- Methodology article
- Open Access
Development of a computer-based quantification method for immunohistochemically-stained tissues and its application to study mast cells in equine wound healing (proof of concept)
BMC Veterinary Research volume 16, Article number: 228 (2020)
There is a growing interest in the scientific community to use computer-based software programs for the quantification of cells during physiological and pathophysiological processes. Drawbacks of computer-based methods currently used to quantify immunohistochemical staining are the complexity of use, expense of software and overly-simplified descriptions of protocol thereby limiting reproducibility. The precise role of mast cells in equine cutaneous wound healing is unknown. Given the contribution of mast cells to the chronic inflammation observed in human keloid, a pathology similar to exuberant granulation tissue (EGT) in horses, mast cells might be present in high numbers in equine limb wounds predisposed to EGT. The main goal of this study was to develop a reliable and reproducible quantification method for immunostained tissues using a computer software that is widely available, at no cost, to the scientific community. A secondary goal was to conduct a proof of concept using the newly-established method to quantify mast cells during wound healing at different anatomical sites (body and limb) in horses to see if a different pattern is observed in limb wounds, which are predisposed to EGT.
A good intraclass correlation coefficient (ICC, 0.67 p < 0.05) was found between the computer-based ImageJ method and manual counting. An excellent intra-operator ICC of 0.90 (p < 0.01) was found for the ImageJ quantification method while a good interoperator ICC of 0.69 (p < 0.01) was measured. No significant difference was observed between the variation of the ImageJ and that of the manual counting method. Mast cells were localized below the epidermis, around cutaneous appendages and blood vessels. Mast cell numbers did not differ significantly in relation to anatomical location or time of healing.
The computer-based quantification method developed is reliable, reproducible, available, cost-free and could be used to study different physiological and pathological processes using immunohistochemistry.
Major drawbacks of computer-based methods currently used to evaluate immunohistochemically-stained tissues are the complexity of use, expense of software and overly-simplified descriptions of protocol limiting reproducibility . Another challenge is to ensure objectivity by validating the method with appropriate parameters including reliability, repeatability and reproducibility . The development of new computer-based methods would be useful to facilitate quantification of cellular populations in immunostained tissues.
Wounds on the limbs of horses suffer a weak and prolonged inflammatory response compared to body wounds, that impairs healing and may lead to the development of exuberant granulation tissue (EGT) . Mast cells are suspected to contribute to chronic inflammation, as is the case in human keloid, a fibroproliferative pathology exhibiting features comparable to EGT .
Mast cells are immune cells that secrete vasodilating factors, and numerous pro- and anti-inflammatory cytokines during wound healing . They also express a c-kit receptor (CD117) that enables cell communication with fibroblasts. Their precise role in cutaneous wound healing in horses is unknown.
The objectives of this study were to 1) develop a reliable and reproducible quantification method for immunostained tissues using a computer software that is widely available, at no cost, to the scientific community, and 2) conduct a proof of concept using the newly-established method to quantify mast cells during skin wound healing at different sites (body and limb) in horses in an effort to determine if a different mast cell pattern is observed in limb wounds predisposed to EGT. It was hypothesized that 1) the computer-based method developed would have excellent repeatability, reproducibility and reliability parameters, and 2) mast cell numbers would be greater and would persist in limb wounds compared to body wounds throughout healing, since this cell is known to contribute to chronic inflammation in human keloids.
One mare developed EGT on its limb wounds from 11 days post wounding until the end of the study and these specimens were excluded from the final analysis (Fig. 1, Additional file 4: Table S1). A total of 16 body and 10 limb wound specimens were included for a total of 26 normally healing wound specimens (Fig. 1).
Validation of the ImageJ computer-based method
The validation of the ImageJ computer-based method was performed on 6 specimens. An intraclass correlation coefficient (ICC) value of 0.67 (p < 0.05) was found between the ImageJ method and the pathologist-based manual counting (Table 1). An intra-operator ICC of 0.90 (p < 0.01) and an inter-operator ICC of 0.69 (p < 0.01) were found for the computer-based method (Table 1). The raw data indicate, that a slight overestimation of mast cell counts occurred with the ImageJ method compared to the manual counting method and may relate to the exclusion of cells, by the pathologist, based on observable morphological parameters (Fig. 2a). Coefficients of variation were calculated for the ImageJ computer-based and the pathologist manual counting-based methods to compare the variation between two counts of each method (Fig. 2b). A greater variation was observed for the ImageJ method (25.19 ± 37.86%) compared to the manual counting method (9.58 ± 7.73%), but no statistically significant difference was observed (p = 0.69).
Proof of concept: quantification of mast cells
Mast cells in intact skin borders were located in the subepidermal and deeper dermal layers and were mostly found around hair follicles, sweat and sebaceous glands, and blood vessels (Fig. 3A-b). Mast cells were also well distributed throughout the granulation tissue (Fig. 3A-c).
In the subepidermal and deep dermal layer, mast cell numbers were greater in body wounds and more variability was seen in limb wounds, but these changes were not statistically significant (Fig. 3B). In the granulation tissue, mast cell kinetics between body and limb wounds were similar, except that the magnitude of the reaction seemed inferior in limb wounds (not statistically significant) (Fig. 3B).
The primary objective of this study was to develop a reliable and reproducible quantification method based on a widely-available cost-free computer software, for the objective evaluation of immunohistochemically-stained tissues. The ImageJ computer-based method developed in this study allows numerical analysis of cells with cell numbers (number of cells/mm2 of tissue) as opposed to previously developed computer-based quantification methods allowing the quantification of the immunostaining intensity .
The intra-operator ICC value confirmed excellent repeatability of the ImageJ method while inter-operator ICC value showed good reproducibility , suggesting that reliability is improved if the method is used by a single operator. This is likely due to the subjective nature of the manual subtraction step required to delete structures considered to be stained non-specifically (Fig. 5d). Few histomorphology studies have validated newly-developed computer-based quantification methods with ICCs to compare the reliability of the methods with the evaluation done by a pathologist or a tissue specialist (e.g. surgeon), mostly considered as the gold standard method [8,9,10,11,12]. To the authors’ knowledge, this study is the first to develop and validate a computer-based quantification method, with ICCs, to evaluate equine wound healing. The ICC found between the ImageJ and the pathologist-based methods (0.67; p < 0.05) confirmed good reliability between the ImageJ and the gold standard methods . Results from this study show no difference in the coefficients of variation between the two methods. The greater mean of coefficients of variation observed in the ImageJ method group could be explained by a single specimen having a coefficient of variation of 96% between count 1 and 2, thus significantly increasing the mean of the group. Other data had a coefficient of correlation between 0 and 40% and had a minimal and maximal difference of 3 and 30%, respectively, with the pathologist group. The manual subtraction step can bring subjectivity and therefore variation between one count to another for the ImageJ method, but the excellent intra-operator ICC found shows that measurements are repeatable. It is difficult to compare such results with other studies aiming to develop computer-based methods applied to cutaneous wound healing since these studies are scarce and ICC agreement intervals may differ in the literature [7, 9, 13, 14]. In short, the hypothesis is partially confirmed since an excellent repeatability and good reproducibility and reliability were found. In light of these results, the newly-developed computer-based ImageJ method presents advantages over manual counting. It yields a precise number of cells in a defined area of tissue and a cell can not be counted twice, limiting human error. Importantly, it requires less expertise and can be used by a variety of personnel when a pathologist or a wound specialist is not available, and it has fewer subjective steps in its process. The ImageJ computer-based method has the manual subtraction step that can introduces subjectivity, while this step as well as the inclusion or exclusion of mast cells contribute to subjectivity in the manual-based counting method. The inclusion or exclusion of mast cells in the counting for the ImageJ method is based on a mathematical parameter (cell area) that is less subjective than morphological parameters like cell shape, visual cell size and nucleus/cytoplasm ratio. It is recognized that in tissues with small numbers of immunostained cells, manual counting could be satisfactory. The advantages discussed above could also make this method a suitable tool for the quantification of other immunostained inflammatory cells like neutrophils and macrophages during physiological and pathological processes. With specific antibodies, the software program can analyze tissues with either numerous or few stained cells since it recognizes the IHC staining. Other mathematical parameters can also be added to the method like the circularity parameter to allow the quantification of multilobed nuclei of cells such as neutrophils [15, 16]. The ImageJ computer-based method we described could eventually help toward monitoring wound healing and response to therapy in the future by the measurement of the immunostained cells present in a small wound biopsy to enhance care.
As a proof of concept, the newly-developed ImageJ method was used to quantitate mast cells, for the first time, in specimens harvested from healing skin wounds of horses. The location of mast cells in the subepidermal and deep dermal layers agrees with other studies conducted on equine skin (Fig. 3A-b) [17, 18]. However, the scattered distribution within the granulation tissue of equine wounds has, to our knowledge, never been reported (Fig. 3A-c). The smaller numbers observed in limb compared to body wounds (Fig. 3b) may reflect a weak inflammatory response seen in limb wounds of horses, refuting our hypothesis . Additional studies are needed to characterize the intrinsic properties of equine mast cells by cell culture and/or other laboratory techniques such as flow cytometry and ELISA to better understand their role in cutaneous wound healing. Although the role of mast cells have been investigated in porcine models where a hypercontractive model was treated with ketotifen (a mast cell stabilizer) and showed significantly less wound contraction than its nontreated control, much remains unknown [19, 20]. Eventually, it may be interesting to compare the kinetics and the number of mast cells of naturally healing equine wounds with those affected by EGT in an effort to establish reference values, a cut off value of mast cell number indicating poor wound prognostic and the clinical role of this cell type.
The principal limitation of our study is the small number of horses, which may explain why no significant differences were found between limb and body wounds in the proof of concept portion of the study, since mast cell numbers are known to show great interindividual variation in horses . However, the multiple sites and the longitudinal nature of the study added robustness for statistical analysis. Results obtained here should help to elucidate the role of mast cells in normal wound healing and future studies might include EGT-affected wounds to document the numbers of mast cells in this healing impairment common to horses.
In conclusion, a reliable, reproducible, available and cost-free computer-based method has been developed to objectively evaluate immunostained tissues. This method could be useful to study and quantify several types of cells implicated in physiological and pathophysiological processes.
Specimens were sourced from a wound tissue bank repository, harvested and archived from four mixed breed mares included in a previous controlled and randomized study . The study’s protocol was approved IACUC of the Université de Montréal and followed the guidelines of the Canadian Council on Animal Care (approval # 15-Rech-1811). Briefly, four full-thickness skin wounds were created, after sedation and local anesthesia, on one lateral mid-thoracic wall (two 15 cm2 wounds) and on the distal extremity of one thoracic limb (two 6 cm2 wounds) per horse (corresponding to the control wounds in the original study). Wounds were left to heal by second intention. Full-thickness wound samples were harvested with an 8 mm diameter biopsy punch from the edge of the 2 wounds on the limb and on the thoracic wall (alternatively) at day 1 (D1), D3, D8 and D17. D0 samples corresponded to normal skin obtained upon wound creation. Wounds and granulation tissue were photographed during the healing process and photographs were visually evaluated at the end of the study for granulation tissue elevation scoring according to the Bigbie et al. scale by a blinded board-certified veterinary surgeon. Elevation scores between 1 and 4 were attributed, score 1 corresponded to an elevation not exceeding the edges of the wound. Score 2 to an elevation at the same level as the wound edge. Score 3 corresponded to an elevation exceeding the edges of the wound. Score 4 corresponded to an elevation exceeding the edges of the wound and covering the epithelium . The experimental model is shown in Fig. 1. A total of 31 specimens were harvested and archived, of which 26 were included in the current study based on inclusion and exclusion criteria (n = 26) (Fig. 1, Additional file 4: Table S1) . Specimens included 3–4 mm of intact skin borders (subepidermal layer and the deep dermal layer) and 3–4 mm of the wound (granulation tissue) and were fixed in 10% formalin, dehydrated, paraffin-embedded and then cut into 4 μm sections .
Immunohistochemistry for mast cell staining
A rabbit polyclonal anti-CD117/c-kit antibody (1/87, #RB-9038, Thermo Fisher Scientific, Rockford, IL, USA) was incubated on the sections for 1 h at room temperature . An equine mast cell tumor was used as a positive control and rabbit serum replaced the primary antibody for the negative control (Fig. 4). The detailed protocol is presented in Additional file 1.
Image analysis and quantification
Immunostained cells were observed under light microscopy at 200x and scanned with the Panoptiq software program v.1. 4. 3 (ViewsIQ, Richmond, BC, CAN). Mast cell characterization and quantification was performed by 2 blinded observers in the wound samples harvested from a former study . In the intact skin borders of the wound, mast cell quantification was performed in the subepidermal layer (0–296 μm under basal membrane) and in the deep dermal layer (296–1184 μm) . In the wound, mast cell quantification was performed in the granulation tissue. From each region, 5 high power fields (HPFs) were selected randomly. Mast cells were quantified using the newly-developed computer-based quantification method based on the ImageJ software program (U. S. National Institutes of Health, Bethesda, MD, USA). Mast cells in six of these specimens were also manually counted by an ACVP-certified pathologist and statistically compared to the ImageJ method. For both methods, identification of mast cells was based on CD117+ staining by immunohistochemistry and the presence of a nucleus. Particles non-specifically-stained and small foci of non-specific staining without a nucleus were manually (ImageJ) or visually (pathologist) subtracted and excluded from the count. For the ImageJ method, CD117+ cells with a cell area between 8 to 300 μm2 were considered mast cells and were included in the count. For the pathologist-based manual counts, inclusion of mast cells in the counts was based on the following morphological parameters: cell shape, cell size and nucleus/cytoplasm ratio. The computer-based quantification method is shown in Fig. 5 and the complete protocol is described in Additional file 2. Additional file 3 provides details about the chosen area range. Mast cell numbers were then compared between limb and body wounds.
Statistical analysis was done with SPSS software program v.25 (IBM, Armonk, NY, USA) to validate the quantification method by calculating ICCs and Wilcoxon signed rank test for paired data, and with SAS software program v.9.3 (SAS Institute Inc., Cary, NC, USA) for mast cell quantification. A linear model for repeated measures was used to detect differences relating to time and to anatomical location of the wound (limb vs body). A p-value < 0.05 was considered statistically significant.
Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Exuberant granulation tissue
Intraclass correlation coefficient
Negative pressure wound therapy
High power field
Choudhury KR, Yagle KJ, Swanson PE, Krohn KA, Rajendran JG. A robust automated measure of average antibody staining in immunohistochemistry images. J Histochem Cytochem. 2010;58(2):95–107.
Jensen EC. Quantitative analysis of histological staining and fluorescence using ImageJ. Anatomical Rec (Hoboken, NJ : 2007). 2013;296(3):378–81.
Wilmink JM, Veenman JN, van den Boom R, Rutten VP, Niewold TA, Broekhuisen-Davies JM, Lees R, Armstrong S, van Weeren PR, Barneveld A. Differences in polymorphonucleocyte function and local inflammatory response between horses and ponies. Equine Vet J. 2003;35(6):561–9.
Shaker SA, Ayuob NN, Hajrah NH. Cell talk: a phenomenon observed in the keloid scar by immunohistochemical study. Appl Immunohistochem Mol Morphol. 2011;19(2):153–9.
Ng MF. The role of mast cells in wound healing. Int Wound J. 2010;7(1):55–61.
Matos LL, Stabenow E, Tavares MR, Ferraz AR, Capelozzi VL, Pinhal MA. Immunohistochemistry quantification by a digital computer-assisted method compared to semiquantitative analysis. Clinics (Sao Paulo, Brazil). 2006;61(5):417–24.
Cicchetti DV, Sparrow SA. Developing criteria for establishing interrater reliability of specific items: applications to assessment of adaptive behavior. Am J Ment Defic. 1981;86(2):127–37.
Prieto MF, Acha B, Gomez-Cia T, Fondon I, Serrano C. A system for 3D representation of burns and calculation of burnt skin area. Burns. 2011;37(7):1233–40.
Straughen JK, Misra DP, Ernst LM, Charles AK, VanHorn S, Ghosh S, Buhimschi I, Buhimschi C, Divine G, Salafia CM. Methods to decrease variability in histological scoring in placentas from a cohort of preterm infants. BMJ Open. 2017;7(3):e013877.
Wang SC, Anderson JAE, Evans R, Woo K, Beland B, Sasseville D, Moreau L. Point-of-care wound visioning technology: reproducibility and accuracy of a wound measurement app. PLoS One. 2017;12(8):e0183139.
Malone M, Schwarzer S, Walsh A, Xuan W, Al Gannass A, Dickson HG, Bowling FL. Monitoring wound progression to healing in diabetic foot ulcers using three-dimensional wound imaging. J Diabetes Complicat. 2020;34(2):107471.
Koetsier KS, Wong JN, Muffley LA, Carrougher GJ, Pham TN, Gibran NS. Prospective observational study comparing burn surgeons' estimations of wound healing after skin grafting to photo-assisted methods. Burns. 2019;45(7):1562–70.
O'Connor SR, Tully MA, Ryan B, Bradley JM, Baxter GD, McDonough SM. Failure of a numerical quality assessment scale to identify potential risk of bias in a systematic review: a comparison study. BMC Res Notes. 2015;8:224.
Cuchna JW, Hoch MC, Hoch JM. The interrater and intrarater reliability of the functional movement screen: a systematic review with meta-analysis. Phys Ther Sport. 2016;19:57–65.
Ulbrich H, Prech P, Luxenburger A, Dannhardt G. Characterization of a computerized assay for rapid and easy determination of leukocyte adhesion to endothelial cells. Biol Pharm Bull. 2005;28(4):718–24.
Garrity-Park M, Loftus EV Jr, Sandborn WJ, Smyrk TC. Myeloperoxidase immunohistochemistry as a measure of disease activity in ulcerative colitis: association with ulcerative colitis-colorectal cancer, tumor necrosis factor polymorphism and RUNX3 methylation. Inflamm Bowel Dis. 2012;18(2):275–83.
Scott DW. Equine dermatology. 2nd ed. St. Louis: Saunders; 2011.
Jorgensen E, Lazzarini G, Pirone A, Jacobsen S, Miragliotta V. Normal microscopic anatomy of equine body and limb skin: a morphological and immunohistochemical study. Ann Anat. 2018;218:205–12.
Kawarai S, Masuda K, Ohmori K, Matsuura S, Yasuda N, Nagata M, Sakaguchi M, Tsujimoto H. Cultivation and characterization of canine skin-derived mast cells. J Vet Med Sci. 2010;72(2):131–40.
Gallant-Behm CL, Hildebrand KA, Hart DA. The mast cell stabilizer ketotifen prevents development of excessive skin wound contraction and fibrosis in red Duroc pigs. Wound Repair Regen. 2008;16(2):226–33.
Kamus L, Rameau M, Theoret C. Feasibility of a disposable canister-free negative-pressure wound therapy (NPWT) device for treating open wounds in horses. BMC Vet Res. 2019;15(1):78.
Bigbie RB, Schumacher J, Swaim SF, Purohit RC, Wright JC. Effects of amnion and live yeast cell derivative on second-intention healing in horses. Am J Vet Res. 1991;52(8):1376–82.
Theoret CL, Olutoye OO, Parnell LK, Hicks J. Equine exuberant granulation tissue and human keloids: a comparative histopathologic study. Vet Surg. 2013;42(7):783–9.
The authors wish to thank Dr. Emma Marchionatti for help with mast cell quantification (proof of concept) and Dr. Guy Beauchamp for help with statistical analyses.
The study was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC 250231) for university-based research. This funding agency played no part in the study.
Ethics approval and consent to participate
The protocol follows the guidelines for the care and use of laboratory animals as established by the Canadian Council on Animal Care and approved by the Institutional Animal Care and Use Committee of the Université de Montréal (# 15-Rech-1811).
Consent for publication
No competing interests have been declared.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Detailed mast cell immunohistochemistry protocol. Full immunohistochemistry protocol for the mast cell identification in limb and body wounds of horses included in the study.
ImageJ quantification protocol used on mast cells. Full protocol of the quantification of mast cells in limb and body wounds of horses included in the study with the ImageJ software program.
Establishment of protocol of mast cell area interval. This document illustrates how the range of mast cell area interval was established by using the ImageJ software program for the later quantification.
Numbers and characteristics of the samples included in the study. Specimen characteristics are grouped in a table to show which specimens were included or excluded in the study and why.
About this article
Cite this article
Dubuc, V., Laverty, S., Richard, H. et al. Development of a computer-based quantification method for immunohistochemically-stained tissues and its application to study mast cells in equine wound healing (proof of concept). BMC Vet Res 16, 228 (2020). https://doi.org/10.1186/s12917-020-02444-x
- Wound healing
- Mast cell