P. Sarder

ComPRePS: Unlocking Scalable AI Analysis for Computational Renal Pathology

ComPRePS improves renal pathology analysis by automating WSI processing and AI-assisted interpretation, enhancing precision and interobserver agreement in key lesion assessments.

Correlating Deep Learning-Based Automated Reference Kidney Histomorphometry with Patient Demographics and Creatinine

This paper uses deep learning to automate kidney histomorphometry and finds significant correlations between morphometric features and patient age, sex, and creatinine levels.

APOL1 kidney risk variants in glomerular diseases modeled in transgenic mice

This paper uses two transgenic mouse models to study APOL1 risk variants' effects on glomerular disease, identifying shared and distinct transcriptional changes in podocytes.

PKR activation-induced mitochondrial dysfunction in HIV-transgenic mice with nephropathy

This study shows that PKR inhibition reduces mitochondrial dysfunction and ameliorates HIV-associated nephropathy in transgenic mice using single-nucleus and bulk RNA-seq.

NAD metabolism modulates inflammation and mitochondria function in diabetic kidney disease

This study shows that nicotinamide riboside (NR) improves mitochondrial function and reduces inflammation in diabetic kidney disease by enhancing NAD metabolism and inhibiting the cGAS-STING pathway.

Structural Knowledge Transfer of Panoptic Kidney Segmentation to Other Stains, Organs, and Species

The paper trains a panoptic segmentation network on kidney biopsy images and shows its ability to generalize across stains, species, and organ types.

Generative modeling of histology tissue reduces human annotation effort for segmentation model development.

This paper uses datasetGAN to reduce annotation effort for glomeruli segmentation, showing it requires less annotation than traditional methods while maintaining high performance.

Endothelial Cell-Specific Molecule-1 Inhibits Albuminuria in Diabetic Mice

This study shows that Esm-1 reduces albuminuria and podocyte injury in diabetic mice, possibly through inhibiting interferon signaling.

A tool for federated training of segmentation models on whole slide images.

The paper demonstrates that federated learning can train segmentation models on whole slide images with performance comparable to centralized training across multiple institutions.

A user-friendly tool for cloud-based whole slide image segmentation, with examples from renal histopathology

Histo-Cloud is a cloud-based tool for whole slide image segmentation that simplifies access for non-experts, demonstrating its use in renal histopathology.

PodoCount: A robust, fully automated whole-slide podocyte quantification tool

PodoCount quantifies podocytes in kidney tissues with high accuracy, showing strong correlation with disease diagnosis and clinical outcomes in both murine models and human diabetic nephropathy.