# Computational Imaging of Renal Structures for Diagnosing Diabetic Nephropathy.

> **NIH NIH R01** · UNIVERSITY OF FLORIDA · 2024 · $708,664

## Abstract

Summary
Diabetic kidney disease (DKD) accounts for $42 billion in annual Medicare spending, and >50% of end-stage
kidney disease (ESKD) cases. The Renal Pathology Society endorses the use of tissue morphometric features
of DKD with discrete classification based on the peak visual semi-quantitative measure of renal tissue
morphometry. These features may lack sensitivity to correlate with clinical biometrics (e.g., eGFR) measured at
the time of biopsy or during follow-up. A multimodal evidence-based quantitative method is required to deliver
continuous scoring by engaging molecular information along with digital histopathology, and summarizing the
contributions of each pixel of a renal tissue histopathology image using quantitative morphometry and omics.
The investigator team has pioneered the unbiased quantitative morphometry of DKD over the last five years in
their existing R01, conducting extensive quantification of novel image pixel features with clinical significance,
and focusing on how computational artificial intelligence (AI) improves precision and accuracy, outperforming
existing diagnostic standards. An opportunity exists to translate their developed tools in a clinically meaningful
form that aids a pathologist in biopsy assessment or a nephrologist in therapy selection. Capitalizing on their
extensive work in the field of DKD digital pathology, the investigator team will: 1) translate their developed digital
pathology tools to clinicians’ desks in the next five years, 2) extend digital pathology by integrating spatial
molecular features to reveal hidden digital image biomarkers, and 3) provide clinicians with additional useful
metrics for biopsy assessment to augment the current treatment strategy of DKD. Recently, this team has
significantly advanced spatial anchoring of cell types and cell states (e.g., injury, recovery, adaptation) in
brightfield histology using paired data from spatial transcriptomics. This effort will be extended to the single-cell
level in this application. The team has also delivered to the community the first cloud application of quantitative
digital pathology tissue assessment via a single-click, promoting FAIR (findable, accessible, interoperable,
reusable) data principles. Based on these preliminary efforts, this renewal application will implement a pipeline
for the clinical use of computational pathology in DKD. The central hypothesis is that computational pathology of
DKD, integrated with spatial omics, offers transformative tools for DKD diagnostic classification, prognosis, and
therapy selection. In particular, the team will: 1) Develop an end-user cloud software for clinical DKD biopsy
assessment using continuous risk scoring; 2) Determine the digital histopathology image pixel features
corresponding to molecular cell processes with clinical significance; 3) Test whether image pixel and spatial
transcriptomics quantitative vectors of glomerular hypertension predict progression and optimal DKD therapy.
Wi...

## Key facts

- **NIH application ID:** 10978599
- **Project number:** 2R01DK114485-06
- **Recipient organization:** UNIVERSITY OF FLORIDA
- **Principal Investigator:** Michael Thomas Eadon
- **Activity code:** R01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2024
- **Award amount:** $708,664
- **Award type:** 2
- **Project period:** 2018-09-15 → 2028-07-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10978599

## Citation

> US National Institutes of Health, RePORTER application 10978599, Computational Imaging of Renal Structures for Diagnosing Diabetic Nephropathy. (2R01DK114485-06). Retrieved via AI Analytics 2026-05-25 from https://api.ai-analytics.org/grant/nih/10978599. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*