Biology subjects

Han, S. S.

Publications and source records attributed to Han, S. S..

4 recordsLinked to original sources

bioRxiv2023

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

BackgroundReference histomorphometric data of healthy human kidneys are largely lacking due to laborious quantitation requirements. Correlating histomorphometric features with clinical parameters through machine learning approaches can provide valuable information about natural population variance. To this end, we leveraged deep learning, computational image analysis, and feature analysis to investigate the relationship of histomorphometry with patient age, sex, and serum creatinine (SCr) in a multinational set of reference kidney tissue sections. MethodsA panoptic segmentation neural network was developed and used to segment viable and sclerotic glomeruli, cortical and medullary interstitia, tubules, and arteries/arterioles in the digitized images of 79 periodic acid-Schiff-stained human nephrectomy sections showing minimal pathologic changes. Simple morphometrics (e.g., area, radius, density) were quantified from the segmented classes. Regression analysis aided in determining the relationship of histomorphometric parameters with age, sex, and SCr. ResultsOur deep-learning model achieved high segmentation performance for all test compartments. The size and density of nephrons and arteries/arterioles varied significantly among healthy humans, with potentially large differences between geographically diverse patients. Nephron size was significantly dependent on SCr. Slight, albeit significant, differences in renal vasculature were observed between sexes. Glomerulosclerosis percentage increased, and cortical density of arteries/arterioles decreased, as a function of age. ConclusionsUsing deep learning, we automated precise measurements of kidney histomorphometric features. In the reference kidney tissue, several histomorphometric features demonstrated significant correlation to patient demographics and SCr. Deep learning tools can increase the efficiency and rigor of histomorphometric analysis. Significance statementAlthough the importance of kidney morphometry is well explored in disease contexts, the definition of variance in reference tissue is not. Advancements in digital and computational pathology have rendered quantitative analysis of unprecedented tissue volumes via the single press of a button. The authors leverage the unique benefits of panoptic segmentation to perform the largest ever quantitation of reference kidney morphometry. Regression analysis identified several kidney morphometric features that varied significantly with patient age and sex, and the results suggested that the set size of nephrons might depend more intricately on creatinine than previously thought.

cell biology↗

bioRxiv2023

Spatial proteomics of human diabetic kidney disease, from health to class III

Aims/HypothesisDiabetic kidney disease (DKD) remains a significant cause of morbidity and mortality in people with diabetes. Though animal models have taught us much about the molecular mechanisms of DKD, translating these findings to human disease requires greater knowledge of the molecular changes caused by diabetes in human kidneys. Establishing this knowledge base requires building carefully curated, reliable, and complete repositories of human kidney tissue, as well as tissue proteomics platforms capable of simultaneous, spatially resolved examination of multiple proteins. MethodsWe used the multiplexed immunofluorescence platform CO-Detection by indexing (CODEX) to image and analyze the expression of 21 proteins in 23 tissue sections from 12 individuals with diabetes and healthy kidneys (DM, 5 individuals), DKD classes IIA, and IIB (2 individuals per class), IIA-B intermediate (2 individuals), and III (one individual). ResultsAnalysis of the 21-plex immunofluorescence images revealed 18 cellular clusters, corresponding to 10 known kidney compartments and cell types, including proximal tubules, distal nephron, podocytes, glomerular endothelial and peritubular capillaries, blood vessels, including endothelial cells and vascular smooth muscle cells, macrophages, cells of the myeloid lineage, broad CD45+ inflammatory cells and the basement membrane. DKD progression was associated with co-localized increase in collagen IV deposition and infiltration of inflammatory cells, as well as loss of native proteins of each nephron segment at variable rates. Compartment-specific cellular changes corroborated this general theme, with compartment-specific variations. Cell type frequency and cell-to-cell adjacency highlighted (statistically) significant increase in inflammatory cells and their adjacency to tubular and SMA+ cells in DKD kidneys. Finally, DKD progression was marked by substantial regional variability within single tissue sections, as well as variability across patients within the same DKD class. The sizable intra-personal variability in DKD severity impacts pathologic classifications, and the attendant clinical decisions, which are usually based on small tissue biopsies. Conclusions/InterpretationsHigh-plex immunofluorescence images revealed changes in protein expression corresponding to differences in cellular phenotypic composition and microenvironment structure with DKD progression. This initial dataset demonstrates the combined power of curated human kidney tissues, multiplexed immunofluorescence and powerful analysis tools in revealing pathophysiology of human DKD.

molecular biology↗

bioRxiv2021

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

BackgroundImage-based machine learning tools hold great promise for clinical applications in nephropathology and kidney research. However, the ideal end-users of these computational tools (e.g., pathologists and biological scientists) often face prohibitive challenges in using these tools to their full potential, including the lack of technical expertise, suboptimal user interface, and limited computation power. MethodsWe have developed Histo-Cloud, a tool for segmentation of whole slide images (WSIs) that has an easy-to-use graphical user interface. This tool runs a state-of-the-art convolutional neural network (CNN) for segmentation of WSIs in the cloud and allows the extraction of features from segmented regions for further analysis. ResultsBy segmenting glomeruli, interstitial fibrosis and tubular atrophy, and vascular structures from renal and non-renal WSIs, we demonstrate the scalability, best practices for transfer learning, and effects of dataset variability. Finally, we demonstrate an application for animal model research, analyzing glomerular features in murine models of aging, diabetic nephropathy, and HIV associated nephropathy. ConclusionThe ability to access this tool over the internet will facilitate widespread use by computational non-experts. Histo-Cloud is open source and adaptable for segmentation of any histological structure regardless of stain. Histo-Cloud will greatly accelerate and facilitate the generation of datasets for machine learning in the analysis of kidney histology, empowering computationally novice end-users to conduct deep feature analysis of tissue slides.

pathology↗

bioRxiv2021

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

BackgroundPodocyte depletion is an established indicator of glomerular injury and predicts clinical outcomes. The semi-quantitative nature of existing podocyte estimation methods or podometrics hinders incorporation of such analysis into experimental and clinical pathologic workflows. Computational image analysis offers a robust approach to automate podometrics through objective quantification of cell and tissue structure. Toward this goal, we developed PodoCount, a computational tool for quantitative analysis of podocytes, and validated the generalizability of the tool across a diverse dataset. MethodsPodocyte nuclei and glomerular boundaries were labeled in murine whole kidney sections, n = 135, from six disease models and human kidney biopsies, n = 45, from diabetic nephropathy (DN) patients. Digital whole slide images (WSIs) of tissues were then acquired. Classical image analysis was applied to obtain podocyte nuclear and glomerular morphometrics. Statistically significant morphometric features, which correlated with each murine disease, were identified. Engineered features were also assessed for their ability to predict outcomes in human DN. PodoCount has been disbursed for other researchers as an open-source, cloud-based computational tool. ResultsPodoCount offers highly accurate quantification of podocytes. Engineered podometric features were benchmarked against routine glomerular histopathology and were found to be significant predictors of disease diagnosis, proteinuria level, and clinical outcomes. ConclusionsPodoCount offers high quantification performance in diverse murine disease models as well as in human DN. Resultant podometric features offers significant correlation with associated metadata as well as outcome. Our cloud-based end-user tool will provide a standardized approach for podometric analysis from gigapixel size WSIs in basic research and clinical practice.

pathology↗