Artificial Intelligence-Aided Recognition of Pathological Characteristics and Subtype Classification of Superficial Perivascular Dermatitis
Open accessAbstract
Background: Superficial perivascular dermatitis, an important type of inflammatory dermatosis, comprises various skin diseases, which are difficult to distinguish by clinical manifestations and need pathological imaging observation. Coupled with its complex pathological characteristics, the subtype classification depends to a great extent on dermatopathologists. There is an urgent need to develop an efficient approach to recognize the pathological characteristics and classify the subtypes of superficial perivascular dermatitis. Methods: 3,954 pathological images (4 × and 10 ×) of three subtypes—psoriasiform, spongiotic and interface—of superficial perivascular dermatitis were captured from 327 cases diagnosed both clinically and pathologically. The control group comprised 1,337 pathological images of 85 normal skin tissue slides taken from the edge of benign epidermal cysts. First, senior dermatologists and dermatopathologists followed the structure–pattern analysis method to label the pathological characteristics that significantly contribute to classifying different subtypes on 4 × and 10 × images. A cascaded deep learning algorithm framework was then proposed to establish pixel-level pathological characteristics' masks and classify the subtypes by supervised learning. Results: 13 different pathological characteristics were recognized, and the accuracy of subtype classification was 85.24%. In contrast, the accuracy of the subtype classification model without recognition was 71.35%. Conclusion: Our cascaded deep learning model used small samples to deliver efficient recognition of pathological characteristics and subtype classification simultaneously. Moreover, the proposed method could be applied to both microscopic images and digital scanned images.
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.
Integrity signals
Registry
Cites
Statistics
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.
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?
Bao et al.'s cascaded deep learning framework classified subtypes of superficial perivascular dermatitis (psoriasiform, spongiotic, and interface types) with an overall accuracy of 85.24%, significantly outperforming a baseline model without recognition capabilities, which had an accuracy of 71.35%.
“The model achieved an overall accuracy of 85.24%, significantly outperforming a baseline model without recognition capabilities, which had an accuracy of 71.35%.”
From Artificial Intelligence in Dermatopathology: An Update and Review of the Current Literature (2026) · cites this paper as [30]
AI can effectively assist in identifying key pathological features and classifying inflammatory skin conditions, potentially aiding dermatopathologists in diagnosing and differentiating between subtypes of superficial perivascular dermatitis.
“The study demonstrated that AI can effectively assist in identifying key pathological features and classifying inflammatory skin conditions, potentially aiding dermatopathologists in diagnosing and differentiating between subtypes of superficial perivascular dermatitis.”
From Artificial Intelligence in Dermatopathology: An Update and Review of the Current Literature (2026) · cites this paper as [30]
In Bao et al.'s model, accuracy was very high for normal skin (>99%), interface dermatitis (83.69%), and spongiotic dermatitis (81.12%).
“In particular, the accuracy was very high for normal skin (>99%), but also for interface dermatitis (83.69%) and spongiotic dermatitis (81.12%)”
From Artificial Intelligence in the Histopathological Assessment of Non-Neoplastic Skin Disorders: A Narrative Review with Future Perspectives (2025) · cites this paper as [18]
Bao et al.'s AI model for classifying subtypes of superficial perivascular dermatitis achieved an overall accuracy of 85.24%, with sensitivity 67.46% and specificity 89.09%.
“Altogether, the accuracy of subtype classification was 85.24%, with a sensitivity of 67.46% and a specificity of 89.09%.”
From Artificial Intelligence in the Histopathological Assessment of Non-Neoplastic Skin Disorders: A Narrative Review with Future Perspectives (2025) · cites this paper as [18]
In Bao et al.'s model, accuracy for the psoriasiform pattern was 64.58%.
“The accuracy for the psoriasiform pattern was instead 64.58%.”
From Artificial Intelligence in the Histopathological Assessment of Non-Neoplastic Skin Disorders: A Narrative Review with Future Perspectives (2025) · cites this paper as [18]
The results of Bao et al. are very close to the actual diagnostic accuracy of dermatopathologists in clinical practice, representing an encouraging preliminary result.
“Overall, the results obtained are very close to the actual diagnostic accuracy of dermatopathologists in clinical practice, representing an encouraging preliminary result in this field [18].”
From Artificial Intelligence in the Histopathological Assessment of Non-Neoplastic Skin Disorders: A Narrative Review with Future Perspectives (2025) · cites this paper as [18]
Bao et al.'s AI accuracy was best in recognizing acanthosis, lichenoid infiltration, hypergranulosis, hyperkeratosis and, to a lesser extent, inflammatory cell infiltration.
“in fact, AI accuracy was the best in recognizing acanthosis, lichenoid infiltration, hypergranulosis, hyperkeratosis and, to a lesser extent, inflammatory cell infiltration, i.e., the key histologic features of interface and spongiotic patterns.”
From Artificial Intelligence in the Histopathological Assessment of Non-Neoplastic Skin Disorders: A Narrative Review with Future Perspectives (2025) · cites this paper as [18]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
The citation lists haven’t been fetched yet
References 14 · Cited by 5
- Authors
- Bao, Y., Zhang, J., Zhang, Q., Chang, J., Lu, D., & Fu, Y.
- Affiliations
- Beijing Hospital (CN)
- Tsinghua University (CN)
- Corresponding author
- Fu, Yu
- Venue
- Frontiers in Medicine
- Publisher
- Frontiers Media SA
- Published
- 2021-07-16
- Volume
- 8
- Pages
- 696305-696305
- ISSN
- 2296-858X (ISSN-L)
- PMC
- PMC8322609
- OpenAlex
- W3183711945
- Cited by
- 5
- References
- 14
- Citations by year
- 2026: 1 · 2025: 3 · 2024: 1
- FWCI
- 0.09
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Frontiers in Medicine (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Natural Science Foundation of Beijing Municipality
- Beijing Nova Program
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › AI in cancer detection
- Health Sciences › Medicine › Dermatology › Medicine and Dermatology Studies History
- Health Sciences › Medicine › Oncology › Cutaneous Melanoma Detection and Management
- Keywords
- dermatopathology, subtype classification, psoriasiform dermatitis, spongiotic dermatitis, interface dermatitis
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
