ASN's Mission

ASN leads the fight to prevent, treat, and cure kidney diseases throughout the world by educating health professionals and scientists, advancing research and innovation, communicating new knowledge, and advocating for the highest quality care for patients.

learn more

Contact ASN

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

email@asn-online.org

202-640-4660

The Latest on Twitter

Kidney Week

Abstract: PO1397

Automatic Artificial Intelligence-Assisted Glomerulosclerosis Analysis in Mice Models with Glomerulopathy

Session Information

Category: Glomerular Diseases

  • 1201 Glomerular Diseases: Fibrosis and Extracellular Matrix

Authors

  • Secher, Thomas, Gubra, Hørsholm, Denmark
  • Salinas, Casper Gravesen, Gubra, Hørsholm, Denmark
  • Christensen, Michael, Gubra, Hørsholm, Denmark
  • Vrang, Niels, Gubra, Hørsholm, Denmark
  • Østergaard, Mette Viberg, Gubra, Hørsholm, Denmark
Background

Glomerulosclerosis (GS) is a hall mark pathological feature in glomerular diseases. In preclinical research, GS is recapitulated in a number of experimental mice models with glomerulopathy, and a cross model characterization of GS would add to our understanding of their translatability to human disease. Here, we report GS quantification using an objective and newly developed automated AI assisted image analysis strategy in three mice models with glomerulopathy.

Methods

AI-assisted GS scoring was performed in three mice models with glomerulopathy and to assess drug treatment effects: 1) Diabetic nephropathy in reninAAV-induced hypertensive uninephrectomized db/db mice (DN/HT). Mice received treatment with vehicle, lisinopril, empagliflozin or combination. 2) I.V. injection of nephrotoxic anti-GBM serum (NTS) and 3) Adriamycin (ADR) in healthy mice. Automatic AI-assisted GS scoring was performed as a two-step process on PAS stained kidney sections. Firstly, segmentation of all glomeruli and next, assignment of a GS score to each glomerulus using trained neutral networks. GS was classified on a five-point scale (GS0-GS4) according to the area of capillary tuft involvement.

Results

The automated AI-assisted scoring performed with high degree of accuracy and allowed for a large number of glomeruli to be evaluated pr. section (>100). We show that the mice models had very distinct GS profiles. The DN/HT model was most severely affected with highest average GS score (DN/HT: 2.5, NTS: 1.76, ADR: 0.68, control: 0.1) and the largest percentage of severely affect glomeruli, GS3+GS4 (DN/HT: 54.1%, NTS: 37.2%, ADR: 16.8%, control: 1.1%). The ADR model was the least affected and NTS was intermediate. AI-assisted scoring was further validated by evaluating the effect of standard of care compounds in the DN/HT model. LIS and the combination treatment with EMPA significantly reduced the average GS score (LIS: 2.1, LIS+EMPA: 1.7).

Conclusion

We here demonstrate an automatic AI-assisted image analysis approach for obtaining GS scores in three mouse models with glomerulopathy, and we show that the models display very distinct GS profiles. An AI-assisted approach allows for rapid and reproducible quantification of GS and, as a tool, could help advance future research.

Funding

  • Commercial Support