PARADIGMA BARU PENDIDIKAN MATEMATIKA DAN APLIKASI ONLINE INTERNET PEMBELAJARAN


Vivin Umrotul M. Maksum, Dian C. Rini Novitasari, Abdulloh Hamid 

Image X-Ray Classification for COVID-19 Detection Using GCLM-ELM 

 

 

CONTACT:  

Dian C. Rini Novitasari,           diancrini@uinsby.ac.id     Department of Mathematics, UIN Sunan Ampel 

Surabaya, Surabaya 60237, Indonesia  

 

       The article can be accessed here. https://doi.org/10.15642/mantik.2021.7.1.74-85 

  Image X-Ray Classification for COVID-19 Detection using 

GLCM-ELM 
 

 

Vivin Umrotul M. Maksum1, Dian C. Rini Novitasari2*, Abdulloh Hamid3   

1,2,3Department of Mathematics, UIN Sunan Ampel Surabaya, Surabaya, Indonesia 

 

 
 

Article history: 

Received Feb 10, 2019 

Revised May 21, 2021 

Accepted Jun 3, 2021 

 

 

Kata Kunci:  

COVID-19, Citra X-

Ray, CAD, GLCM, ELM 

 

Abstrak. COVID-19 merupakan penyakit atau virus yang baru-baru ini 

menyebar hampir di seluruh dunia. Penyakit ini juga telah memakan 

banyak korban karena virus ini terkenal mematikan. Pemeriksaan dapat 

dilakukan menggunakan chest X-Ray karena biaya yang dikeluarkan 

untuk chest X-Ray lebih murah dibandingkan dengan tes swab dan 

PCR. Pada penelitian ini data yang digunakan adalah data citra chest X-

Ray. Citra chest X-Ray dapat diidentifikasi menggunakan Computer-

Aided Diagnosis System dengan memanfaatkan klasifikasi machine 

learning. Langkah awal yang dilakukan adalah tahap preprocessing, 

serta ekstraksi fitur menggunakan Gray Level Co-Occurrence Matrix 

(GLCM). Hasil dari proses ekstraksi fitur tersebut akan digunakan pada 

tahap klasifikasi. Proses klasifikasi yang digunakan adalah Extreme 

Learning Machine (ELM). Extreme Learning Machine (ELM) 

merupakan salah satu jaringan saraf tiruan dengan umpan maju 

(feedforward) yang mana memiliki satu lapisan tersembunyi yang 

disebut dengan Single Hidden Layer Feedforward Neural Networks 

(SLFNs). Pada penelitian ini hasil yang diperoleh dengan ekstraksi fitur 

GLCM dan klasifikasi menggunakan ELM menghasilkan akurasi 

terbaik sebesar 91.21%, sensitifitas 100%, dan spesifisitas 91% pada 

rotasi 135° menggunakan fungsi aktivasi sin dengan percobaan node 

hidden 15. 

 

Keywords:  

COVID-19, X-Ray 

image, CAD, GLCM, 

ELM 

 

 

Abstract. COVID-19 is a disease or virus that has recently spread 

worldwide. The disease has also taken many casualties because the 

virus is notoriously deadly. An examination can be carried out using a 

chest X-Ray because it costs cheaper compared to swab and PCR tests. 

The data used in this study was chest X-Ray image data. Chest X-Ray 

images can be identified using Computer-Aided Diagnosis by utilizing 

machine learning classification. The first step was the preprocessing 

stage and feature extraction using the Gray Level Co-Occurrence 

Matrix (GLCM). The result of the feature extraction was then used at 

the classification stage. The classification process used was Extreme 

Learning Machine (ELM). Extreme Learning Machine (ELM) is one of 

the artificial neural networks with advanced feedforward which has one 

hidden layer called Single Hidden Layer Feedforward Neural Networks 

(SLFNs).  The results obtained by GLCM feature extraction and 

classification using ELM achieved the best accuracy of 91.21%, the 

sensitivity of 100%, and the specificity of 91% at 135° rotation using 

linear activation function with 15 hidden nodes. 

  

 

How to cite: 

V. U. M. Maksum, D. C. R. Novitasari, and A. Hamid,“Image X-Ray Classification for COVID-

19 Detection Using GCLM-ELM”, J. Mat. Mantik, vol. 7, no. 1, pp. 74-85, May 2021. 

 

Jurnal Matematika MANTIK 

Vol. 7, No. 1, May 2021, pp. 74-85 

 

ISSN: 2527-3159 (print) 2527-3167 (online) 

mailto:diancrini@uinsby.ac.id
https://doi.org/10.15642/mantik.2021.7.1.74-85
http://u.lipi.go.id/1458103791




Vivin Umrotul M. Maksum, Dian C. Rini Novitasari, Abdulloh Hamid 

Image X-Ray Classification for COVID-19 Detection Using GCLM-ELM 

75 
 

1. Introduction  

The disease discussed by people around the world that was first discovered in Wuhan, 

Hubei Province, China which was reported to the World Health Organization (WHO) on 

December 31, 2019 is an outbreak of acute respiratory infection (pneumonia) caused by the 

new corona virus severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). On 

February 12, 2020, WHO officially declared the disease coronavirus disease 2019 

(COVID-19). COVID-19 is an acute infectious disease that attacks the respiratory system 

[1-2]. The most common symptoms of patients infected with COVID-19 are fever, myalgia 

or fatigue, dry cough, headache, hemoptysis, diarrhea, and the patients slowly experience 

severe shortness of breath [3-4]. COVID-19 has a very fast growth rate but can be 

suppressed to balance medical care capabilities. Thus the death rate can be lowered [5]. 

Several ways have been carried out to reduce the increased rate of people with COVID-

19 diseases rapidly rising. One way is by swab or polymerase chain reaction (PCR) tests 

[6]. However, both of the tests require considerable time and cost. Therefore, the latest way 

to detect COVID-19 disease is by using X-Ray. X-Rays are considered capable of 

describing the condition of patients infected with COVID-19 and can also be a clinical 

diagnostic tool [7].  

The first step in detecting COVID-19 based on X-Ray images is pre-processing image 

data [8]. In the pre-processing stage, feature extraction from X-Ray images is carried out 

using the Gray Level Co-occurrence Matrix (GLCM) method. GLCM is a second-order 

statistical method that calculates the relationship between two pixels in an image with 

degrees of grey. In previous research studies, GLCM has a good performance in extracting 

image features, and it was implemented to extract colposcopy image features before 

carrying out the classification process. The best accuracy results in the classification 

process are 95%, which shows that the features extracted using the GLCM algorithm can 

represent images well [9]. The classification process can be done by several methods, one 

of which is the Extreme Learning Machine (ELM) method. 

ELM is a feedforward neural network with a hidden layer with hidden node parameters 

randomly generated and the weights calculated analytically [10]. The classification using 

ELM has good performance with a relatively short computation time [11]. Based on 

previous research and the above explanation, this study classifies X-Ray images using the 

GLCM and ELM algorithms as feature extraction and classification to detect the presence 

of COVID-19. This research is expected to be implemented so that patients exposed to 

COVID-19 receive immediate and appropriate treatment. 
 

2. Preliminaries 

2.1 Corona Virus Disease 2019 (COVID-19) 

Corona Virus Disease 2019 (COVID-19) is an infectious disease caused by a new 
virus and is known as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) 

[12]. The disease attacks on the chest so it is almost similar to acute pneumonia [13]. The 

virus was first discovered in China on December 31, 2019 and on February 12, 2020, 

the World Health Organization (WHO) officially declared the disease COVID-19 [14]. 
COVID-19 is an animal-borne disease, but it can spread to humans. The most common 

symptoms of COVID-19 are fever, fatigue, and dry cough [15]. Some people may 

experience pain, nasal congestion, runny nose, headache, phlegm cough, sore throat, 

shortness of breath, and diarrhea [16]. These symptoms usually occur gradually, but some 

patients infected with COVID-19 do not show these symptoms and do not feel any 

problems with their bodies. People who are susceptible to COVID-19 are those who have 

a history of diseases related to the heart and blood vessels. COVID-19 cannot be seen only 

with the naked eye, but the disease can be seen using Chest X-Ray. Since COVID-19 



Jurnal Matematika MANTIK 

Vol. 7, No. 1, May 2021, pp. 74-85 
 

76 

 

attacks the epithelial cells lining the respiratory tract, it can be analyzed using Chest X-Ray 

[17]. The following are the results of the COVID-19 Chest X-Ray, as shown in Figure 1. 

 

Figure 1. COVID-19 Chest X-Ray  

 

2.2 Gray Level Co-Occurrence Matrix (GLCM) 

Gray Level Co-occurrence Matrix (GLCM) is a very popular feature-based feature 

extraction method [18-19]. GLCM determines the texture relationship between pixels that 

have the same degree of grayness so that statistical features or features are obtained in an 
image [20]. GLCM is stored in the form of an 𝑖 × 𝑗 × 𝑛 matrix, where n is a GLCM number 

of different rotation directions [21]. The rotation direction is formed by four shifting 

directions of 0°,45°,90° and 135°. There are several extraction features commonly 
used for GLCM extraction as shown from Equation(1) to Equation (4) [22] : 

a. Contrast 
Contrast is a variation of the local intensity value in the texture matrix. The lower 

texture contrast means the neighboring pixels on the matrix have similar local intensity 

values and vice versa. The following is an Equation to calculate the value of the 

contrast. 

𝐶𝑜𝑛𝑡𝑟𝑎𝑠𝑡 = ∑ ∑ 𝑃𝑎,𝑏 (𝑎 − 𝑏)
2

𝑏𝑎

 (1) 

Where 𝑃𝑎,𝑏  represents the pixel values in row a and column b in the co-currency matrix. 
b. Homogeneity  

Homogeneity is the degree of homogeneity of the repetition of textures. 

𝐻𝑜𝑚𝑜𝑔𝑒𝑛𝑒𝑖𝑡𝑦 = ∑ ∑(
𝑃𝑎,𝑏

1 + (𝑎 − 𝑏)2 
)

𝑏𝑎

 (2) 

c. Energy 
Energy is a measure of the uniformity of texture. The high value of energy represents 

a high degree of texture uniformity in an image. The following is an Equation to 

calculate the value of energy. 

𝐸𝑛𝑒𝑟𝑔𝑦 = ∑ ∑(𝑃𝑎,𝑏)
2

𝑏𝑎

 (3) 

d.  Correlation  
Correlation is a measure of the degree of linear relationship of the grayness of one pixel 

relative to another. 

𝐶𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛 = ∑ ∑  𝑃𝑎,𝑏  
(𝑎 − 𝜇𝑖 )(𝑏 − 𝜇𝑗 )

√𝜎𝑎
2𝜎𝑏

2𝑏𝑎

 
(4) 

where, 

𝜇𝑎 : ∑ ∑ 𝑎. 𝑃𝑎,𝑏𝑏𝑎  

𝜇𝑏 : ∑ ∑ 𝑏. 𝑃𝑎,𝑏𝑏𝑎  

𝜎𝑎  : √∑ ∑ (𝑎 − 𝜇𝑎)
2𝑃𝑎,𝑏𝑏𝑎  

𝜎𝑏  : √∑ ∑ (𝑏 − 𝜇𝑏)
2𝑃𝑎,𝑏𝑏𝑎  



Vivin Umrotul M. Maksum, Dian C. Rini Novitasari, Abdulloh Hamid 

Image X-Ray Classification for COVID-19 Detection Using GCLM-ELM 

77 
 

2.3 K-Fold Cross Validation 

K-fold cross validation or also called rotation estimation is a technique to 

minimize bias by random sampling of training and testing data [23]. One technique 

of k-fold cross validation is to estimate how accurate a model is when executed. In 

addition, the technique of k-fold cross validation breaks the data into K parts of the 

same size. Training and testing are conducted as many times as K times. The first 

experiment is a subset of D1 data into testing data and the other subset of data into 

training data, the second experiment is a subset of D2 data into testing data and a 

subset of  D1, D3 data up to a subset of  Dk data into training data [24].  The 

illustration of the k-fold cross validation is shown in Figure 2 

 
Figure 2. Data partition using K-Fold Cross Validation  

 

2.4   Extreme Learning Machine (ELM) 

Extreme Learning Machine (ELM) is a feedforward neural network with one hidden 

layer or commonly referred to as Single Hidden Layer Feedforward Neural Networks 
(SLFNs) [25]. 

In ELM algorithms, hidden node learning parameters including weight input 

and bias input can be randomly assigned. For network output weights can be 

determined analytically using simple general inverse operations [26]. The 

advantage of using ELM is efficient training without time-consuming learning for 

the process. In addition, on a universal approach, ELM capabilities have been 

successfully applied in many real-world applications, such as classification and 

regression issues [11]. 

ELM has a mathematical model that differs from backpropagation with a 

simpler and more effective model [27]. ELM neural network model with  neuron 

inputs, hidden layer neurons, and activation function 𝑔(𝑥). Purpose 𝑛 =
 [𝑥1, 𝑥2, 𝑥3, ⋯ , 𝑥𝑛] with 𝑥𝑛  is the input value on the network, 𝐻 is a matrix connecting the 
input layer and hidden layer, then the matrix has a size  𝑛 × 𝑚.  The determination of the 
values of the matrix elements is done randomly which can be formulated as follows [11]: 

∑ 𝛽𝑖
𝑚

𝑖=1
𝑔(𝑤𝑖𝑥𝑗 +  𝑏𝑗 ) =  𝑌 (5) 

where, 

𝑌 : predicted data, 
𝛽𝑖   : weight that connect hidden nodes and output nodes with 𝛽𝑖 =  (𝛽1, 𝛽2, ⋯ , 𝛽𝑚 ), 
𝑤𝑖  : the weight that connects the hidden node to the i and the input nodes with  
   𝑤𝑖 =  (𝑤1𝑖, 𝑤2𝑖, ⋯ , 𝑤𝑝𝑖) 

𝑥𝑗  : vector input data with 𝑥𝑗 =  𝑥𝑗(𝑡−1), 𝑥𝑗(𝑡−2), ⋯ , 𝑥𝑗(𝑡−𝑝) 

𝑗 : numbers 1, 2, ⋯ , 𝑚 



Jurnal Matematika MANTIK 

Vol. 7, No. 1, May 2021, pp. 74-85 
 

78 

 

𝑏𝑖  : bias weight to the i hidden node with 𝑏𝑖 =  (𝑏1, 𝑏2, ⋯ , 𝑏𝑚) 
Equation (5) can be converted into a matrix as Equation (6) 

 

𝐻𝛽 = 𝑌 (6) 
 

where, 

 

𝐻 =  [

𝑔(𝑤1𝑥1 +  𝑏1) ⋯ 𝑔(𝑤𝑚𝑥1 +  𝑏𝑚)
⋮ ⋱ ⋮

𝑔(𝑤𝑛𝑥𝑛,1 +  𝑏𝑛) ⋯ 𝑔(𝑤𝑚,𝑛𝑥𝑛 +  𝑏𝑚)
], dan 𝛽 =  [

𝛽1
⋮

𝛽𝑚

] 

 

𝛽 the matrix of the output weights and T is the matrix of the target. In ELM, weights and 
bias values are determined randomly. So that the output weight associated with the hidden 

layer can be determined by Equation [28]: 

 

𝛽 = 𝐻+ 𝑌 (7) 
 

Where 𝐻+ is a matrix that uses Moore Penrose Pseudo Inverse theory which has the 
best generalization results with fast computation time. In the ELM architecture described 

above, each node on the input is connected to a node in the hidden layer using the activation 

function 𝑔(𝑤𝑚𝑥𝑛 +  𝑏𝑚 ) thus resulting in an H matrix of many nodes × lots of data [29]. 
 

2.4 Confusion Matrix 

Confusion matrix with a size of 2 × 2 associated with the classifier shows the results of 
classification and actual data where TP (True Positive) is a class which is actually positive 

and predicted positive, so class labels are predicted correctly. And TN (True Negative) is a 

class of actually negative and predicted negative, so class labels are predicted correctly. 

Whereas, FP (False Positive) is a class of actually negative but predicted positive, so class 

labels are predicted wrong, and FN (False Negative) is a class of  actually positive but 

predicted negative, so class labels are predicted wrong [30]. This can be seen in Table 1 as 

follows: 
Table 1. Confusion Matrix 

 

Predicted Value 
Actual Value 

True False 

True True Positive (TP) False Positive (FP) 

False False Negative (FN) True Negative (TN) 

 

From Table 2.1,  the level of accuracy, sensitivity, and specificity of an algorithm 

model can be calculated using Equation 8, 9, and 10 [18] . 

Accuracy = 
𝑇𝑃+𝑇𝑁

𝑇𝑃+𝑇𝑁+𝐹𝑃+𝐹𝑁
 × 100 (8) 

Sensitivity =  
𝑇𝑃

𝑇𝑃 + 𝐹𝑁
 × 100 (9) 

Specificity =  
𝑇𝑁

𝑇𝑁 + 𝐹𝑃
 × 100 (10) 

 

 

3. Research Method 

3.1 Data 

This study used chest X-Ray image data consisting of 1031 data derived 

from [31]. Table 2 shows the data used in this study. 



Vivin Umrotul M. Maksum, Dian C. Rini Novitasari, Abdulloh Hamid 

Image X-Ray Classification for COVID-19 Detection Using GCLM-ELM 

79 
 

Table 2. Research Data Details 
 

 Data Type Data Format  Number of Data 

COVID-19 JPEG 119 

Non COVID-19 JPEG 912 

 

The data was taken on 03 April 2020 from [32-35]. The data increased every week so 

that when retrieving at different times the number of data increased. 

3.2 Research Type 

This type of research is applied research. In this research report, a system is created to 

check whether a person is suffering from COVID-19 by using ELM method. Figure 3 is 

the flowchart for the detection of COVID-19. 

 
 

Figure 3. COVID-19 Detection using GLCM-ELM Flowchart 

 

4. Results and Discussion 

Experiments of k = 5 were conducted in this study using hidden nodes of 5, 10, 15, 

20, 25, 30 and 35, then experiments were also conducted using several activation functions, 

that is sigmoid activation function, linear, sin, and radial basis. The results on the 

experiments using the hidden nodes of 5, 10, 15, 20, 25, 30 and 35, as well as those of the 

activation functions consisting of sigmoid, linear, sin and radial basis using four rotation 

directions of 0°,45°,90° and 135° are shown in the following table 3, table 4, table 5 and 

table 6.  

  



Jurnal Matematika MANTIK 

Vol. 7, No. 1, May 2021, pp. 74-85 
 

80 

 

Table 3. COVID-19 and Non COVID-19 Classification using GLCM-ELM with 0° Direction 
 

 Hidden Nodes 
Activation 

Function 
Accuracy Sensitivity Specificity  

5 

Sigmoid 88.35 50.00 89.10 

Linear 88.83 100.00 88.78 

Sin 88.83 66.67 89.16 

Radial Basis 87.86 33.33 88.67 

10 

Sigmoid 88.84 50.00 89.60 

Linear 87.86 100.00 88.72 

Sin 88.34 40.00 89.55 

Radial Basis 88.83 50.00 90.00 

15 

Sigmoid 91.26 85.71 91.46 

Linear 88.83 50.00 89.22 

Sin 89.86 71.42 90.5 

Radial Basis 89.37 75.00 89.66 

20 

Sigmoid 91.26 80.00 91.83 

Linear 88.83 100.00 88.78 

Sin 90.29 75.00 90.90 

Radial Basis 89.80 66.67 90.86 

25 

Sigmoid 88.84 54.55 90.77 

Linear 88.84 100.00 88.78 

Sin 87.86 42.86 89.45 

Radial Basis 87.86 44.44 89.85 

30 

Sigmoid 91.26 77.78 91.88 

Linear 88.84 66.67 89.16 

Sin 91.75 73.33 93.19 

Radial Basis 90.29 64.28 92.19 

35 

Sigmoid 91.75 81.82 92.31 

Linear 88.44 50.00 88.72 

Sin 89.81 58.82 92.59 

Radial Basis 88.35 50.00 91.15 

 
Table 4. COVID-19 and Non COVID-19 Classification Results using GLCM–ELM with 45° 

Direction 
 

 Hidden Nodes 
Activation 

Function 
Accuracy Sensitivity Specificity 

5 

Sigmoid 89.32 100.00 89.21 

Linear 89.32 75.00 89.6 

Sin 88.84 66.67 89.16 

Radial Basis 88.83 100.00 88.78 

10 

Sigmoid 87.37 34.00 89.00 

Linear 88.35 50.00 89.10 

Sin 87.86 43.00 89.45 

Radial Basis 87.37 34.00 89.00 

15 

Sigmoid 90.34 75.00 90.95 

Linear 88.40 50.00 88.78 

Sin 90.82 77.78 91.41 

Radial Basis 91.30 80.00 91.87 

20 

Sigmoid 92.72 83.33 93.30 

Linear 89.81 100.00 89.71 

Sin 92.23 76.92 93.23 

Radial Basis 91.75 80.00 92.35 

25 

Sigmoid 90.82 66.67 92.70 

Linear 89.37 100.00 89.27 

Sin 92.27 75.00 93.72 

Radial Basis 91.30 71.43 92.75 



Vivin Umrotul M. Maksum, Dian C. Rini Novitasari, Abdulloh Hamid 

Image X-Ray Classification for COVID-19 Detection Using GCLM-ELM 

81 
 

Table 4. COVID-19 and Non COVID-19 Classification Results using GLCM–ELM with 45° 

Direction (continued) 
 

 Hidden Nodes 
Activation 

Function 
Accuracy Sensitivity Specificity 

30 

Sigmoid 90.78 72.72 91.79 

Linear 89.32 100.00 89.22 

Sin 90.78 72.73 91.79 

Radial Basis 90.29 70.00 91.32 

35 

Sigmoid 90.78 72.73 91.80 

Linear 89.32 100.00 89.22 

Sin 90.78 72.73 91.79 

Radial Basis 90.29 70.00 91.33 

 

Table 5. COVID-19 and Non COVID-19 Classification Results using GLCM-ELM 90° Direction 
 

Hidden Nodes 
Activation 

Function 
Accuracy Sensitivity Specificity 

5 

Sigmoid 89.80 100.00 89.65 

Linear 89.32 75.00 89.60 

Sin 88.40 50.00 88.78 

Radial Basis 90.24 100.00 90.00 

10 

Sigmoid 89.32 66.67 90.00 

Linear 90.29 100.00 90.00 

Sin 90.29 75.00 90.90 

Radial Basis 88.40 50.00 88.78 

15 

Sigmoid 91.75 88.9 91.88 

Linear 89.32 100.00 89.21 

Sin 90.29 83.40 90.50 

Radial Basis 90.30 75.00 90.90 

20 

Sigmoid 89.81 80.00 90.04 

Linear 89.32 100.00 89.22 

Sin 89.81 80.00 90.04 

Radial Basis 89.81 80.00 90.04 

25 

Sigmoid 89.80 63.64 91.28 

Linear 88.84 100.00 88.78 

Sin 89.32 58.33 91.24 

Radial Basis 89.32 62.50 90.90 

30 

Sigmoid 88.84 54.55 90.78 

Linear 88.35 50.00 88.73 

Sin 88.84 55.56 90.36 

Radial Basis 88.84 54.55 90.77 

35 

Sigmoid 92.72 80.00 93.72 

Linear 88.84 66.68 89.16 

Sin 93.20 85.71 93.75 

Radial Basis 92.72 80.00 93.72 

 
  



Jurnal Matematika MANTIK 

Vol. 7, No. 1, May 2021, pp. 74-85 
 

82 

 

Table 6. COVID-19 and Non COVID-19 Classification Results using GLCM-ELM with 135° 

Direction 
 

Hidden Nodes 
Activation 

Function 
Accuracy Sensitivity Specificity 

5 

Sigmoid 88.89 100.00 88.84 

Linear 89.37 100.00 89.27 

Sin 88.83 100.00 88.89 

Radial Basis 89.32 75.00 89.60 

10 

Sigmoid 88.83 57.14 89.95 

Linear 88.35 50.00 89.10 

Sin 89.32 100.00 89.21 

Radial Basis 89.80 100.00 89.65 

15 

Sigmoid 90.73 100.00 90.54 

Linear 89.75 100.00 89.66 

Sin 91.21 100.00 91.00 

Radial Basis 89.80 66.67 90.86 

20 

Sigmoid 92.23 81.82 92.82 

Linear 89.32 100.00 89.27 

Sin 92.23 81.82 92.82 

Radial Basis 90.78 70.00 91.84 

25 

Sigmoid 89.86 58.82 92.63 

Linear 88.89 60.00 89.60 

Sin 91.30 71.43 92.75 

Radial Basis 90.82 64.71 93.16 

30 

Sigmoid 90.29 62.50 92.63 

Linear 87.38 33.33 89.00 

Sin 90.78 66.67 92.67 

Radial Basis 91.75 76.92 92.75 

35 

Sigmoid 90.78 77.78 91.37 

Linear 89.32 100.00 89.22 

Sin 90.78 77.78 91.37 

Radial Basis 90.78 77.78 91.37 

 

The best results are seen from the values of sensitivity, accuracy, and specificity. In 

this COVID-19 detection system, the best sensitivity value is to avoid misdiagnosis that 

patients with COVID-19 disease are detected incorrectly as Non COVID-19. Table 3 

represents the test results at 0°, using the linear activation function and 25 hidden nodes, 

and give the sensitivity, accuracy, and specificity values are 100%, 88.84%, and 88.78%, 

respectively. Table 4 represents the test results at 45°, using the linear activation function 

and 20 hidden nodes, and give sensitivity, accuracy, and specificity values are 100%, 

89.81%, and 89.71%, respectively. GLCM feature extraction with an angle of 90° using 

linear activation function and 10 hidden nodes obtained sensitivity, accuracy, and 

specificity values of 100%, 90.29%, and 90%. Meanwhile, in Table 6, the feature extraction 

at an angle of 135° with 15 hidden nodes and the activation function sin obtained the best 

sensitivity, accuracy, and specificity results of 100%, 91.21%, and 91%. 

The best results from each trial, based on the feature extraction direction, were 

obtained using the activation function parameters sin, 10 hidden nodes, and the feature 

extraction direction of 135°, namely with sensitivity, accuracy, and specificity values of 

100%, 91.21%, and 91%. Future research is expected to increase the sensitivity and 

accuracy with other feature extraction methods such as Gray Level Run-Length Matrix 

(GLRLM). In a study conducted by [34], GLRLM had better accuracy quality than the 

GLCM method. The future work is by applying ELM development methods, namely the 

Kernel Extreme Learning Machine (K-ELM) [35] and Multi-Layer Extreme Learning 

Machine (MLLEM) [36]. 

 



Vivin Umrotul M. Maksum, Dian C. Rini Novitasari, Abdulloh Hamid 

Image X-Ray Classification for COVID-19 Detection Using GCLM-ELM 

83 
 

5. Conclusions 

The feature extraction process using GLCM method can analyze the images of 

COVID-19 and Non COVID-19. Based on the extraction results of the GLCM feature with 

the four features used, it is shown that the extracted images produced good value. Good 

sensitivity was achieved from the feature extraction with 90°. 
Covid-19 classification using Extreme Learning Machine (ELM) method shows good 

results. The classification of COVID-19 in this study is divided into two classes, namely 

COVID-19 and Non COVID-19 which produced the best results at the rotation of 135° 

with the accuracy of 91.21%, the sensitivity of 100%, and the specificity of 91% on hidden 

nodes 15 using the activation function of sin. 

 
References 

[1] Z. Y. Zu et al., “Coronavirus disease 2019 (COVID-19): a perspective from China,” 

Radiology, vol. 296, no. 2, pp. E15–E25, 2020. 

[2] P. Boldog, T. Tekeli, Z. Vizi, A. Dénes, F. A. Bartha, and G. Röst, “Risk assessment 

of novel coronavirus COVID-19 outbreaks outside China,” J. Clin. Med., vol. 9, no. 

2, p. 571, 2020. 

[3] K.-C. Liu et al., “CT manifestations of coronavirus disease-2019: a retrospective 

analysis of 73 cases by disease severity,” Eur. J. Radiol., vol. 126, p. 108941, 2020. 

[4] C. Huang et al., “Clinical Features of Patients Infected with 2019 Novel Coronavirus 

in Wuhan, China,” pp. 497–506, 2020. 

[5] D. Wang et al., “Clinical characteristics of 138 hospitalized patients with 2019 novel 

coronavirus–infected pneumonia in Wuhan, China,” Jama, vol. 323, no. 11, pp. 

1061–1069, 2020. 

[6] J. E. Corral et al., “COVID-19 polymerase chain reaction testing before endoscopy: 

an economic analysis,” Gastrointest. Endosc., vol. 92, no. 3, pp. 524–534, 2020. 

[7] D. C. R. Novitasari et al., “Detection of covid-19 chest x-ray using support vector 

machine and convolutional neural network,” Commun. Math. Biol. Neurosci., vol. 

2020, p. Article-ID, 2020. 

[8] A. H. Asyhar, A. Z. Foeady, M. Thohir, A. Z. Arifin, D. Z. Haq, and D. C. R. 

Novitasari, “Implementation LSTM Algorithm for Cervical Cancer using 

Colposcopy Data,” 2020 Int. Conf. Artif. Intell. Inf. Commun., pp. 485–489, 2020. 

[9] M. Thohir, A. Z. Foeady, D. C. R. Novitasari, A. Z. Arifin, B. Y. Phiadelvira, and 

A. H. Asyhar, “Classification of Colposcopy Data Using GLCM-SVM on Cervical 

Cancer,” in 2020 International Conference on Artificial Intelligence in Information 

and Communication (ICAIIC), 2020, pp. 373–378. 

[10] Y. Wang, Z. Li, L. Feng, C. Zheng, and W. Zhang, “Automatic detection of epilepsy 

and seizure using multiclass sparse extreme learning machine classification,” 

Comput. Math. Methods Med., vol. 2017, 2017. 

[11] S. Ding, H. Zhao, Y. Zhang, X. Xu, and R. Nie, “Extreme learning machine: 

algorithm, theory and applications,” Artif. Intell. Rev., vol. 44, no. 1, pp. 103–115, 

2015. 

[12] B. S. Chhikara, B. Rathi, J. Singh, and F. N. U. Poonam, “Corona virus SARS-CoV-

2 disease COVID-19: Infection, prevention and clinical advances of the prospective 

chemical drug therapeutics,” Chem. Biol. Lett., vol. 7, no. 1, pp. 63–72, 2020. 



Jurnal Matematika MANTIK 

Vol. 7, No. 1, May 2021, pp. 74-85 
 

84 

 

[13] L. Lin, L. Lu, W. Cao, and T. Li, “Hypothesis for potential pathogenesis of SARS-

CoV-2 infection–a review of immune changes in patients with viral pneumonia,” 

Emerg. Microbes Infect., vol. 9, no. 1, pp. 727–732, 2020. 

[14] Y. Jee, “WHO international health regulations emergency committee for the 

COVID-19 outbreak,” Epidemiol. Health, vol. 42, 2020. 

[15] H. Al-Najjar and N. AL-Rousan, “Can Covid-19 Virus be created in the Laboratory?: 

A Theoretical Experimental Study,” A Theor. Exp. Study (April 22, 2020), 2020. 

[16] W. Ji, G. Bishnu, Z. Cai, and X. Shen, “Analysis clinical features of COVID-19 

infection in secondary epidemic area and report potential biomarkers in evaluation,” 

MedRxiv, 2020. 

[17] H. Dai et al., “High-resolution chest CT features and clinical characteristics of 

patients infected with COVID-19 in Jiangsu, China,” Int. J. Infect. Dis., vol. 95, pp. 

106–112, 2020. 

[18] A. Z. Foeady, D. C. R. Novitasari, A. H. Asyhar, and M. Firmansjah, “Automated 

Diagnosis System of Diabetic Retinopathy Using GLCM Method and SVM 

Classifier,” Proceeding Electr. Eng. Comput. Sci. Informatics, vol. 5, no. 1, pp. 154–

160, 2018. 

[19] D. C. R. Novitasari, W. T. Puspitasari, P. Wulandari, A. Z. Foeady, and M. F. Rozi, 

“Klasifikasi Alzheimer dan Non Alzheimer Menggunakan Fuzzy C-Mean, Gray 

Level Co-Occurence Matrix dan Support Vector Machine,” J. Mat. MANTIK, vol. 4, 

no. 2, pp. 83–89, 2018. 

[20] Ş. Öztürk and B. Akdemir, “Application of feature extraction and classification 

methods for histopathological image using GLCM, LBP, LBGLCM, GLRLM and 

SFTA,” Procedia Comput. Sci., vol. 132, pp. 40–46, 2018. 

[21] A. Harshavardhan, S. Babu, and T. Venugopal, “Analysis of feature extraction 

methods for the classification of brain tumor detection,” Int. J. Pure Appl. Math., 

vol. 117, no. 7, pp. 147–155, 2017. 

[22] M. Hall-Beyer, “Practical guidelines for choosing GLCM textures to use in 

landscape classification tasks over a range of moderate spatial scales,” Int. J. Remote 

Sens., vol. 38, no. 5, pp. 1312–1338, 2017. 

[23] B. Ghojogh and M. Crowley, “The theory behind overfitting, cross validation, 

regularization, bagging, and boosting: tutorial,” arXiv Prepr. arXiv1905.12787, 

2019. 

[24] T.-T. Wong and N.-Y. Yang, “Dependency analysis of accuracy estimates in k-fold 

cross validation,” IEEE Trans. Knowl. Data Eng., vol. 29, no. 11, pp. 2417–2427, 

2017. 

[25] M. A. A. Albadra and S. Tiuna, “Extreme learning machine: a review,” Int. J. Appl. 

Eng. Res., vol. 12, no. 14, pp. 4610–4623, 2017. 

[26] F. Kang, J. Liu, J. Li, and S. Li, “Concrete dam deformation prediction model for 

health monitoring based on extreme learning machine,” Struct. Control Heal. Monit., 

vol. 24, no. 10, p. e1997, 2017. 

[27] Y. Kuang, Q. Wu, J. Shao, J. Wu, and X. Wu, “Extreme learning machine 

classification method for lower limb movement recognition,” Cluster Comput., vol. 

20, no. 4, pp. 3051–3059, 2017. 

[28] S. Handika, I. Gririantari, and A. Dharma, “Perbandingan Metode Extreme Learning 

Machine dan Particle Swarm Optimization Extreme Learning Machine untuk 



Vivin Umrotul M. Maksum, Dian C. Rini Novitasari, Abdulloh Hamid 

Image X-Ray Classification for COVID-19 Detection Using GCLM-ELM 

85 
 

Peramalan Jumlah Penjualan Barang,” Maj. Ilm. Teknol. Elektro, vol. 15, no. 1, p. 

84, 2016. 

[29] J. Cao, K. Zhang, M. Luo, C. Yin, and X. Lai, “Extreme learning machine and 

adaptive sparse representation for image classification,” Neural networks, vol. 81, 

pp. 91–102, 2016. 

[30] X. Deng, Q. Liu, Y. Deng, and S. Mahadevan, “An improved method to construct 

basic probability assignment based on the confusion matrix for classification 

problem,” Inf. Sci. (Ny)., vol. 340–341, pp. 250–261, 2016. 

[31] D. S. Kermany et al., “Identifying medical diagnoses and treatable diseases by 

image-based deep learning,” Cell, vol. 172, no. 5, pp. 1122–1131, 2018. 

[32] P. Mooney, “Chest X-Ray Images (Pneumonia),” Kaggle, 2018. . 

[33] Github, “Covid Chest X-Ray Dataset,” 2019. . 

[34] D. C. R. Novitasari, A. Lubab, A. Sawiji, and A. H. Asyhar, “Application of feature 

extraction for breast cancer using one order statistic, GLCM, GLRLM, and GLDM,” 

Adv. Sci. Technol. Eng. Syst. J., vol. 4, no. 4, pp. 115–120, 2019. 

[35] D. C. R. Novitasari, A. H. Asyhar, M. Thohir, A. Z. Arifin, H. Mu’jizah, and A. Z. 

Foeady, “Cervical Cancer Identification Based Texture Analysis Using GLCM-

KELM on Colposcopy Data,” in 2020 International Conference on Artificial 

Intelligence in Information and Communication (ICAIIC), 2020, pp. 409–414. 

[36] S. Ding, N. Zhang, X. Xu, L. Guo, and J. Zhang, “Deep Extreme Learning Machine 

and Its Application in EEG Classification,” Math. Probl. Eng., vol. 2015, 2015.