Back
The distribution of glomeruli in the kidney can be either in the superficial cortex, associated with short-looped nephrons (SLN), or in the deep cortex, associated with long-looped nephrons (LLN). Previous research has shown that LLN tend to exhibit greater glomerular hypertrophy and increased susceptibility to progressive changes. One common pathological feature in diabetic nephropathy (DN) is the alteration of mesangial morphology. This study uses artificial intelligence (AI)-based methods to compare mesangial morphomics between SLN and LLN in rat DN samples.
Wholemounts of kidney sections stained with periodic acid-Schiff (PAS) were obtained from two groups: control rats (CONT, n=6) and rats with high salt plus streptozotocin-induced DN (DN, n=8). Our pre-trained AI algorithms were used to accurately segment SLN and LLN. Subsequently, custom AI algorithms based on CellProfiler were developed to quantify mesangial morphomics.
In normal kidneys, mesangium within LLN showed larger area, higher intensity, increased texture contrast, greater texture variance, and more complex branching patterns compared to mesangium in SLN. These findings suggest a denser matrix, higher and non-uniform intensity, and more complex branching structure in the mesangium of LLN. When comparing DN with CONT, a significant 59.4% increase in mesangial area was observed. In DN, mesangial morphomics in SLN showed increased texture contrast, texture entropy, texture variance, and more branching patterns compared to SLN in CONT. However, the morphomics of the LLN remained similar between the two groups.
In the context of diabetic nephropathy, SLN glomeruli show more pronounced mesangial changes compared to LLN. Our AI-driven platform serves as a valuable tool for comparative analysis of diverse morphomic characteristics between SLN and LLN glomeruli.