CoChat
Sign Up
Cited by 9AI in cancer detectionCell Image Analysis TechniquesArtificial Intelligence in Healthcare and Education

The DOI matches its Crossref record; 18 of the 18 references it cites were found in a registry or an index.

An open source pipeline for quantitative immunohistochemistry image analysis of inflammatory skin disease using artificial intelligence

Open access
Ding, Yuchun, Dhawan, Gaurav, Jones, Claire, Ness, Thomas, Nichols, Esme, Krasnogor, Natalio, Reynolds, Nick J.·2023·Journal of the European Academy of Dermatology and Venereology
DOI 10.1111/jdv.18726PMC PMC10947200
Open this paper in CoChat — free accountOpen PDF

Abstract

BACKGROUND: The application of artificial intelligence (AI) to whole slide images has the potential to improve research reliability and ultimately diagnostic efficiency and service capacity. Image annotation plays a key role in AI and digital pathology. However, the work-streams required for tissue-specific (skin) and immunostain-specific annotation has not been extensively studied compared with the development of AI algorithms. OBJECTIVES: The objective of this study is to develop a common workflow for annotating whole slide images of biopsies from inflammatory skin disease immunostained with a variety of epidermal and dermal markers prior to the development of the AI-assisted analysis pipeline. METHODS: A total of 45 slides containing 3-5 sections each were scanned using Aperio AT2 slide scanner (Leica Biosystems). These slides were annotated by hand using a commonly used image analysis tool which resulted in more than 4000 images blocks. We used deep learning (DL) methodology to first sequentially segment (epidermis and upper dermis), with the exclusion of common artefacts and second to quantify the immunostained signal in those two compartments of skin biopsies and the ratio of positive cells. RESULTS: We validated two DL models using 10-fold validation runs and by comparing to ground truth manually annotated data. The models achieved an average (global) accuracy of 95.0% for the segmentation of epidermis and dermis and 86.1% for the segmentation of positive/negative cells. CONCLUSIONS: The application of two DL models in sequence facilitates accurate segmentation of epidermal and dermal structures, exclusion of common artefacts and enables the quantitative analysis of the immunostained signal. However, inaccurate annotation of the slides for training the DL model can decrease the accuracy of the output. Our open source code will facilitate further external validation across different immunostaining platforms and slide scanners.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1111/jdv.18726

Cites

18 of 18 references verified0 citation errors · 0 not checkable

Statistics

Figures

FIGURE 1 Study flow chart for training of Deeplab V3 AI. Numbers of immunostained slides, sections and images derived are provided for segmentation of epidermis and dermis on the left and cell segmentation on the right. The two Deeplab AI V3 models were run sequentially and then validated using random sampling to provide quantitative outputs and comparisons to ground truth images.
FIGURE 1 Study flow chart for training of Deeplab V3 AI. Numbers of immunostained slides, sections and images derived are provided for segmentation of epidermis and dermis on the left and cell segmentation on the right. The two Deeplab AI V3 models were run sequentially and then validated using random sampling to provide quantitative outputs and comparisons to ground truth images.CC BY · PubMed Central
FIGURE 2 Using QuPath to annotate the epidermis, cellular structures and positively and negatively stained cells within lesional eczema skin. (a) Representative image of γ‐H2AX immunohistochemical staining. Magnification = 12×, scale bar = 100 μm. (b) Representative image of γ‐H2AX immunohistochemical staining with manual annotations for AI training. Epidermis labelled in blue, dermal papilla in purple and cross section dermal papilla in red. Magnification = 12×, scale bar = 100 μm. (c) Higher magnification representative image of γ‐H2AX immunohistochemical staining. Magnification = 40×, scale bar = 20 μm. (d) Higher magnification representative image of γ‐H2AX immunohistochemical staining with manual annotations for AI training. Box within which cells were annotated labelled in green. Strong positive cells labelled in light blue with strong positive nuclei labelled in red. Weak positive nuclei labelled in orange and negative nuclei labelled in turquoise. Magnification = 40×, scale bar = 20 μm.
FIGURE 2 Using QuPath to annotate the epidermis, cellular structures and positively and negatively stained cells within lesional eczema skin. (a) Representative image of γ‐H2AX immunohistochemical staining. Magnification = 12×, scale bar = 100 μm. (b) Representative image of γ‐H2AX immunohistochemical staining with manual annotations for AI training. Epidermis labelled in blue, dermal papilla in purple and cross section dermal papilla in red. Magnification = 12×, scale bar = 100 μm. (c) Higher magnification representative image of γ‐H2AX immunohistochemical staining. Magnification = 40×, scale bar = 20 μm. (d) Higher magnification representative image of γ‐H2AX immunohistochemical staining with manual annotations for AI training. Box within which cells were annotated labelled in green. Strong positive cells labelled in light blue with strong positive nuclei labelled in red. Weak positive nuclei labelled in orange and negative nuclei labelled in turquoise. Magnification = 40×, scale bar = 20 μm.CC BY · PubMed Central
FIGURE 3 Representative image outputs from the deep learning models. (a) Representative image of cyclin B1 immunohistochemical staining in psoriatic lesional skin. Magnification = 6×, scale bar = 200 μm. (b) Image outputs from the trained deep learning (DL) model for slide (a). Magnification = 6×, scale bar = 200 μm. (c) Representative image of γ‐H2AX immunohistochemical staining in psoriatic lesional skin. Magnification = 6×, scale bar = 200 μm. (d) Image outputs from the trained DL model for slide (c). Magnification = 6×, scale bar = 200 μm. (e) Slide (c) (representative image of γ‐H2AX immunohistochemical staining in psoriatic lesional skin) at higher magnification. Magnification = 15×, scale bar = 50 μm. (f) Image output from the trained DL model on slide (e) (representative image of γ‐H2AX immunohistochemical staining in psoriatic lesional skin) at high magnification. Magnification = 15×, scale bar = 50 μm. Epidermis (red), positive cells inside epidermis (purple), negative cells inside epidermis (blue), positive cells outside epidermis (cyan) and negative cells outside epidermis (yellow).
FIGURE 3 Representative image outputs from the deep learning models. (a) Representative image of cyclin B1 immunohistochemical staining in psoriatic lesional skin. Magnification = 6×, scale bar = 200 μm. (b) Image outputs from the trained deep learning (DL) model for slide (a). Magnification = 6×, scale bar = 200 μm. (c) Representative image of γ‐H2AX immunohistochemical staining in psoriatic lesional skin. Magnification = 6×, scale bar = 200 μm. (d) Image outputs from the trained DL model for slide (c). Magnification = 6×, scale bar = 200 μm. (e) Slide (c) (representative image of γ‐H2AX immunohistochemical staining in psoriatic lesional skin) at higher magnification. Magnification = 15×, scale bar = 50 μm. (f) Image output from the trained DL model on slide (e) (representative image of γ‐H2AX immunohistochemical staining in psoriatic lesional skin) at high magnification. Magnification = 15×, scale bar = 50 μm. Epidermis (red), positive cells inside epidermis (purple), negative cells inside epidermis (blue), positive cells outside epidermis (cyan) and negative cells outside epidermis (yellow).CC BY · PubMed Central
FIGURE 4 Quantitative analysis of deep learning (DL) model used to label negative and positive immunostained cells compared with ground truth (manually annotated images). (a) Percentage of positive cell and (b) Average signal. *p<0.05, unpaired t‐test.
FIGURE 4 Quantitative analysis of deep learning (DL) model used to label negative and positive immunostained cells compared with ground truth (manually annotated images). (a) Percentage of positive cell and (b) Average signal. *p<0.05, unpaired t‐test.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Cited by

Sorted by citations

See all 9

Related

See all

Metadata

Authors
Ding, Y., Dhawan, G., Jones, C., Ness, T., Nichols, E., Krasnogor, N., & Reynolds, N. J.
Affiliations
  • Newcastle University (GB)
  • Royal Victoria Infirmary (GB)
  • Engineering and Physical Sciences Research Council (GB)
Corresponding author
  • Ding, Yuchun
  • Reynolds, Nick J.
ORCID
Venue
Journal of the European Academy of Dermatology and Venereology
Publisher
Wiley
Published
2022-12-03
Volume
37
Issue
3
Pages
605-614
ISSN
  • 0926-9959 (ISSN-L)
  • 1468-3083
DOI
10.1111/jdv.18726
PMC
PMC10947200
OpenAlex
W4308772283
Cited by
9
References
18
Citations by year
2026: 2 · 2025: 4 · 2024: 1 · 2023: 1 · 2022: 1
FWCI
0.58
Type
journal article
Language
en
Open access
Licence
Locations
  • Journal of the European Academy of Dermatology and Venereology (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Newcastle University ePrints (Newcastle Univesity) (cc-by)
Funders
Medical Research Council Canada (MC_PC_17168, MR/N005872/1)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › AI in cancer detection
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Biophysics › Cell Image Analysis Techniques
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
Keywords
immunohistochemistry image analysis, inflammatory skin diseases, deep learning, skin biopsy, digital pathology
MeSH terms
Artificial Intelligence, Humans, Immunohistochemistry, Skin Diseases, Software, Reproducibility of Results

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review