Artificial intelligence assists identification and pathologic classification of glomerular lesions in patients with diabetic nephropathy

Abstract

Background: Glomerular lesions are the main injuries of diabetic nephropathy (DN) and are used as a crucial index for pathologic classification. Manual quantification of these morphologic features currently used is semi-quantitative and time-consuming. Automatically quantifying glomerular morphologic features is urgently needed.

Methods: A series of convolutional neural networks (CNN) were designed to identify and classify glomerular morphologic features in DN patients. Associations of these digital features with pathologic classification and prognosis were further analyzed.

Results: Our CNN-based model achieved a 0.928 F1-score for global glomerulosclerosis and 0.953 F1-score for Kimmelstiel-Wilson lesion, further obtained a dice of 0.870 for the mesangial area and F1-score beyond 0.839 for three glomerular intrinsic cells. As the pathologic classes increased, mesangial cell numbers and mesangial area increased, and podocyte numbers decreased (p for all < 0.001), while endothelial cell numbers remained stable (p = 0.431). Glomeruli with Kimmelstiel-Wilson lesion showed more severe podocyte deletion compared to those without (p < 0.001). Furthermore, CNN-based classifications showed moderate agreement with pathologists-based classification, the kappa value between the CNN model 3 and pathologists reached 0.624 (ranging from 0.529 to 0.688, p < 0.001). Notably, CNN-based classifications obtained equivalent performance to pathologists-based classifications on predicting baseline and long-term renal function.

Conclusion: Our CNN-based model is promising in assisting the identification and pathologic classification of glomerular lesions in DN patients.

Keywords: Artificial intelligence; Diabetic nephropathy; Glomerulus; Identification; Pathology.

Fig. 1

Schematic illustration of our CNN model for identifying glomerular morphological features. A The training procedure of EfficientNet architecture for identifying different glomeruli types. B The training procedure of U-Net architecture for segmenting mesangial area and the verification procedure of our prior V-Net architecture for identifying three glomerular intrinsic cells. Scale bars mean 750 μm in WSIs and 50 μm in single glomerulus images

Fig. 2

Our CNN architectures identify intraglomerular features. Original images derived directly from WSI slides (A NOA glomerulus, B KW glomerulus). Prediction images (C, D) describe the predicted results of intraglomerular features from the original images (cyan: Bowman capsules, red: podocytes, blue: mesangial cells, green: endothelial cells, yellow: mesangial regions). Scale bar: 50 μm

Fig. 3

Violin plots depict the distribution of each intraglomerular feature between NOA and KW glomeruli from a pooled 11,188 midsection glomerular image (NOA: 8181, KW: 3007). M: mesangial, E: endothelial, P: podocytes. The purple line: the median value, the orange line: the interquartile range. Inter-group comparisons were performed by Mann Whitney U test

Fig. 4

Intraglomerular lesions within one patient are not parallel to each other. The common presentation showed that a patient in class III had Kimmelstiel–Wilson (KW) lesions (A) accompanied by severe mesangial expansion (B). While in uncommon presentation, a patient in class III had a KW lesion (C) accompanied by mild mesangial expansion (D). Cyan: Bowman capsule; red: podocytes (P); blue: mesangial cells (M); green: endothelial cells (E); yellow: mesangial area; asterisk, KW lesions. M^f: mesangial area fraction. Scale bar: 50 μm