Kidney Week

Abstract: SA-PO299

Exploring the Subtle, Novel, and Early Kidney Pathological Changes in Diabetic Nephropathy Using Clustering with Deep Learning

Session Information

• Diabetic Kidney Disease: Clinical Pathology, Diagnostic and Treatment Advances
  October 26, 2024 | Location: Exhibit Hall, Convention Center
  Abstract Time: 10:00 AM - 12:00 PM

Category: Diabetic Kidney Disease

• 702 Diabetic Kidney Disease: Clinical

Authors

• Yabe, Tomohisa, Kanazawa Ika Daigaku, Kahoku-gun, Ishikawa, Japan

Tomohisa Yabe, MD

• Fujii, Ai, Kanazawa Ika Daigaku, Kahoku-gun, Ishikawa, Japan

Ai Fujii, MD, PhD

• Okada, Keiichiro, Kanazawa Ika Daigaku, Kahoku-gun, Ishikawa, Japan

Keiichiro Okada, MD, PhD

• Hayashi, Norifumi, Kanazawa Ika Daigaku, Kahoku-gun, Ishikawa, Japan

Norifumi Hayashi, MD

• Fujimoto, Keiji, Kanazawa Ika Daigaku, Kahoku-gun, Ishikawa, Japan

Keiji Fujimoto, MD, PhD

• Yokoyama, Hitoshi, Kanazawa Ika Daigaku, Kahoku-gun, Ishikawa, Japan

Hitoshi Yokoyama, MD, PhD

• Furuichi, Kengo, Kanazawa Ika Daigaku, Kahoku-gun, Ishikawa, Japan

Kengo Furuichi, MD, PhD

Background

As diabetes mellitus (DM) is a leading cause of CKD, early diagnosis of diabetic kidney disease is essential to decrease the number of chronic kidney disease (CKD). However, the pathological changes occurred in early stages of diabetic nephropathy have not been fully elucidated yet.

Methods

Nephrectomized kidneys (partial/total) in Kanazawa Medical University from 1998 to 2019 were used in this study. We performed invariant information clustering (IIC)-based clustering on glomerular images obtained from nephrectomized kidneys of patients with and without diabetes. Visualizing techniques (gradient-weighted class activation mapping (Grad-CAM) and generative adversarial networks (GAN)) were also used to identify the novel and early pathological changes on light microscopy in diabetic nephropathy.

Results

Overall, 13,251 glomerular images (7,799 images from diabetes cases and 5,542 images from non-diabetes cases) obtained from 45 patients were clustered into 10 clusters by IIC. Diabetic clusters that mainly contained glomerular images from diabetes cases (Clusters 0, 1, and 2) and non-diabetic clusters that mainly contained glomerular images from non-diabetes cases (Clusters 8 and 9) were distinguished in the t-distributed stochastic neighbor embedding (t-SNE) analysis. Grad-CAM demonstrated that the outer portions of glomerular capillaries in diabetic clusters had characteristic lesions. Cycle-GAN showed that compared to Bowman’s space, smaller glomerular tufts was a characteristic lesion of diabetic clusters.

Conclusion

The findings in this study would be the novel and early pathological changes on light microscopy in diabetic nephropathy and could be key to its early diagnosis to reduce the number of patients with CKD.

Funding

• Government Support – Non-U.S.