Multi-Task Learning-Based Immunofluorescence Classification of Kidney Disease.

Sai Pan,Xiaofei Wang,Yibing Fu,Jiaona Liu,Mai Xu,Yong Wang,Weicen Liu,Jie Wu,Zhong Yin,Shuwei Duan,Pu Chen,Guangyan Cai,Li Tang,Xiangmei Chen,Lai Jiang,Ning Dai

doi:10.3390/ijerph182010798

Abstract

Chronic kidney disease is one of the most important causes of mortality worldwide, but a shortage of nephrology pathologists has led to delays or errors in its diagnosis and treatment. Immunofluorescence (IF) images of patients with IgA nephropathy (IgAN), membranous nephropathy (MN), diabetic nephropathy (DN), and lupus nephritis (LN) were obtained from the General Hospital of Chinese PLA. The data were divided into training and test data. To simulate the inaccurate focus of the fluorescence microscope, the Gaussian method was employed to blur the IF images. We proposed a novel multi-task learning (MTL) method for image quality assessment, de-blurring, and disease classification tasks. A total of 1608 patients’ IF images were included—1289 in the training set and 319 in the test set. For non-blurred IF images, the classification accuracy of the test set was 0.97, with an AUC of 1.000. For blurred IF images, the proposed MTL method had a higher accuracy (0.94 vs. 0.93, p < 0.01) and higher AUC (0.993 vs. 0.986) than the common MTL method. The novel MTL method not only diagnosed four types of kidney diseases through blurred IF images but also showed good performance in two auxiliary tasks: image quality assessment and de-blurring.

Highlights

Chronic kidney disease (CKD) is a non-communicable disease that contributes to high morbidity and mortality worldwide, including in China [1,2]
Immunofluorescence (IF) is one of the most important methods used for the diagnosis of these four kidney diseases
The accuracy of diagnosing kidney disease using seven types of IF images was as high as 0.97, which suggested that convolutional neural networks (CNN) might identify distinctive IF images and image features that the human eye misses

Summary

Introduction

Chronic kidney disease (CKD) is a non-communicable disease that contributes to high morbidity and mortality worldwide, including in China [1,2]. IgA nephropathy (IgAN) are the most common forms of primary glomerulonephritis, whereas lupus nephropathy (LN) and diabetes nephropathy (DN) are the most common secondary types of glomerulonephritis in China [6,8]. These four pathological types contribute to more than 60% of all cases of CKD [9]. It is imperative to identify and diagnose these four kidney diseases. By 2017, there were only 3.94 and 4.81 pathologists per 100,000 people in the United States and Canada, respectively [10]

Methods

Results

Discussion

Conclusion