations.
This work forms the foundation for further studies into the use of IR imaging as a novel imaging tool to assess emerging anti-fibrotic treatments in large-scale model system testing without the requirements of an experienced pathologist. Furthermore, this may offer novel insight into the biochemical processes occurring in fibrosis that cannot be readily accessed using conventional methodologies. Further work will need to characterize the biochemical changes being detected and assess whether any of these spectral markers are present prior to the manifestation of a morphological changes. More importantly, future studies will translate these findings from the murine bleomycin model to lung fibrotic disorders such as idiopathic pulmonary fibrosis of human tissues and correlate them to clinical outcomes such as response to novel therapeutic agents. With further investigations, IR imaging holds the potential to be a powerful adjunct to current histopathological techniques in finding better ways to understand, manage and develop novel therapeutics for fibrotic diseases.
Acknowledgements
We would like to acknowledge Dr. Thomas Royston for feedback about the study. Imaging services were provided by the Research Resources Center – Research Histology and Tissue Imaging Core at the University of Illinois at Chicago established with the support of the Vice Chancellor of Research.
Disclosures
The authors declare no conflicts of interest.
References
1. C. B. Nanthakumar, R. J. Hatley, S. Lemma, J. Gauldie, R. P. Marshall, and S. J. Macdonald, “Dissecting fibrosis: therapeutic insights from the small-molecule toolbox,” Nat. Rev. Drug Discovery 14(10), 693–720 (2015). [CrossRef]
2. R. Tikhomirov, B. R. Donnell, F. Catapano, G. Faggian, J. Gorelik, F. Martelli, and C. Emanueli, “Exosomes: From Potential Culprits to New Therapeutic Promise in the Setting of Cardiac Fibrosis,” Cells 9(3), 592 (2020). [CrossRef]
3. M. Zeisberg and R. Kalluri, “Cellular mechanisms of tissue fibrosis. 1. Common and organ-specific mechanisms associated with tissue fibrosis,” Am. J. Physiol. Cell Physiol. 304(3), C216–C225 (2013). [CrossRef]
4. T. A. Wynn, “Cellular and molecular mechanisms of fibrosis,” J. Pathol. 214(2), 199–210 (2008). [CrossRef]
5. S. Tomassetti, S. Piciucchi, P. Tantalocco, A. Dubini, and V. Poletti, “The multidisciplinary approach in the diagnosis of idiopathic pulmonary fibrosis: a patient case-based review,” Eur Respir Rev. 24(135), 69–77 (2015). [CrossRef]
6. L. Berchtold, I. Friedli, J. P. Vallee, S. Moll, P. Y. Martin, and S. de Seigneux, “Diagnosis and assessment of renal fibrosis: the state of the art,” Swiss Med. Wkly. 147(1920), w14442 (2017). [CrossRef]
7. B. M. Elicker, K. G. Kallianos, and T. S. Henry, “The role of high-resolution computed tomography in the follow-up of diffuse lung disease: Number 2 in the Series “Radiology” Edited by Nicola Sverzellati and Sujal Desai,” Eur Respir Rev 26(144), 170008 (2017). [CrossRef]
8. J. Wilder and K. Patel, “The clinical utility of FibroScan((R)) as a noninvasive diagnostic test for liver disease,” Med. Devices: Evidence Res. 7, 107–114 (2014). [CrossRef]
9. R. J. Perea, J. T. Ortiz-Perez, M. Sole, M. T. Cibeira, T. M. de Caralt, S. Prat-Gonzalez, X. Bosch, A. Berruezo, M. Sanchez, and J. Blade, “T1 mapping: characterisation of myocardial interstitial space,” Insights Imaging 6(2), 189–202 (2015). [CrossRef]
10. Y. P. Huang, Y. P. Zheng, S. F. Leung, and A. P. Choi, “High frequency ultrasound assessment of skin fibrosis: clinical results,” Ultrasound Med. Biol. 33(8), 1191–1198 (2007). [CrossRef]
11. S. Tomassetti, A. U. Wells, U. Costabel, A. Cavazza, T. V. Colby, G. Rossi, N. Sverzellati, A. Carloni, E. Carretta, M. Buccioli, P. Tantalocco, C. Ravaglia, C. Gurioli, A. Dubini, S. Piciucchi, J. H. Ryu, and V. Poletti, “Bronchoscopic Lung Cryobiopsy Increases Diagnostic Confidence in the Multidisciplinary Diagnosis of Idiopathic Pulmonary Fibrosis,” Am. J. Respir. Crit. Care Med. 193(7), 745–752 (2016). [CrossRef]
12. R. Kaarteenaho, “The current position of surgical lung biopsy in the diagnosis of idiopathic pulmonary fibrosis,” Respir Res 14(1), 43 (2013). [CrossRef]
13. Y. Sumida, A. Nakajima, and Y. Itoh, “Limitations of liver biopsy and non-invasive diagnostic tests for the diagnosis of nonalcoholic fatty liver disease/nonalcoholic steatohepatitis,” World J. Gastroenterol. 20(2), 475–485 (2014). [CrossRef]
14. G. Theophilou, M. Paraskevaidi, K. M. Lima, M. Kyrgiou, P. L. Martin-Hirsch, and F. L. Martin, “Expert review of molecular diagnostics,” 15, 693–713 (2015).
15. M. J. Nasse, M. J. Walsh, E. C. Mattson, R. Reininger, A. Kajdacsy-Balla, V. Macias, R. Bhargava, and C. J. Hirschmugl, “High-resolution Fourier-transform infrared chemical imaging with multiple synchrotron beams,” Nat. Methods 8(5), 413–416 (2011). [CrossRef]
16. D. C. Fernandez, R. Bhargava, S. M. Hewitt, and I. W. Levin, “Infrared spectroscopic imaging for histopathologic recognition,” Nat. Biotechnol. 23(4), 469–474 (2005). [CrossRef]
17. M. J. Walsh, R. K. Reddy, and R. Bhargava, “Label-free biomedical imaging with mid-IR spectroscopy,” IEEE J. Sel. Top. Quantum Electron. 18(4), 1502–1513 (2012). [CrossRef]
18. M. J. Baker, J. Trevisan, P. Bassan, R. Bhargava, H. J. Butler, K. M. Dorling, P. R. Fielden, S. W. Fogarty, N. J. Fullwood, and K. A. Heys, “Using Fourier transform IR spectroscopy to analyze biological materials,” Nat. Protoc. 9(8), 1771–1791 (2014). [CrossRef]
19. D. Mayerich, M. J. Walsh, A. Kadjacsy-Balla, P. S. Ray, S. M. Hewitt, and R. Bhargava, “Stain-less staining for computed histopathology,” Technology 03(01), 27–31 (2015). [CrossRef]
20. B. Bird and M. J. Baker, “Quantum cascade lasers in biomedical infrared imaging,” Trends Biotechnol. 33(10), 557–558 (2015). [CrossRef]
21. K. Yeh, S. Kenkel, J.-N. Liu, and R. Bhargava, “Fast infrared chemical imaging with a quantum cascade laser,” Anal. Chem. 87(1), 485–493 (2015). [CrossRef]
22. F. L. Martin, J. G. Kelly, V. Llabjani, P. L. Martin-Hirsch, I. I. Patel, J. Trevisan, N. J. Fullwood, and M. J. Walsh, “Distinguishing cell types or populations based on the computational analysis of their infrared spectra,” Nat. Protoc. 5(11), 1748–1760 (2010). [CrossRef]
23. S. Pahlow, K. Weber, J. Popp, B. R. Wood, K. Kochan, A. Ruther, D. Perez-Guaita, P. Heraud, N. Stone, A. Dudgeon, B. Gardner, R. Reddy, D. Mayerich, and R. Bhargava, “Application of Vibrational Spectroscopy and Imaging to Point-of-Care Medicine: A Review,” Appl Spectrosc 72, 52–84 (2018). [CrossRef]
24. W. Querido, J. M. Falcon, S. Kandel, and N. Pleshko, “Vibrational spectroscopy and imaging: applications for tissue engineering,” Analyst 142(21), 4005–4017 (2017). [CrossRef]
25. A. Sala, D. J. Anderson, P. M. Brennan, H. J. Butler, J. M. Cameron, M. D. Jenkinson, C. Rinaldi, A. G. Theakstone, and M. J. Baker, “Biofluid Diagnostics by FTIR Spectroscopy: A Platform Technology for Cancer Detection,” Cancer Lett. 477, 122–130 (2020). [CrossRef]
26. C. Krafft, D. Codrich, G. Pelizzo, and V. Sergo, “Raman and FTIR imaging of lung tissue: methodology for control samples,” Vib. Spectrosc. 46(2), 141–149 (2008). [CrossRef]
27. S. Koljenovic, T. C. B. Schut, J. P. van Meerbeeck, A. P. Maat, S. A. Burgers, P. E. Zondervan, J. M. Kros, and G. J. Puppels, “Raman microspectroscopic mapping studies of human bronchial tissue,” J. Biomed. Opt. 9(6), 1187–1198 (2004). [CrossRef]
28. S. Kaminaka, T. Ito, H. Yamazaki, E. Kohda, and H. O. Hamaguchi, “Near-infrared multichannel Raman spectroscopy toward real-time in vivo cancer diagnosis,” J. Raman Spectrosc. 33(7), 498–502 (2002). [CrossRef]
29. Z. Huang, A. McWilliams, H. Lui, D. I. McLean, S. Lam, and H. Zeng, “Near-infrared Raman spectroscopy for optical diagnosis of lung cancer,” Int. J. Cancer 107(6), 1047–1052 (2003). [CrossRef]
30. E. Kaznowska, J. Depciuch, K. Łach, M. Kołodziej, A. Koziorowska, J. Vongsvivut, I. Zawlik, M. Cholewa, and J. Cebulski, “The classification of lung cancers and their degree of malignancy by FTIR, PCA-LDA analysis, and a physics-based computational model,” Talanta 186, 337–345 (2018). [CrossRef]
31. F. Großerueschkamp, A. Kallenbach-Thieltges, T. Behrens, T. Brüning, M. Altmayer, G. Stamatis, D. Theegarten, and K. Gerwert, “Marker-free automated histopathological annotation of lung tumour subtypes by FTIR imaging,” Analyst 140(7), 2114–2120 (2015). [CrossRef]
32. X. Sun, Y. Xu, J. Wu, Y. Zhang, and K. Sun, “Detection of lung cancer tissue by attenuated total reflection–Fourier transform infrared spectroscopy—a pilot study of 60 samples,” J. Surg. Res. 179(1), 33–38 (2013). [CrossRef]
33. J. Sule-Suso, “Synchrotron Based FTIR Spectroscopy in Lung Cancer. Is there a Niche?” Biomedical Applications of Synchrotron Infrared Microspectroscopy: A Practical Approach 279 (2010).
34. J. Pijanka, G. D. Sockalingum, A. Kohler, Y. Yang, F. Draux, G. Parkes, K.-P. Lam, D. Collins, P. Dumas, and C. Sandt, “Synchrotron-based FTIR spectra of stained single cells. Towards a clinical application in pathology,” Lab. Invest. 90(5), 797–807 (2010). [CrossRef]
35. P. D. Lewis, K. E. Lewis, R. Ghosal, S. Bayliss, A. J. Lloyd, J. Wills, R. Godfrey, P. Kloer, and L. A. Mur, “Evaluation of FTIR spectroscopy as a diagnostic tool for lung cancer using sputum,” BMC Cancer 10(1), 640 (2010). [CrossRef]
36. Y. Yang, J. Sulé-Suso, G. D. Sockalingum, G. Kegelaer, M. Manfait, and A. J. El Haj, “Study of tumor cell invasion by Fourier transform infrared microspectroscopy,” Biopolymers 78, 311–317 (2005). [CrossRef]
37. D. Hanahan and R. A. Weinberg, “Hallmarks of cancer: the next generation,” Cell 144(5), 646–674 (2011). [CrossRef]
38. S. S. Nazeer, H. Sreedhar, V. K. Varma, D. Martinez-Marin, C. Massie, and M. J. Walsh, “Infrared spectroscopic imaging: Label-free biochemical analysis of stroma and tissue fibrosis,” Int. J. Biochem. Cell Biol. 92, 14–17 (2017). [CrossRef]
39. V. K. Varma, A. Kajdacsy-Balla, S. Akkina, S. Setty, and M. J. Walsh, “Predicting Fibrosis Progression in Renal Transplant Recipients Using Laser-Based Infrared Spectroscopic Imaging,” Sci. Rep. 8(1), 686 (2018). [CrossRef]
40. J. T. Kwak, A. Kajdacsy-Balla, V. Macias, M. Walsh, S. Sinha, and R. Bhargava, “Improving prediction of prostate cancer recurrence using chemical imaging,” Sci. Rep. 5(1), 8758 (2015). [CrossRef]
41. S. E. Holton, M. J. Walsh, and R. Bhargava, “Subcellular localization of early biochemical transformations in cancer-activated fibroblasts using infrared spectroscopic imaging,” Analyst 136(14), 2953–2958 (2011). [CrossRef]
42. S. Holton, M. Walsh, A. Kajdacsy-Balla, and R. Bhargava, “Label-free characterization of cancer-activated fibroblasts using infrared spectroscopic imaging,” Biophys. J. 101(6), 1513–1521 (2011). [CrossRef]
43. S. E. Holton, A. Bergamaschi, B. S. Katzenellenbogen, and R. Bhargava, “Integration of molecular profiling and chemical imaging to elucidate fibroblast-microenvironment impact on cancer cell phenotype and endocrine resistance in breast cancer,” PLoS One 9(5), e96878 (2014). [CrossRef]
44. A. W. Jones and N. L. Reeve, “Ultrastructural study of bleomycin-induced pulmonary changes in mice,” J. Pathol. 124(4), 227–233 (1978). [CrossRef]
45. G. Izbicki, M. J. Segel, T. G. Christensen, M. W. Conner, and R. Breuer, “Time course of bleomycin-induced lung fibrosis,” Int. J. Exp. Pathol. 83(3), 111–119 (2002). [CrossRef]
46. V. Suryadevara, L. Huang, S. J. Kim, P. Cheresh, M. Shaaya, M. Bandela, P. Fu, C. Feghali-Bostwick, G. Di Paolo, D. W. Kamp, and V. Natarajan, “Role of phospholipase D in bleomycin-induced mitochondrial reactive oxygen species generation, mitochondrial DNA damage, and pulmonary fibrosis,” Am. J. Physiol. Cell Physiol. 317(2), L175–L187 (2019). [CrossRef]
47. P. Lasch, W. Haensch, D. Naumann, and M. Diem, “Imaging of colorectal adenocarcinoma using FT-IR microspectroscopy and cluster analysis,” Biochim. Biophys. Acta, Mol. Basis Dis. 1688, 176–186 (2004). [CrossRef]
48. B. Bird and J. Rowlette, “A protocol for rapid, label-free histochemical imaging of fibrotic liver,” Analyst 142(8), 1179–1184 (2017). [CrossRef]
49. V. K. Varma, A. Kajdacsy-Balla, S. K. Akkina, S. Setty, and M. J. Walsh, “A label-free approach by infrared spectroscopic imaging for interrogating the biochemistry of diabetic nephropathy progression,” Kidney Int. 89(5), 1153–1159 (2016). [CrossRef]
50. N. Krishnakumar, N. Sulfikkarali, S. Manoharan, and R. M. Nirmal, “Screening of chemopreventive effect of naringenin-loaded nanoparticles in DMBA-induced hamster buccal pouch carcinogenesis by FT-IR spectroscopy,” Mol. Cell. Biochem. 382(1-2), 27–36 (2013). [CrossRef]
51. A. J. Lazenby, E. C. Crouch, J. A. McDonald, and C. Kuhn, “Remodeling of the lung in bleomycin-induced pulmonary fibrosis in the rat,” Am. Rev. Respir. Dis. 142(1), 206–214 (1990). [CrossRef]
