Chronic Kidney Disease Detection

Jaya Jeswani(B) , Mohammed Multazim Ansari, Rushikesh Durgade,

and Alisha Fatima Ansari

Department of Information Technology, Xavier Institute of Engineering, University of

Mumbai, Mumbai, Maharashtra, India
jaya.j@xavier.ac.in

Abstract. The impact of technological advancement, particularly

machine learning, on health can be seen in the efficient analysis of differ-
ent chronic diseases that allows for more precise diagnosis and effective
treatment. People aged 60 and above are most affected by kidney disease,
a serious chronic condition linked to ageing, hypertension, and diabetes.
Early diagnosis of CKD enables patients to receive immediate treatment,
which slows the disease’s further development. This study employs the
machine learning techniques of artificial neural networks, support vector
machines, and k-Nearest Neighbour to identify CDK early. The signifi-
cance of detecting these frequently fatal illnesses reflects the significance
of AI. These four processes of image pre-processing, feature extraction,
classification, and diagnosis are used to identify the type of disease. Con-
volution Neural Network (CNN), which has a number of prediction-based
layers, is used for categorisation and image pre-processing to improve the
image’s quality. At the very end, the user is encouraged to get a cure.

Keywords: Kidney · Disease · Machine Learning · Image Processing ·

Chronic Kidney Disease

1 Introduction
Chronic renal disease is an irreparable kidney ailment that raises the risk of a
wide range of illnesses, including heart failure, anaemia, and bone disease. The
kidneys are incredibly versatile. However, kidney damage won’t be immediately
obvious due to symptoms. Patients frequently don’t have symptoms until the
condition is nearly terminal. Avoiding symptoms is a treatment option for some
renal diseases. By restoring a few kidney functions, it aids patients in preventing
the condition from getting worse.
Dialysis and kidney transplantation are two of the main treatments for end-
stage kidney disease, particularly in cases with CKD. More than 800 million
people worldwide suffer from chronic kidney disease, which is a progressive con-
dition [1]. Due to the high cost of treatment, only 10% of patients receive dialysis
or a kidney transplant, and it is predicted that the number of kidney failure cases
will rise disproportionately in developing nations like China and India, where the
elderly population is growing [2].
c The Author(s) 2023
S. Tamane et al. (Eds.): ICAMIDA 2022, ACSR 105, pp. 4–10, 2023.
https://doi.org/10.2991/978-94-6463-136-4_2
 Chronic Kidney Disease Detection 5

Due to a lack of glomeruli filters, also known as nephrons, patients with

acquired immunodeficiency syndrome (AIDS) experience increased complica-
tions from kidney illness. Early detection and management of CKD develop-
ment are crucial. Healthcare has benefited greatly from the burgeoning interest
in automated diagnosis and the quick development of machine learning tech-
niques. Even though several studies have divided CKD into different stages using
machine learning approaches.
The technical support tools that are based on data must be able to swiftly,
accurately, and affordably support the decision-making process in the early diag-
nostics. They enable the patient to receive therapy for the disease before it
advances to a level where there is no turning back since they shorten the time
needed for diagnosis. The three main categories of machine learning are rein-
forcement learning, unsupervised learning, and supervised learning. The most
popular type of machine learning applied in medical research is supervised learn-
ing. In order to lower the mortality rate and costs for the healthcare system, this
research intends to create and deploy a machine learning model that, based on
data like ultrasound from clinical laboratories, allows forecasting the possibility
of diagnosing CKD in its early stages.

2 Literature Review

Chronic Kidney Disease Prediction using Data Mining research paper published
at International Conference on Emerging Trends at 2020 proposed a system that
makes use of back propagation neural networks and the Random Forest method,
among other data mining approaches. Here, they evaluate the two algorithms
and discover that the Back-Propagation technique produces the greatest results
since it makes use of the Feedforward Neural Network, a supervised learning
network [1].
Development of Chronic Kidney Disease Prediction Using Machine Learn-
ing research paper published at International Conference on Intelligent Data
Communication Technologies on 2019 proposed creating a system for predicting
the presence of CKD using machine learning methods including Naive Bayes,
K-Nearest Neighbour, Logistic Regression, Decision Tree, Random Forest, and
Multi-Layer Perceptron Algorithm. These are used, and the effectiveness of each
is evaluated in relation to the outcomes for accuracy, precision, and recall. The
system is finally implemented using Random Forest [2].
M. N. Amin, A. Al Imran and F. T. Johora et al. analyse model performance
on real (imbalanced) data and model performance on oversampled (balanced)
data using logistic regression and feed forward neural networks. Feed forward
neural networks showed the best results for both real and oversampled data,
with 0.99 Recall, 0.97 Precision, 0.99 F1-Score and 0.99 AUC score [3].
A dataset from the UCI machine learning repository containing data on
roughly 400 patients is used by Kayaalp et al. to study kidney disease using
hybrid classification approaches. They choose the most pertinent feature in the
dataset using the relief and gain ratio approach and the support vector machine
6 P. J. Jeswani et al.

and KNN classifier. They get to the conclusion that, in terms of f-measure, pre-
cision, and contrast matrix, the KNN approach outperforms other algorithms
for a set of features [4].
To identify CKD, A. Salekin and J. Stankovic tested three classifiers: neural
network, random forest, and K-nearest neighbours. They made use of a dataset
from UCI with 400 patients and 24 attributes. The attributes that accurately
identify this disease have been discovered by the use of the wrapper approach
in a feature reduction study. They may predict the presence of CKD with a.98
F1 and a 0.11 RMSE by taking factors like albumin, specific gravity, diabetes
mellitus, haemoglobin, and hypertension into account [5].

3 Problem Statement

Chronic kidney disease (CKD) is now recognised as a severe threat to society’s

health. Regular laboratory examinations are able to identify chronic kidney dis-
ease, and there are also treatments that can halt the development of the condi-
tion, slow down its progression, lessen complications brought on by a decreased
Glomerular Filtration Rate (GFR), reduce the risk of cardiovascular disease,
and enhance survival and quality of life. CKD can be brought on by drinking
little water, smoking, eating poorly, sleeping too little, and many other factors.
Globally, this illness affected 753 million people in 2016, 336 million of whom
were men and 417 million of them were women. Renal failure can occasionally
occur since the condition is typically only detected when it is advanced.
End-stage kidney disease (ESKD), which requires kidney replacement ther-
apy, can develop in patients with chronic kidney disease (CKD) who gradually
lose kidney function (KRT). Prompt management may increase the quality of life
for CKD patients who have a high risk of developing ESKD and may also reduce
morbidity, mortality, and healthcare costs related to KRT. Because the disease
generally progresses silently, a solid prediction model for the risk of ESKD at the
early stage of CKD can be clinically crucial. This model is projected to assist
physicians in providing high-risk patients with more tailored treatment options,
improving overall prognosis and reducing the cost burden of this condition.

4 Proposed System

This research paper presents a proof-of-concept study with the major goal to
establish ML models for predicting the risk of ESKD on a Chinese CKD dataset.
The ML models were trained and tested based on easily obtainable variables,
including the baseline characteristics and routine ultra sound images. Results
obtained from this study suggest not only the feasibility of ML models in per-
forming this clinically critical task, but also the potential in facilitating person-
alised medicine. Instead of doing multiple clinical test such as blood test, urine
test, this system will work on ultra sound report of kidney which will process
the report and give the result of kidney disease is present or not.
 Chronic Kidney Disease Detection 7

The below diagram shows the block of the system. The dataset contains the
ultrasound images of the patients. The data processing and feature selection
block process the image through machine learning algorithm and forward the
analysed data to the classification section. The classification section verifies the
analysed data by using CNN method and provide the actual outcome of the
report.

Fig. 1. Block diagram of chronic kidney disease prediction system using CNN [8].

5 Methodology

There are three stages in the intricate block diagram in Fig. 2. The primary
responsibility of the initial stage of data collection and transmission is to collect
data from ultrasonic pictures from mobile devices, etc., in the selected size and
resolution. Pictures from the dataset for the training set the model architecture
was first built with exceptional accuracy in mind.
The training dataset was used in the third step of the procedure, and numer-
ous images were then used to test or validate the model. Finished by adding the
model to a web page for use. In order to take the proper action, the output sys-
tem data will identify whether the symptoms are showing up on the ultrasound
report. We’ll take the digital camera image that needs to be uploaded and run
the CNN algorithm to determine whether the disease is present or not.
8 P. J. Jeswani et al.

Fig. 2. The flow chart of chronic kidney disease prediction system using CNN [8].

5.1 CNN Model

The term “Convolution” in CNN denotes the mathematical function of con-

volution which is a special kind of linear operation wherein two functions are
multiplied to produce a third function which expresses how the shape of one
function is modified by the other. In simple terms, two images which can be
represented as matrices are multiplied to give an output that is used to extract
features from the image. There are two main parts to a CNN architecture.

1. A convolution tool that separates and identifies the various features of the
image for analysis in a process called as Feature Extraction.
2. The network of feature extraction consists of many pairs of convolutional or
pooling layers.
3. A fully connected layer that utilizes the output from the convolution process
and predicts the class of the image based on the features extracted in previous
stages.
 Chronic Kidney Disease Detection 9

4. This CNN model of feature extraction aims to reduce the number of features
present in a dataset. It creates new features which summarizes the existing
features contained in an original set of features. There are many CNN layers
as shown in the CNN architecture diagram.

5.2 Overview
In the ultrasound image, the characteristics of the normal kidney are in the shape
of an oval composed of the renal cortex, which shows low echo compared to the
liver, and the renal sinus, which shows high echo. The boundary between the
cortex and the renal sinus is clear, and a high echo in the centre is visible. On the
other hand, in the kidney that is continuously damaged, the ultrasonic echo of
the renal cortex increases due to fibrosis. As a result, the boundary between the
brightened renal cortex and the renal cortex is unclearly observed. In addition,
due to the decline in function, the size decreases, and kidney atrophy is observed
[14]. Because these features are used in the diagnosis of chronic kidney disease,
the ROI area was set to 50 × 50 and set to 3 locations. The Fig. 4 shows non-
chronic kidney disease ultrasound and 5 shows chronic kidney disease ultrasound
image.

Fig. 3. The overview of chronic kidney disease prediction system using CNN [8].

Fig. 4. NCKD Fig. 5. CKD

10 P. J. Jeswani et al.

6 Conclusion
Early diagnosis and treatment of chronic kidney disease are possible, but as
the condition worsens, recovery is unattainable. The use of dialysis or renal
replacement therapy is ultimately required. To put it another way, it’s critical to
identify and manage chronic renal illness as soon as possible. Using information
such kidney size and internal echo characteristics, ultrasound is used to exam-
ine information on the degree of inflammation when diagnosing kidney cancer,
inflammatory diseases, nodular diseases, chronic renal disease, etc.

References
1. Csaba P. Kovesdy, “Epidemiology of chronic kidney disease: an update 2022”,
National Library of Medicine, ncbi.nlm.nih.gov, April 2022.
2. Global Facts about kidney, kidney. Org, September 2021.
3. J. Snegha, “Chronic Kidney Disease Prediction using Data Mining”, International
Conference on Emerging Trends, 2020.
4. Baisakhi Chakraborty, “Development of Chronic Kidney Disease Prediction Using
Machine Learning”, International Conference on Intelligent Data Communication
Technologies, 2019.
5. M. N. Amin, A. Al Imran, and F. T. Johora, “Classification of Chronic Kidney
Disease using Logistic Regression, Feedforward Neural Network and Wide Deep
Learning,” 2018 Int. Conf. Innov. Eng. Technol. ICIET 2018, pp. 1–6, 2019, https://
doi.org/10.1109/CIET.2018.8660844.
6. Kayaalp F, Basarslan MS and Polat K, “A hybrid classification example in describ-
ing chronic kidney disease”, IEEE Electric Electronics, Computer Science, Biomed-
ical Engineerings’ Meeting (EBBT), (2018), pp. 1–4.
7. A. Salekin and J. Stankovic, “Detection of chronic kidney disease and selecting
important predictive attributes,” in Healthcare Informatics (ICHI), 2016 IEEE
International Conference On, 2016.
8. Phung and Ree,“A High-Accuracy Model Average Ensemble of Convolutional Neu-
ral Networks for Classification of Cloud Image Patches on Small Datasets”, in MDPI
Journal On October 2019.

Open Access This chapter is licensed under the terms of the Creative Commons
Attribution-NonCommercial 4.0 International License (http://creativecommons.org/
licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation,
distribution and reproduction in any medium or format, as long as you give appro-
priate credit to the original author(s) and the source, provide a link to the Creative
Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the
chapter’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the chapter’s Creative Commons license and
your intended use is not permitted by statutory regulation or exceeds the permitted
use, you will need to obtain permission directly from the copyright holder.