ASN's Mission

To create a world without kidney diseases, the ASN Alliance for Kidney Health elevates care by educating and informing, driving breakthroughs and innovation, and advocating for policies that create transformative changes in kidney medicine throughout the world.

learn more

Contact ASN

1401 H St, NW, Ste 900, Washington, DC 20005

email@asn-online.org

202-640-4660

The Latest on Twitter

Kidney Week

Abstract: TH-PO562

Development of a Multiple Convolutional Neural Network Facilitated Immunofluorescence Assessment of Glomerular Diseases

Session Information

  • Pathology and Lab Medicine
    November 03, 2022 | Location: Exhibit Hall, Orange County Convention Center‚ West Building
    Abstract Time: 10:00 AM - 12:00 PM

Category: Pathology and Lab Medicine

  • 1700 Pathology and Lab Medicine

Authors

  • Xia, Peng, Peking Union Medical College Hospital, Beijing, Beijing, China
  • Zhang, Xueyuan, Beijing Zhijian Life Technology Co. LTD, Beijing, Beijing, China
  • Wen, Yubing, Peking Union Medical College Hospital, Beijing, Beijing, China
  • Zhang, Baichuan, Beijing Zhijian Life Technology Co. LTD, Beijing, Beijing, China
  • Zhao, Xuesong, Peking Union Medical College Hospital, Beijing, Beijing, China
  • Ren, Fei, Institute of Computing Technology Chinese Academy of Sciences, Beijing, Beijing, China
  • Chen, Limeng, Peking Union Medical College Hospital, Beijing, Beijing, China
Background

Immunofluorescence (IF) tests of renal tissue are important in diagnosing glomerular diseases. We developed a multiple CNN facilitated program to generate suggested IF diagnosis of glomerular diseases.

Methods

A dataset of 2747 IF images of glomerulus, including IgA Nephropathy (IgAN, n=759), Idiopathic Membranous Nephropathy (IMN, n=1050), anti-glomerular basement membrane antibody disease (anti-GBM disease, n=64), Mesangial Proliferative Glomerular Nephritis (MPGN, n=63), Poststreptococcal Glomerulonephritis (PSGN, n=83) and other glomerular diseases (n=728) including Lupus Nephritis, Diabetic Nephropathy and Secondary MN from Peking Union Medical College Hospital were used for training and validation datasets. Another 869 images from 183 patients were used as test dataset. The program included a CNN trained as glomeruli location module and a CNN trained as deposition appearance and location classifier (Figure 1). The performance of the program was evaluated by sensitivity, specificity and F1 score. F1 was computed as 2×True Positive / (2×True Positive+False Positive+False Negative).

Results

The program was accurate common glomerular diseases. The sensitivity of diagnosing suspected IgAN and IMN were 95.9% and 91.1%, and the specificity were 99.1% and 97.1%. The corresponding F1 score were 0.972 and 0.911. The program showed specific diagnostic performance for uncommon glomerular diseases. The specificity of diagnosing suspected anti-GBM disease, MPGN, PSGN and C3 glomerulonephritis were 100%, 100%, 100%, and 98.9% (Table 1).

Conclusion

This multiple CNN facilitated diagnostic program were useful in IF assessment of glomerular diseases.

The proposed program showed good IF diagnostic performance of glomerular diseases
 AccuracySensitivitySpecificityF1 Score
IgAN (n=73)97.8%95.9%99.1%0.972
IMN (45)95.6%91.1%97.1%0.911
Anti-GBM disease (n=14)98.9%85.7%100%0.923
MPGN (n=9)98.4%66.7%100%0.800
PSGN (n=5)98.9%60.0%100%0.750
C3 glomerulonephritis (n=5)98.4%80.0%98.9%0.727
Other glomerular diseases (n=32)92.3%90.6%92.7%0.806

Figure 1 Overview of the CNN trained for deposition appearance and location classification