AI Science Tracker
EN

Pathology networks identify gastric-cancer lymph-node metastases

Peking University Cancer Hospital; Tsinghua University · 23 Jan 2021 · medicine

Discovery

Pathology networks identify gastric-cancer lymph-node metastases

Source check: Source checked. Not independent scientific confirmation.

Finding: A cascade of Faster R-CNN/DeepLab and fused Xception/DenseNet models analyzed lymph-node slides from 222 gastric-cancer patients. In the reported 327-slide test cohort from 94 patients, metastasis classification had positive and negative predictive values of 93.53% and 97.99%. This is a single-center archival clinical validation; the paper’s “prospective” terminology does not establish a new frozen-model clinical trial. Poorly differentiated and scattered tumor cells were more often missed, and no improved patient outcome was demonstrated.

Technology: Faster R-CNN; DeepLab; fused Xception and DenseNet-121

Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
Faster R-CNN/DeepLab locate and segment lymph nodes; fused Xception/DenseNet learned features classify metastatic tissue against expert pathology labels.
Test and result
Pathology-reference classification of lymph-node metastases in archival 2009–2019 gastric-cancer slides. The source describes 594 training slides and 327 test slides, with reported test-cohort PPV 93.53% and NPV 97.99%; these are not patient-level screening-benefit estimates.
Limitation / still unknown
External-center generalization, prospective frozen-model performance and patient benefit remain unestablished. The Methods alternate image-random-split and patient-cohort descriptions, leaving the exact grouping of repeated slides incompletely documented. Hard-to-distinguish histological subtypes had more false negatives.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://link.springer.com/article/10.1007/s10120-021-01158-9

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 21 Nov 2026. Review scope: Primary publisher full relevant cohort/reference, model and validation Methods, Results and discussion; supplementary materials not inspected. Publisher canonical front DOI/title and online date verified. Exact primary identity/date and fresh primary PubMed correction/retraction relationships checked..

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open in the full catalog for filters, charts and related discoveries. Browse medicine discoveries.