ASN's Mission

To create a world without kidney diseases, the ASN Alliance for Kidney Health elevates care by educating and informing, driving breakthroughs and innovation, and advocating for policies that create transformative changes in kidney medicine throughout the world.

learn more

Contact ASN

1401 H St, NW, Ste 900, Washington, DC 20005

email@asn-online.org

202-640-4660

The Latest on X

Kidney Week

Abstract: TH-PO033

Deep Learning-Based Quantitative Assessment of Renal Chronicity Indices in Lupus Nephritis

Session Information

Category: Augmented Intelligence, Digital Health, and Data Science

  • 300 Augmented Intelligence, Digital Health, and Data Science

Authors

  • Tu, Tianqi, Peking University First Hospital, Beijing, Beijing, China
  • Tan, Ying, Peking University First Hospital, Beijing, Beijing, China
Background

Lupus nephritis (LN) is a common condition in patients with SLE. Evaluating kidney chronicity indices (CI) is crucial for diagnosing LN and predicting outcomes. Deep learning (DL) has shown promise in improving the workflow of pathologists. In renal pathology, DL has demonstrated high accuracy in identifying various histopathologic lesions. However, existing studies have not adequately addressed the challenges of assessing CI in LN, such as integrating features at different levels and dealing with class imbalances among tissue compartments.

Methods

This study enrolled 141 patients from two distinct cohorts. Training of the CNN model involved annotating 22,410 images from 30 biopsy slides, which were subsequently tested on 111 slides.
The evaluation process encompassed the segmentation of tissue compartments, correlation analysis with pathologists' assessments, and investigating associations with patient outcomes utilizing CoxPH models.

Results

1) The developed pipeline exhibited an accuracy of over 0.91 in identifying six tissue compartments and histopathologic lesions, surpassing previous segmentation performance benchmarks.
2) Consistent with the evaluations by expert pathologists, our pipeline exhibited an agreement of over 0.85 in assessing CI in both cohorts, highlighting its reliability and consistency.
3) The application of the CoxPH model to predict patients' outcomes achieved a score of 0.82 and 0.78 in the two cohorts.

Conclusion

Our study showcases the state-of-the-art performance of our deep learning pipeline in identifying tissue compartments and histopathologic lesions in lupus nephritis. The pipeline demonstrates strong agreement with pathologists' CI assessments and maintains the correlation between CI and patient outcomes. This automated approach has the potential to enhance pathological assessment and disease stratification in lupus nephritis.

Funding

  • Government Support – Non-U.S.