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: FR-PO016

Deep Learning Model for Evaluating Histopathology of Acute Renal Tubular Injury

Session Information

Category: Augmented Intelligence, Digital Health, and Data Science

  • 300 Augmented Intelligence, Digital Health, and Data Science

Authors

  • Kim, Hyeongwan, Jeonbuk National University Hospital, Jeonju, Jeollabuk-do , Korea (the Republic of)
  • Jung, Yujin, Jeonbuk National University Hospital, Jeonju, Jeollabuk-do , Korea (the Republic of)
  • Park, Woong, Jeonbuk National University Hospital, Jeonju, Jeollabuk-do , Korea (the Republic of)
  • Yeom, Jihyun, Jeonbuk National University Hospital, Jeonju, Jeollabuk-do , Korea (the Republic of)
  • Lee, Soo jin, Jeonbuk National University Hospital, Jeonju, Jeollabuk-do , Korea (the Republic of)
  • Shin, Yujin, Jeonbuk National University Hospital, Jeonju, Jeollabuk-do , Korea (the Republic of)
  • Kim, Won, Jeonbuk National University Hospital, Jeonju, Jeollabuk-do , Korea (the Republic of)
Background

Acute tubular injury is the most common cause of acute kidney injury (AKI). Histopathological diagnosis may help distinguish between the different types of acute kidney injury and aid in treatment. Currently, no studies have utilized deep learning models to assist in the histopathological diagnosis of acute kidney injury. The aim of this study is to perform histopathological segmentation to identify the four structures of acute renal tubular injury using deep learning models.

Methods

We use a segmentation model (glomerulus, healthy tubules, tubules with cast, and necrotic tubules) to classify tubule-specific injuries after cisplatin treatment. A total of 45 whole slide images with 400 generated patches are used in the segmentation model and 27,478 annotations are created for four classes, namely glomerulus, healthy tubules, necrotic tubules and tubules with cast. A segmentation model was developed utilizing the DeepLabV3 architecture with the MobileNetv3-Large backbone to accurately identify four histopathological structures associated with acute renal tubular injury in mouse PAS-stained samples. In the segmentation model for four structures, the highest Intersection over Union and the Dice coefficient are obtained for the segmentation of the “glomerulus” class, followed by “necrotic tubules,” “healthy tubules,” and “tubules with cast” classes.

Results

The overall performance of the segmentation algorithm for all classes in the test set includes an Intersection over Union of 0.7968 and Dice coefficient of 0.8772. The Dice scores for the glomerulus, healthy tubules, necrotic tubules, and tubules with cast are 91.78 ± 11.09%, 87.37 ± 4.02%, 88.08 ± 6.83%, and 83.64 ± 20.39%, respectively.

Conclusion

The utilization of deep learning in a predictive model has demonstrated promising performance in accurately identifying the histopathological structures of injured renal tubules. These results may provide new opportunities for applying the proposed methods to more effectively evaluate renal pathology in the future.