CoChat

Impact of a deep learning assistant on the histopathologic classification of liver cancer

Open access

The DOI matches its Crossref record; 35 of the 35 references it cites were found in a registry or an index.

Abstract

Artificial intelligence (AI) algorithms continue to rival human performance on a variety of clinical tasks, while their actual impact on human diagnosticians, when incorporated into clinical workflows, remains relatively unexplored. In this study, we developed a deep learning-based assistant to help pathologists differentiate between two subtypes of primary liver cancer, hepatocellular carcinoma and cholangiocarcinoma, on hematoxylin and eosin-stained whole-slide images (WSI), and evaluated its effect on the diagnostic performance of 11 pathologists with varying levels of expertise. Our model achieved accuracies of 0.885 on a validation set of 26 WSI, and 0.842 on an independent test set of 80 WSI. Although use of the assistant did not change the mean accuracy of the 11 pathologists (p = 0.184, OR = 1.281), it significantly improved the accuracy (p = 0.045, OR = 1.499) of a subset of nine pathologists who fell within well-defined experience levels (GI subspecialists, non-GI subspecialists, and trainees). In the assisted state, model accuracy significantly impacted the diagnostic decisions of all 11 pathologists. As expected, when the model’s prediction was correct, assistance significantly improved accuracy (p = 0.000, OR = 4.289), whereas when the model’s prediction was incorrect, assistance significantly decreased accuracy (p = 0.000, OR = 0.253), with both effects holding across all pathologist experience levels and case difficulty levels. Our results highlight the challenges of translating AI models into the clinical setting, and emphasize the importance of taking into account potential unintended negative consequences of model assistance when designing and testing medical AI-assistance tools.

Overview

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

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

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

Integrity signals

Registry

Crossref · 10.1038/s41746-020-0232-8

Cites

35 of 35 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Graphical user interfaces for the experiment. The pathologists navigated the slide using the ObjectiveView image viewer. After identifying a tumor region of interest (ROI), they saved a 500 × 500 μm image patch at ×10 objective magnification (a) containing the ROI using the ‘crop’ tool. (The horizonal scale bar denotes 200 μm). After uploading the image patch to the diagnostic assistant’s user interface (b), they received a probability of each diagnosis (here, HCC), with an accompanying class activation map to assist with interpretation.
Fig. 1 Graphical user interfaces for the experiment. The pathologists navigated the slide using the ObjectiveView image viewer. After identifying a tumor region of interest (ROI), they saved a 500 × 500 μm image patch at ×10 objective magnification (a) containing the ROI using the ‘crop’ tool. (The horizonal scale bar denotes 200 μm). After uploading the image patch to the diagnostic assistant’s user interface (b), they received a probability of each diagnosis (here, HCC), with an accompanying class activation map to assist with interpretation.CC BY · PubMed Central
Fig. 2 Experimental design. Each of the 11 pathologists was randomly assigned to either test Order 1 or 2. Each test began with a brief practice block of 4 (2 HCC and 2 CC) practice whole-slide images (WSI), followed by 8 experiment blocks of 10 WSI each, with Order 1 beginning with assistance and Order 2 beginning without assistance. The same 80 experiment WSI were reviewed in the same sequence during Tests 1 and 2, across both test Orders.
Fig. 2 Experimental design. Each of the 11 pathologists was randomly assigned to either test Order 1 or 2. Each test began with a brief practice block of 4 (2 HCC and 2 CC) practice whole-slide images (WSI), followed by 8 experiment blocks of 10 WSI each, with Order 1 beginning with assistance and Order 2 beginning without assistance. The same 80 experiment WSI were reviewed in the same sequence during Tests 1 and 2, across both test Orders.CC BY · PubMed Central
Fig. 3 Pathologist diagnostic workflow with assistance. After reviewing the H&E whole-slide image (a), the pathologist extracts a tumor patch at ×10 objective magnification (b) and uploads it to the cloud-based model, which outputs predicted probabilities for cholangiocarcinoma (CC) and hepatocellular carcinoma (HCC) into the user interface (c), as well as corresponding class activation maps (not shown). These outputs are integrated with the pathologist’s diagnostic impression, to result in a final assisted diagnosis.
Fig. 3 Pathologist diagnostic workflow with assistance. After reviewing the H&E whole-slide image (a), the pathologist extracts a tumor patch at ×10 objective magnification (b) and uploads it to the cloud-based model, which outputs predicted probabilities for cholangiocarcinoma (CC) and hepatocellular carcinoma (HCC) into the user interface (c), as well as corresponding class activation maps (not shown). These outputs are integrated with the pathologist’s diagnostic impression, to result in a final assisted diagnosis.CC BY · PubMed Central
Fig. 4 Impact of assistance on individual pathologist diagnostic performance. The average diagnostic accuracy (across the set of 80 experiment WSI) for each pathologist is plotted as follows: gray circle (unassisted) = accuracy of the unassisted pathologist, star (model) = accuracy of the model alone (based on pathologist selected input patches), purple diamond (assisted) = accuracy of the pathologist with model assistance.
Fig. 4 Impact of assistance on individual pathologist diagnostic performance. The average diagnostic accuracy (across the set of 80 experiment WSI) for each pathologist is plotted as follows: gray circle (unassisted) = accuracy of the unassisted pathologist, star (model) = accuracy of the model alone (based on pathologist selected input patches), purple diamond (assisted) = accuracy of the pathologist with model assistance.CC BY · PubMed Central

Verification

Claims the paper makes

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

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

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

Show all 35 references

Cited by

See all 327

Metadata

Authors
Kiani, A., Uyumazturk, B., Rajpurkar, P., Wang, A., Gao, R., Jones, E., Yu, Y., Langlotz, C. P., Ball, R. L., Montine, T. J., Martin, B. A., Berry, G. J., Ozawa, M. G., Hazard, F. K., Brown, R. A., Chen, S. B., Wood, M., Allard, L. S., Ylagan, L., Ng, A. Y., & Shen, J.
Affiliations
  • Stanford University (US)
  • Stanford Medicine (US)
Corresponding author
  • Ng, Andrew Y.
  • Shen, Jeanne
Venue
npj Digital Medicine
Published
2020-02-26
Volume
3
Issue
1
Pages
23-23
ISSN
2398-6352 (ISSN-L)
OpenAlex
W3007464329
Cited by
325
References
34
Citations by year
2026: 41 · 2025: 64 · 2024: 62 · 2023: 55 · 2022: 40 · 2021: 38 · 2020: 25
FWCI
15.53
Type
journal article
Language
en
Open access
Licence
Locations
  • npj Digital Medicine (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › AI in cancer detection
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Radiomics and Machine Learning in Medical Imaging
Keywords
liver cancer, hepatocellular carcinoma, cholangiocarcinoma, histological classification, whole slide images, clinical implementation of AI, H&E staining

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

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Impact of a deep learning assistant on the histopathologic classification of liver cancer. https://app.cochat.ai/papers/10.1038/s41746-020-0232-8
The paper
Kiani, A., Uyumazturk, B., Rajpurkar, P., Wang, A., Gao, R., Jones, E., Yu, Y., Langlotz, C. P., Ball, R. L., Montine, T. J., Martin, B. A., Berry, G. J., Ozawa, M. G., Hazard, F. K., Brown, R. A., Chen, S. B., Wood, M., Allard, L. S., Ylagan, L., . . . Shen, J. (2020). Impact of a deep learning assistant on the histopathologic classification of liver cancer. npj Digital Medicine. https://doi.org/10.1038/s41746-020-0232-8

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.