Robust Cell Detection and Segmentation for Image Cytometry Reveal Th17 Cell Heterogeneity

Our plain-English summary of Robust Cell Detection and Segmentation for Image Cytometry Reveal Th17 Cell Heterogeneity, by T. Tsujikawa, G. Thibault, V. Azimi, et al (Cytometry. Part a, 2019).

The study

• Authors: T. Tsujikawa, G. Thibault, V. Azimi, S. Sivagnanam, G. Banik, C. Means, R. Kawashima, D.R. Clayburgh, J.W. Gray, L.M. Coussens, Y.H. Chang
• Journal: Cytometry. Part a
• Published: 2019;95(4):389-398
• Cited by: 13 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Image cytometry enables quantitative cell characterization with preserved tissue architecture; thus, it has been highlighted in the advancement of multiplex immunohistochemistry (IHC) and digital image analysis in the context of immune-based biomarker monitoring associated with cancer immunotherapy. However, one of the challenges in the current image cytometry methodology is a technical limitation in the segmentation of nuclei and cellular components particularly in heterogeneously stained cancer tissue images. To improve the detection and specificity of single-cell segmentation in hematoxylin-stained images (which can be utilized for recently reported 12-biomarker chromogenic sequential multiplex IHC), we adapted a segmentation algorithm previously developed for hematoxlin and eosin-stained images, where morphological features are extracted based on Gabor-filtering, followed by stacking of image pixels into n-dimensional feature space and unsupervised clustering of individual pixels. Our proposed method showed improved sensitivity and specificity in comparison with standard segmentation methods. Replacing previously proposed methods with our method in multiplex IHC/image cytometry analysis, we observed higher detection of cell lineages including relatively rare TH 17 cells, further enabling sub-population analysis into TH 1-like and TH 2-like phenotypes based on T-bet and GATA3 expression. Interestingly, predominance of TH 2-like TH 17 cells was associated with human papilloma virus (HPV)-negative status of oropharyngeal squamous cell carcinoma of head and neck, known as a poor-prognostic subtype in comparison with HPV-positive status. Furthermore, TH 2-like TH 17 cells in HPV-negative head and neck cancer tissues were spatiotemporally correlated with CD66b+ granulocytes, presumably associated with an immunosuppressive microenvironment. Our cell segmentation method for multiplex IHC/image cytometry potentially contributes to in-depth immune profiling and spatial association, leading to further tissue-based biomarker exploration. © 2019 International Society for Advancement of Cytometry.

© 2019 International Society for Advancement of Cytometry.

In plain English

This methods study did not examine asbestos exposure. The authors built a computer method that outlines individual cell nuclei in microscope images of tissue stained only with hematoxylin. They compared it with two standard programs, CellProfiler and ImageJ/FIJI, on images of head and neck squamous cell carcinoma, malignant pleural mesothelioma and pancreatic ductal adenocarcinoma. Three physicians marked cells by hand, agreeing on 70 to 76.7% of them. With cells marked by at least two physicians as the reference, the new method's sensitivity was 65–78% across the three cancer types, against 55% for CellProfiler. With cells marked by at least one physician, its specificity was about 90%. In 38 head and neck cancer samples it detected more cells, including significantly more of the rare TH17 immune cells. HPV-positive tumors (21) had a significantly higher share of TH1-like TH17 cells than HPV-negative tumors (17). In 9 further HPV-negative head and neck cancers, TH2-like TH17 cells lay significantly closer to CD66b+ granulocytes than TH1-like ones. The authors called the work a proof of concept. They noted that poor-quality staining can cause failures.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Methods development and validation study of an image-analysis (cell segmentation) algorithm, applied to human cancer tissue images
• Who was studied: Hematoxylin-stained images from three cancer types (head and neck squamous cell carcinoma, malignant mesothelioma, pancreatic ductal adenocarcinoma) for the method comparison; a previously published tissue microarray of head and neck squamous cell carcinoma (N = 38: 21 HPV-positive, 17 HPV-negative); and a further cohort of HPV-negative head and neck squamous cell carcinoma (N = 9). Samples came from Oregon Health & Science University under IRB approval.
• Main results:
  • When the reference was cells marked by two or all three physicians, the new method's sensitivity was 65–78% across the three cancer types, while CellProfiler's stayed at 55%.
  • The new method's specificity was approximately 90% when the reference was cells marked by one or more physicians.
  • The new method and ImageJ/FIJI had equivalent sensitivity in malignant mesothelioma and head and neck squamous cell carcinoma tissue. The new method outperformed ImageJ/FIJI in pancreatic ductal adenocarcinoma tissue.
  • Across the 38 head and neck cancer samples, the new method detected a statistically significantly higher number of TH17 cells than CellProfiler.
  • HPV-positive head and neck tumors (N = 21) showed a significantly higher frequency of TH1-like TH17 cells than HPV-negative tumors (N = 17).
  • In the 38 samples, TH2-like TH17 cell density correlated with CD66b+ granulocyte percentage (Spearman r 0.372, 95% CI 0.04957 to 0.6242, P 0.0215). Total TH17 cells did not correlate with CD66b+ granulocytes (r 0.05749, P 0.7317, not significant).
  • In the 9 HPV-negative head and neck cancers, TH2-like TH17 cells were significantly closer to CD66b+ granulocytes than TH1-like TH17 cells were.
• The authors' conclusion: The authors concluded that their approach improved cell segmentation for image cytometry based on multiplex immunohistochemistry and increased detection of rare cell populations, including TH17 subpopulations. They said this may support more detailed immune profiling and further tissue-based biomarker research.
• Limitations the authors state:
  • The method depends on the quality of hematoxylin staining. Low-quality images can lead to failed cell detection or over- or under-segmentation.
  • The manual reference annotations by the three physicians were only moderately consistent (70 to 76.7% matched), reflecting technical challenges in heterogeneous cancer tissue.
  • The work is an early proof of concept, and no software package had been released at the time.

Topics

Medical terms in this study, explained in our Lexicon: Immunohistochemistry, Biomarker, Immunotherapy, GATA3, Squamous Cell Carcinoma.

Related studies

• Diffusion-weighted MRI of malignant pleural mesothelioma: preliminary assessment of apparent diffusion coefficient in histologic subtypes — R.R. Gill, S. Umeoka, et al. (2010)
• Can semi-automated image cytometry on induced sputum become a screening tool for lung cancer? Evaluation of quantitative semi-automated sputum cytometry on radon- and uranium-exposed workers — W. Marek, N. Kotschy-Lang, et al. (2001)
• Current evidence base of FDG-PET/CT imaging in the clinical management of malignant pleural mesothelioma: emerging significance of image segmentation and global disease assessment — S. Basu, B. Saboury, et al. (2011)
• Novel cell imaging techniques show induction of apoptosis and proliferation in mesothelial cells by asbestos — J.L. Goldberg, C.L. Zanella, et al. (1997)
• The analysis of novel microRNA mimic sequences in cancer cells reveals lack of specificity in stem-loop RT-qPCR-based microRNA detection — P. Winata, M. Williams, et al. (2017)
• A simple "expert system" for morphometric evaluation of cells in pleural effusions — M. Oberholzer, G. Feichter, et al. (1989)

Browse return to medical research.