An interpretable deep learning framework for intestinal metaplasia detection in gastric histopathology images - Report - MDSpire

An interpretable deep learning framework for intestinal metaplasia detection in gastric histopathology images

  • By

  • Alia Al-Mohtaseb

  • Fahad T. Alotaibi

  • Salem Alhatamleh

  • Hatem Malkawi

  • Amal Alishwait

  • Ala Meshal Aljehani

  • Rola Madain

  • Mohammad Amin

  • June 26, 2026

  • 0 min

Share

Clinical Report: A Deep Learning Approach for the Automated Identification of Intestinal Metaplasia in Gastric Histopathology Images

Overview

The CNXTGeM framework demonstrates high accuracy in detecting intestinal metaplasia in gastric histopathology images, achieving 100% sensitivity and 99.04% accuracy in a validation cohort.

Background

Intestinal metaplasia (IM) is a precancerous condition that can progress to gastric cancer, making its accurate identification crucial for timely intervention. Traditional methods of detecting IM can be subjective and prone to variability.

Data Highlights

MetricCNXTGeMBaseline ConvNeXt
Accuracy99.04%90.53%
Precision98.08%N/A
Specificity98.11%N/A
F1-score99.03%N/A
Sensitivity100%91.49%

Key Findings

  • CNXTGeM achieved an accuracy of 99.04% in detecting intestinal metaplasia.
  • The model demonstrated 100% sensitivity, significantly reducing missed cases compared to baseline models.
  • External validation on the GasHisSDB dataset yielded an accuracy of 99.34%.
  • Gradient-based visualization techniques confirmed that the model focused on relevant histopathological features.
  • CNXTGeM outperformed baseline models including VGG16, VGG19, DenseNet121, and MobileNetV2.

Clinical Implications

The CNXTGeM framework may enhance the accuracy of intestinal metaplasia detection in clinical settings.

Conclusion

The CNXTGeM framework demonstrates the potential for automated detection of intestinal metaplasia.

Related Resources & Content

  1. Gastric Cancer — Advancements in Deep Learning Techniques for the Pathological Assessment of Gastric Endoscopic Submucosal Dissection Samples
  2. Gastric Cancer — Utilizing Semi-Supervised Deep Learning for the Diagnosis and Assessment of Gastric Atrophy and Intestinal Metaplasia in Pathological Images: A Development and Validation Study
  3. Frontiers in Medicine — Lightweight deep learning model for gastrointestinal precancerous lesion screening with attention enhancement
  4. Gastric Cancer — Utilizing Convolutional Neural Networks for Early Detection of Gastric Cancer via Enhanced Narrow Band Imaging Techniques
  5. ACG Clinical Guideline: Diagnosis and Management of Gastric Premalignant Conditions - PMC
  6. ACG Clinical Guideline: Diagnosis and Management of Gastric Premalignant Conditions - PMC
  7. Effect of Helicobacter pylori eradication on gastric cancer risk in patients with intestinal metaplasia or dysplasia: a meta-analysis of randomized controlled trials - PMC
  8. 25053 504..554
  9. Severity of gastric intestinal metaplasia predicts the risk of gastric cancer: a prospective multicentre cohort study (GCEP) | Gut
  10. OLGA and OLGIM staging systems on the risk assessment of gastric cancer: a systematic review and meta‑analysis of prospective cohorts - PMC
  11. Official journal of the American College of Gastroenterology | ACG
  12. Frontiers | Helicobacterpylori eradication following endoscopic resection might prevent metachronous gastric cancer: a systematic review and meta-analysis of studies from Japan and Korea

Original Source(s)

An interpretable deep learning framework for intestinal metaplasia detection in gastric histopathology images

  • by Alia Al-Mohtaseb, Fahad T. Alotaibi, Salem Alhatamleh, Hatem Malkawi, Amal Alishwait, Ala Meshal Aljehani, Rola Madain, Mohammad Amin

  • June 26, 2026

Related Content

Integrative oncology in colorectal cancer: evidence-based strategies from prevention through survivorship

Misdiagnosis analysis and multidisciplinary collaborative treatment of primary cutaneous diffuse large B-cell lymphoma: case report

Clinical and prognostic implications of pituitary tumor-transforming gene 1 in solid tumors