https://doi.org/10.31449/inf.v47i8.4840                                                                                            Informatica 47 (2023) 77–88 77 
Provably Efficient Multi-Cancer Image Segmentation Based on 
Multi-Class Fuzzy Entropy 
Hend Muslim Jasim
1
, Mudhafar Jalil Jassim Ghrabat 
2,3
, Luqman Qader Abdulrahman
4
, Vincent Omollo Nyangaresi
5
, 
Junchao Ma
6, *
, Zaid Ameen Abduljabbar
1,7,8, *
, Iman Qays Abduljaleel
9 
1
Department of Computer Science, College of Education for Pure Sciences, University of Basrah, Basrah, 61004, Iraq 
2
Iraqi Commission for Computers and Informatics, the Informatics Institute for Postgraduate Studies, Baghdad, Iraq 
3
Computer Science Department, Al-Turath University College, Baghdad 10013, Iraq 
4
College of Health Science, Hawler Medical University, Erbil, Iraq 
5
Jaramogi Oginga Odinga University of Science & Technology, Bondo, 40601, Kenya 
6
College of Big Data and Internet, Shenzhen Technology University, Shenzhen, 518118, China 
7
Technical Computer Engineering Department, AL-Kunooze University College, Basrah 6100, Iraq 
8
Huazhong University of Science and Technology, Shenzhen Institute, Shenzhen, 518000, China 
9
Department of Computer Science, College of Computer Science and Information Technology, University of Basrah, 
Basrah 61004, Iraq 
E-mail: majunchao@sztu.edu.cn; zaid.ameen@uobasrah.edu.iq 
* Corresponding author’s  
 
Keywords: cancer segmentation, medical images, image segmentation, fuzzy entropy, cancer tumours 
Received: May 5, 2023 
 
One of the segmentation techniques with the greatest degree of success used in numerous recent 
applications is multi-level thresholding. The selection of appropriate threshold values presents difficulties 
for traditional methods, however, and, as a result, techniques have been developed to address these 
difficulties multidimensionally. Such approaches have been shown to be an efficient way of identifying the 
areas affected in multi-cancer cases in order to define the treatment area. Multi-cancer methods that 
facilitate a certain degree of competence are thus required. This study tested storing MRI brain scans in a 
multidimensional image database, which is a significant departure from past studies, as a way to improve 
the efficacy, efficiency, and sensitivity of cancer detection.  The evaluation findings offered success rates 
for cancer diagnoses of 99.08%, 99.87%, 94%; 97.08%, 98.3%, and 93.38% sensitivity; the success rates 
of in particular were 99.99%; 98.23%, 99.53%, and 99.98%. 
 
Povzetek: Raziskava se ukvarja z večdimenzionalnim shranjevanjem slik MRI možganov za izboljšanje 
učinkovitosti odkrivanja raka, dosega visoke stopnje uspešnosti in občutljivosti, kar predstavlja 
pomemben premik od tradicionalnih metod segmentacije. 
 
1 Introduction 
Within the segmentation of images, several 
characteristics, including colour, texture, and precision, 
are consistent features. Images are thus employed as a 
communication medium for many different reasons, 
including in the design of buildings and textiles, as well 
as in the field of medicine. The most common uses of 
direct imaging are seen in medical settings and 
photography, with the former focused on observation 
with the purpose of finding irregularities or verifying 
medical conditions [1][2]. The most common method 
takes pictures of the human body from various angles, 
and such pictures may serve as teaching tools for medical 
professionals in a variety of settings. Medical imaging is 
often used to aid in diagnosis; however, training and 
experience is required in how to interpret such images 
[3]. Several radioactive techniques may be seen in 
current medical imaging technologies, specifically in  
 
pictures of procedures is cysts, cancer, cancer, skin  
cancer, and breast cancer, as this offers options for 
making the group of useful images larger. The quantity  
of each kind of object in the collection is thus the target 
of any improvements [4][5]. 
Image segmentation is a crucial yet challenging step in 
the image processing pipeline, which has created 
significant issues among the segment of the scientific 
community concerned with visual material 
comprehension. Difficulties with segmentation hinder the 
widespread use of technologies such as 3D 
reconstruction, while effective segmentation may be used 
to divide a large picture into smaller, more manageable 
pieces by identifying comparable features across the 
image. The term can also refer to the process of 
separating an image's subject from its backdrop. 
Advancements in picture segmentation algorithms have 
made this process more efficient and accurate, while 

78   Informatica 47 (2023) 77–88                                                                                   H.M. Jasim et al. 
combining various recent advances in theory and 
technology is facilitating the development of a universal 
segmentation method applicable to all picture types [1]. 
Research published in Multi-Cancer recently described 
an efficient approach to monitoring treatment sites, based 
on a multi-touch technique for measuring specific values 
[6]. The efficacy, efficiency, and sensitivity of brain 
regions and their respective sizes as measured against 
prior research and outcomes show that distinct 
differences exist between various cancers in these 
regards [7]. 
Magnetic Resonance Imaging (MRI) scans, however, 
suffer from noise, weak borders, and artefacts due to the 
complexity of human brain tissues and the Nuclear 
Magnetic Resonance (NMR) inherent in this imaging 
technology. As a result, there is a substantial benefit to 
working to minimise the clinical risks associated with 
utilising brain imaging to predict disease. The novel 
features tested in this work include an improved model 
for brain image processing and brain disease diagnostic 
prediction based on combining Hybrid Pyramid U-Net 
Model for Brain Tumour Segmentation (HPU-Net) and 
improved fuzzy clustering, which offers improvements 
over fuzzy clustering in terms of noise, weak boundaries, 
and artefacts in brain images [3].  
The main contribution of this paper is thus a model 
addressing multi-cancer cases in an efficient approach 
that combines images from patients and pre-processed 
segments to classify the intensity of cancer spread. Each 
stage of the multi-cancer in this efficient approach is thus 
subject to optimisation, which enables high-accuracy 
results at all respective stages. A simulation was 
conducted to test the efficacy of the model against state-
of-art models, and the proposed multi-cancer model thus 
demonstrated higher accuracy than these other methods. 
 
2 Related works 
     Many scholars have examined the ways in which 
medical pictures might be clarified and shared, offering 
guidelines for doing so to various professionals. As a 
consequence, significant criteria and other measurements 
have been developed using a set of instruments, which 
satisfy pooled values value was satisfied. Wu et al. [8] 
provided an effective mode of motion for separating 
sediment in accordance with the CT measurements 
performed in accordance with skin malnutrition (FCM) 
using low-key visuals, while the identification of 
diseases by Nammalwar [9] offers another innovative 
combination of image types. 
Kengal and Jha [10] provided a comprehensive set of 
procedures for segmenting the brain and detecting 
diseased and healthy regions using skin secretions, while 
Minajags and Gudar argued that the use of DWT and 
FCM is the best way to approach imaging, while 
Vadgore used MRI and Takri used computer science 
techniques to create a technique for exchanging brain' 
medical profiles [11].  
A further novel method of disentangling the various 
overlapping realms of vision and sound that have 
plagued scholars since the days of Mirza emerged in [12] 
to promote the efficacy and accuracy of entropy in MRI 
image separation, which is questionable in the context of 
allergies. 
For the purposes of implementing and evaluating medical 
devices, Benjara presented a CDB-based distribution 
standard (DFCM) algorithm [13]. In terms of issues such 
as kidney tumours, and breast cancers, decay in medical 
images often clarified in film [14]. New research and 
imaging methods, however, rely on a means that may not 
offer reliable details on physical appearance [14,15]. 
Objects may sometimes not be attached to the relevant 
connections due to the presence of gaps and parasitic 
characteristics. Noise in medical equipment may be 
mitigated in a number of ways, however, making 
cleansing on a moderate scale an essential procedure 
[16]. 
Based on the evaluation of current methodologies, 
however, there are a number of significant problems with 
medical devices, including high computational 
complexity, faulty medical devices, long calculation 
times, high costs, semantic gaps, and a lack of 
dependability. 
Earlier studies examined items that could not be attached 
to the relevant connection using various algorithms; 
however, the resulting standard medical devices 
approach has certain drawbacks, including inaccurate 
feature extraction, manual picture annotation issues, and 
reduced accuracy. To address these problems and to 
improve accuracy and dependability, noise must be 
removed using a pre-processing approach with a median 
filter. The resulting medical approach is effective due to 
the fact that it uses a trained classifier to identify certain 
traits to collect the necessary data. 
The research gaps and shortcomings of earlier studies 
highlight a lack of research on the accurate identification 
of breast cancer utilising automated methods. Virtually 
all accessible datasets are also unbalanced, with the 
number of instances in one class far outnumbering those 
in all other classes. 
These research gaps are thus addressed in this work: 
As histopathologists are skilled in labelling 
histopathology images and lesion reports, deep learning 
on documents created by these expert is used to 
overcome the overfitting caused by over-parameterised 
networks and insufficient generalisation ability in other 
models. 
 
Table1: Classification related work. 
Models Accuracy 
(%) 
sensitivity 
(%) 
Specificity 
(%) 
 Li et al.[17] 86.4 87.4 ---- 
Kuruvilla, 
Gunavathi[1
8] 
93.3 91.3 100 
Choi, 
Choi[19] 
97.00 95.2 96.3 
CNN[20] 84.2 84.00 84.3 
Regularized 
Extreme 
94.23 91.00 93 

Provably Efficient Multi-Cancer Image Segmentation Based…                                                Informatica 47 (2023) 77–88 79 
Learning 
Machine 
(RELM) 
[21] 
Deep 
Convolution
al Neural 
Network 
(DCNN) 
[22] 
95 96.00 95.9 
 
In Table 1 shown a comparison of other relevant literary 
works. Different methods have been used in the overview 
of the results, but the traditional method has drawbacks 
such inaccurate feature extraction, human image 
annotation, and poorer accuracy. To solve these 
problems, noise is removed for improved accuracy and 
reliability using a pre-processing procedure with the 
types of cancer. The effective the types of cancer of 
images is a result of the use of a trained classifier for the 
classification of specific features to obtain the pertinent 
information. 
3 Methodology 
    This section discusses the research methods in order to 
identify a wide range of cancers addressed by different 
types of medicine.  
 
A. Level set method utilised to select multi-cancer 
region on medical images 
     In 1987, Oheser and Sedgin demonstrated how to 
drastically alter the user experience by improving 
images. To achieve this assumption gm = (x, y) is used to 
compare the majority of medical phenomena, which 
describes the situation(s) at a given moment. An 
illustration of the concept is given by 
(𝑀 (𝐷 ),𝐷 )=0                                                                         (1) 
 Boundary distribution is another example. The height of 
the ground is equal on the edge near m to (X i, Y i). 
 Thus,  
((Xi,Yi,t=0)=±d                                                                (2) 
The X picture depicts the time and space between the 
coordinates x, y, and 0, showing probability distribution 
functions (x, y, t = 0) and other information (PDEs) 
about the strength of these attributes as approximations 
in real time (t). The primary function is located at a 
distance of p from the maximum height of the active 
field (x, y), when (t, x, y) equals zero. The primary 
objective is to produce an adequate image the first time, 
with no user input if the observed body is healthy and the 
original function and m (t) are both still in place. As well 
as playing a role in anti-cancer efforts, this is useful in 
other areas. However, cancer and other issues in the skin 
take greater effort to observe. More importantly, there 
may be no picture area, with a null result ensuing. 
Addressing this in a systematic way is as easy as 
counting to one, however, as the starting function, e= 0, 
may be replaced by additional time, and may use the 
sentence(s) and legal statuses defined in [23] and [24]. 
Thus, 
 
∂∅(m(e),e)
∂e
=0 
∂∅
∂m(e)
∂m(e)
∂e
+
∂∅
𝑒 𝑒 e
=0     
∂∅
∂m(e)
m
t
+∅
𝑒 =0 
 (3) 
From the velocity equation (0)/ee = m (e), if Y is a 
property of some surface, then m (e) = Y (m (e)) n, n = (| 
|). The starting sign of the limit function determines the 
magnitude of the progressive force Y necessary to create 
the evolutionary system at the level closest to the ideal 
solution, as shown in [25]. 
∅
𝑒 +β∅m
t
  = 0 
∅
𝑒 +β∅Yn=0 
∅
𝑒 +Yβ∅
∇∅
|∇∅|
=0 
∅
e
+Y|β∅|=0 
 (4) 
Further, e = 0 is the process time, and the suffix can thus 
determine the movement of x (x, z, e) at any time without 
changing ∅ (x, z, e = 0) after the circulation of CF. The 
curvature of the surface as a result of movement and joint 
training to examine the level of compliance of any part of 
the cancer can thus be assessed [26]. 
U=β
β∅
|β∅|
  =
∅
xx
∅
z
2
−2∅
xz
∅
x
∅
z
+∅
zz
∅
x
2
(∅
x
2
+∅
z
)
1
2
⁄
 
 (5) 
 
B. Fuzzy entropy-based multi-cancer thresholding 
Pixels in both colour and grayscale may be recorded 
using this approach, which leverages the many methods 
available for image search applications [27], such as 
image visual processing, medical archiving, document 
image analysis, and map processing. The simplest 
approach to quickly applying a medical picture is to save 
it, and this also allows the sharing of images with other 
professionals. Thus, D = (J, I), where J = (0.1... M - 1), I 
= (0.1 ... N - 1). The number of grey levels used in 
medical imaging is expressed by the equation G = (0.1... 
L - 1), which is in turn determined by the range and 
height peak of the image, denoted via the two numbers m 
and n. The image value of a pixel (x, z) is such that DU = 
(x, z); (X, z) = k, in position (x, z) = D U = J (x, z), as 
these factors are intertwined in the process of obtaining 
pictures in the range (0,1, ..., L-1). The use of Ti and Tj 
supports medical imaging, except where D is the original 
medical picture, which has been split into three zones: m 
F m, b F b, and d F d. Gray values greater than 2 T = W 3 
F_, F b, and the F d - distribution (1) thus describe the 
probability distribution n [28]: the area as a whole 
contains F pixels with grey values less than Ti, F M thus 
covers the average grey value of pixels between Ti and 
Tj , and the gap between pixel F b and the grey value is 
greater than Ti. 
Thus, 

80   Informatica 47 (2023) 77–88                                                                                   H.M. Jasim et al. 
pd = P(F
d
) 
pm = P(F
m
) 
pb = P(F
b
) 
 (6) 
For F_, F_V, and F_b to α_v, are more important in 
comparison with α _b, α _d and the standard parameters 
of this procedure, while ai, bi, ci, aj, bj, and cj are also all 
very important. Within the framework of partnership, the 
functions (α) of the thresholds Ti and Tj are also variable 
by pixel number such that k = 0.1... 255. 
Where p (m) K p (b | K) and p (d.K), all pixels as noted 
by FM, F_b and F_d, are one of shaded, light and dark, 
and the pixel reports given by D _D_E and p (me) + p 
(bk) + p (d | e) = 1 (e = 0.1.2 ... 255), offer results as 
shown in [29]: 
  P
em
=P(D
d
)= P
e
∗ P
m|e
 
  P
eb
=P(D
m
)= P
ek
∗ P
b|e
 
  P
ed
=P(D
b
)= P
e
∗ P
d|e
 
 (8) 
The importance of large shaded squares (F_M), as well 
as those later areas that are clear (F_b) and dark (F_d), 
may be seen by examining equations (m_E), p_ (b | e), 
and p_ (d | e) [30]. A full comparison is offered in 
Equation (9) 
P
m
=∑P
𝐸 255
e=0
∗ P
m|E 
= ∑P
𝐸 255
e=0
∗ μ
m
(E) 
P
b
=∑P
𝑒 255
e=0
∗ P
b|e 
= ∑P
𝑒 255
e=0
∗ μ
b
(E) 
P
d
=∑P
𝑒 255
e=0
∗ P
d|e 
= ∑P
𝑒 255
e=0
∗ μ
d
(e) 
 (9) 
Z, K1, A1, B2, and C2 are equivalent to U, S and Z, D = 
(K, Ai, Bi, Ci, Aj, Bj, C2), and S (K, Ai) is substance (μ). 
The health group Z (e, Ai, Bj, Cj)), with μ leaders (e) rea 
(_d), can perform three health care tasks: 
μ
m
(𝑒 )=
{
 
 
 
 
 
 
 
 
 
 
 
 
         0,                                 e≤a1
(e−a1)
2
(c1−a1)∗(b1−a1)
,             ai<e≤bj
1−
(k−c1)
2
(c1−a1)∗(c1−b1)
,     bj<e≤aj
                 1,                                  ci<e≤aj
1−
(k−a2)
2
(c2−a2)∗(b2−a2)
,    ai<e≤bj
(k−c2)
2
(c2−a2)∗c2−b2)
,              bj<e≤cj
       0,                                  e>cj
 
 (10) 
       μ
b
(k)=
{
 
 
 
 
 
 
0,                                        e≤aj
(e−aj)
2
(cj−aj)∗(bj−aj)
,         aj<e≤bj
1−
(e−cj)
2
(cj−aj)∗(cj−bj)
,    bj<e≤cj
1,                                          e>cj
 
 (11) 
μ
d
(e)=
{
 
 
 
 
 
 
1,                                 e≤ai
1−
(e−ai)
2
(ci−ai)∗(ci−bi)
,   ai<e≤bi
(e−ci)
2
(ci−ai)∗(ci−bi)
,       bi<e≤ci
               0,                     e>ai
 
 (12) 
Using six boundary lines (0 <Ai <bi <Ci <Aj <Bj <cj 
lines255) for the pixels. Entropy is described in every 
section by means of the comparison between light and 
darkness [31]. 
H
m
=−∑
P
𝑒 ∗μ
m
(e)
P
m
∗In(
P
𝑒 ∗μ
m
(e)
P
m
)
255
e=0
  
H
b
=−∑
P
𝑒 ∗μ
b
(e)
P
b
∗In(
P
𝑒 ∗μ
b
(e)
P
b
)
255
e=0
 
H
d
=−∑
P
𝑒 ∗μ
d
(E)
P
d
∗In(
P
e
∗μ
d
(E)
P
d
)
255
e=0
 
 
 (13) 
All expressions can thus be presented as a single digit 
and row. 
 
 H((C1,A2),(B2,C2),(A1,b1)==H
m
H
b
+              (14) 
 
The intention of extraction of the above equations is to 
counteract the fact that the large entropy value is not 
suitable for all groups. The following changes were thus 
introduced per [32]: 
 
(d(Ti)=μ_m(Ti)=0.5 
μ_m(Tj)=μ_b(Tj)=0.5                                      (15) 
 
From Equations (10-12), the gur_m shaded section can 
be used as the foundation of the Ti and Tj level (K), thus 
allowing ih gihanna (bb (K)) and the dark segment ((_d 
(K)) [32] to be calculated. Ti and Tj are calculated as 
follows: 
Ti=
{
 
 
 
 
a1+
√
(ci−ai)∗(bi−ai)
2
,   
(ai+ci)
2
≤bi≤ci
ci−
√
(ci−ai)∗(ci−bi)
2
,   ai≤bi≤
(ai+ci)
2
 
Tj=
{
 
 
 
 
aj+
√
(cj−aj)∗(bj−aj)
2
,
(aj+cj)
2
≤bj≤cj
cj−
√
(cj−aj)∗(cj−bj)
2
,aj≤bj≤
(aj+cj)
2
 
 
 (16) 
      D
d
={(x,y):I(x,y) ≤Ti,   (x,y) ∈D
k
} 
      D
m
={(x,y):Ti<𝐼 (x,y) ≤Tj,   (x,y) ∈D
k
} 
      D
b
={(x,y):I(x,y)>𝑇𝑗 ,   (x,y) ∈D
k
} 
       
         
(
7
) 

Provably Efficient Multi-Cancer Image Segmentation Based…                                                Informatica 47 (2023) 77–88 81 
4   Proposed method 
 Researchers have created and applied a wide variety of 
cancer detection methods across the field of segmental 
medical imaging. The suggested procedure may be used 
in the assessment of cases featuring several cancers, even 
where these are represented in separate medical videos. 
The proposed method of segmenting medical images and 
properly identifying cancer has the potential to provide 
urgently required details regarding the structure of 
tumours [33]. This research also used a variety of 
approaches simultaneously to develop a practical 
technique for dissemination. Figure 1 provides a 
graphical representation of this procedure. 
 
Algorithm 1 is based on the FEL input value. 
 Input: medical images 
Output: identification of cancerous areas 
Step 1: Input doctor’s opinion. 
Step 2: Examine the relevant medical results. 
Step 3: Start searching for medical images (4). If the 
energy value is positive (F = 1), the meter should be 
included in the picture. If the energy is negative (F = -1), 
the energy meter must be outside of the picture. 
Step 4: Evaluate the numerically equivalent N-
function using equations (2) and (3). 
Step 5: If the function has a zero value for _t cp 
border (4), select a hundred or so cancerous tissues and 
apply the following equations: 
k = (∇∅) / | | = (_Xx ∅_y ^ 2-2Ø_xy _x y_a + ∅_yy 
∅_x ^ 2) / (∅_x ^ 2 + ∅_y) ^ (1/2) 
Step 6: Identify three small entropy input modes, 
such as U, S, and Z, such that the values of all member 
functions, such as μ_m (s); cleaning, μ 〗 _b (k); and 
darkness, μ_d k can be represented using equations (10 to 
12). 
Step 7: Calculate the double values to find the lower 
and upper limits (15). 
Step 8: Collectivise the resulting images of the 
relevant cancer and its parts. 
Step 9: End. 
 
Graph 1 shows the estimated cancer field using such 
medical images segmentation. The approach is divided 
into three stages: in the first, the algorithm is used to find 
and select parts of images of cancer. The second step 
analyses the entropy compression in a given cancer and 
selects a control sample. 
Algorithm 1 uses double caps to collect the details of the 
last stages of cancer. Details of the proposed procedure 
are as follows: 
 
Adjustable scale algorithms provide a looped search for 
appropriate medical images when searching for particular 
cancerous tissues. 
Each grayscale area is assessed according to both the 
background and foreground. 
For each grayscale area of the members of a medical 
resource, for each input value, the entropy group 
algorithm is recalculated and the threshold value 
selected. 
The cancer component is thus defined as the first 
grayscale area, i.e., that with the lowest entropy value, 
thus dividing this from healthy tissue and expanding the 
field of cancer research.  
5 Experimental results and 
discussion 
According to previous researchers, "the proposed 
algorithm defines three types of images out of the 
described 84 films” [34]. These images were based on 28 
photographs [35] of breast cancer in ultrasound images 
of three types, with 32 separated after the administration 
of an anticancer drug of which there were three types: 
cancer or small cell imaging can be used for the first 
stage of Kuppatarkun medical imaging, the three 
protocol aims are to improve efficiency, effectiveness, 
and accuracy in research, a process elaborated on in the 
remainder of this section. The computer used featured 
8GB of RAM running under Windows 7 64-bit OS. 
 
 
Figure 1: Proposed approach for extracting the part with cancer through medical image segmentation. 

82   Informatica 47 (2023) 77–88                                                                                   H.M. Jasim et al. 
5.1 Experimental results 
In this research, the analysis can broke down using the 
first step employed in the procedure to calculate the 
distance between the threshold and the actual 
measurement. Ultrasound was used for this analysis, 
which opened the door to three different applications of 
clinical physics (breast cancer, dermoscopy, and skin 
image detection through magnetic resonance). Different 
compression rates were also tested using the suggested 
technique for compressing the test results. Figures 2 to 4 
illustrate the process. Multiple viewing regions were 
discovered based on the research and digitisation data 
and the latter’s influence on the values of the parameter 
mask (M) (GD). In the medical field, these procedures 
represent cancer GT chronology, in this case offering a 
pixel-perfect image [36]. As a means of regulating levels, 
the sensitivity of this procedure allows it to impact 
cancer visualisation be increasing clarity (Figures. 2 to 4 
c). Improved experimental graphics with the use of 
optimised visuals for displaying activity allows results 
with a much higher degree of accuracy. Figure 2 thus 
depicts the sensitivity, precision, accuracy, and accuracy 
features of each medicine tested, while Figure 3 provides 
a succinct overview of the data. 
 
Figure 2: Multi-cancer segmentation for ultrasound 
breast imaging as a result of FEL thresholding the initial 
work is done by thresholding (40:160,190:240) for 
extraction of the cancer region, as shown in (a) the 
original ultrasound picture with several cancers, (b) the 
wider search (c) the multi-cancer segment after 500 
iterations, and (d) the initialisation of the process. 
 
 
Figure 3: The multi-cancer segmentation for MRI 
baseline was thresholded using FELs. The original MRI 
picture with several cancers (a) was used in a search to 
detect the cancerous region (b), while segmentation of 
multiple cancers (c) after 500 iterations, and initialisation 
were shown to be triggered by thresholding (240:390, 
200:290) in terms of cancer region extraction (d).  
 
Figure 4: Multi-cancer segmentation for skin 
dermoscopy colour pictures based on FEL thresholding 
The original skin picture with several cancers is shown in 
(a), the search to detect cancer is seen in the (b) region, 
cancer found and segmented can be seen in (c) after 700 
iterations and initialisation by thresholding (60:243, 
80:385) for various cancer region extractions, as seen in 
in (d). 
 

Provably Efficient Multi-Cancer Image Segmentation Based…                                                Informatica 47 (2023) 77–88 83 
 
Figure 5: Segmentation of breast cancer with ultrasound 
illustrated using iterative measurements. 
 
Figure 6: Iterative segmentation of brain cancer results. 
 
Figure 7: Iteration of skin cancer segmentation using 
dermoscopy colour images. 
Figure 7 displays the pre-processing and segmentation 
results of three example input images. As shown Figure 
4, FEL thresholding results in the multi-cancer 
segmentation of skin dermoscopy colour images, with 
the original skin images showing several tumours. As 
shown Figure 5, the dataset was split into training and 
testing groups with a 50:50 ratio. Following that, the 
training images for each class were split into benign (85 
images), malignant (46 images), and normal (46 images) 
groups (185 images in total).  
A data augmentation step was then used due to the fact 
that the dataset was insufficient to train a deep learning 
model. To expand the variety of the original dataset, 
three operations were implemented and applied to the 
original ultrasound images: horizontal flip, vertical flip, 
and rotation by 90°. These processes were repeated 
several times until there were 3,200 images in each class. 
The dataset thus contained10,385 images after the 
augmentation procedure. Examples of classified images 
from the original dataset using record-wise 10-fold cross-
validation are shown in. Figure 6, which displays the 
confusion matrices and shows what the intended target 
was, thus usefully determining classified images. 
 
Figure 8: Results of segmentation of performance for the 
medical imaging algorithms using FELs. 
Table 2: The results of multi-cancer segmentation using 
Focused Echocardiography in Life Support. 
Method for 
Cancer 
Segmentation 
Performance measures 
Jaccard - Dice 
 
Colour k means 0.524357 0.670071 
fuzzy c-means 0.572415 0.690864 
Total variation 
fuzzy c-means 
0.580086 0.694207 
Texture Features 0.235247 0.354119 
A fourth order 
partial differential 
equation fuzzy c-
means 
 
0.596755 
 
0.736681 
Focused 
Echocardiography 
in Life Support 
for Ultrasound 
0.9865 0.966 
Focused 
Echocardiography 
in Life Support 
for MRI 
 
0.9891 
0.9509 
Focused 
Echocardiography 
in Life Support 
for Dermoscopy 
0.9757 0.9482 

84   Informatica 47 (2023) 77–88                                                                                   H.M. Jasim et al. 
Table 3: Comparison of the suggested approach with 
existing conventional methods. 
Image Ultrasound Brain Dermoscopy 
Jaccard 
0.9765 0.9891 0.9657 
Dice 
0.956 0.9409 0.9282 
Accuracy 
0.9979 0.9987 0.9794 
Sensitivity 
0.9708 0.983 0.9338 
Specificity 
0.9992 0.9997 0.9999 
Precision 
0.9823 0.9953 0.9998 
 
5.2 Comparison of segmentation results 
Three medical image types were utilised experimentally 
in the suggested FEL approach to see boundaries in 
various settings. The findings were compiled, and FEL 
was then compared to other drug production techniques. 
The second component of the picture makes use of the 
Dachshund Jacuzzi technique, which was first presented 
in 1912; this is, however, based on the Dice-method, a 
highly repetitious process; thus, for the purposes of 
describing the real branch of work, a report is commonly 
used to assess and utilise both group outcomes and 
related records [37], as shown Figure 8 and Table 1. 
These range from zero contracts through single results to 
full contracts. Combining distance measurements with 
GT and segmentation (cs) yields both uppercase and 
Jaccard equations [38]. There are four criteria that go 
towards determining the CBD index. Current and real 
conditions detected include viruses, as shown by the 
positive outcomes. However, such incorrect information 
may have dire repercussions for cancer patients: such 
scores indicate the presence of aberrant risks and suggest 
that the findings are flawed. However, the 
counterexample is false results [39][40], as seen in the 
original data. 
For the Jaccard coefficient, the GT represents ground 
truth, while CS, as indicated in Table 2, illustrates cancer 
segmentation; the Jaccard coefficient is thus defined as 
follows [41]: 
 
Jaccard =
|𝐺𝑇 ∩𝐶𝑆 |
|𝐺𝑇 ∪𝐶𝑆 |
 
 (17) 
Dice similarity coefficient 
 
Dice =
2∗|GT∗CS|
|GT|+|CS|
 
 (18) 
 
If CS is true colour of the kernel and the GT, the search 
results show that the Jacquard coefficient and the dice 
differ from the 0 and 1 values. The value 0 corresponds 
to the contract, but it is worth 1 specification (complete 
contract) [42]. For the relevant section, alongside the 
existing hypothesis, the proposed FEL method should 
also determine the sensitivity, precision, accuracy and 
precision of the SCS analysis. Positive results indicate 
that the WSG scores are positive [43, 44, 45, 46, 47].  
 
Sensitivity= 
𝑇𝑃
𝑇𝑃 +𝐹𝑁
 
 (19) 
Precision= 
𝑇𝑃
𝑇𝑃 +𝐹𝑃
 
 (20) 
Specificity= 
𝑇𝑁
𝑇𝑁 +𝐹𝑃
 
 (21) 
Accuracy= 
𝑇𝑃 +𝑇𝑁
𝑇𝑃 +𝑇𝑁 +𝐹𝑃 +𝐹𝑁
 
 (22) 
If the medical images are correctly separated, the 
results are described as "correct". 
 
Correct Rate = ( Xc/ X ) ∗ 100%  (23) 
Rrror  Rate = ( XR / X ) ∗ 100%  (24) 
 
The proposed approach was used on 28 ultrasound 
results from 27.4 cases of breast cancer, giving an 
average rate of 96.4% with only 3.6% error. Accuracy 
was at 96.9%, with an estimated 95.8% accuracy in sets 
featuring fewer than 23 people: there were 32 cases of 
brain cancer from the actual results of the RM, with 31 
cases detected, giving an error of just 3.1%, as shown 
Figure 9. 
 
Figure 9: Comparison of results for the proposed method 
with other standard segmentation cancer methods by 
Dice-coefficient, and Jaccard-coefficient. 
Table 4: A comparison between the current models and 
our suggested 
Models Accuracy (%) 
Deep Convolutional Neural 
Network (DCNN)[21] 
95% 
Deep Convolutional Neural 
Network (DCNN)[22] 
96% 
Deep Convolutional Neural 
Network (DCNN)[48] 
96.6 
Deep Convolutional Neural 
Network (DCNN)[49] 
97.5% 
Deep Convolutional Neural 
Network (DCNN)[50] 
95% 
Our Proposal Method 0.9979 
 

Provably Efficient Multi-Cancer Image Segmentation Based…                                                Informatica 47 (2023) 77–88 85 
Table 4, The our proposed method classified and the 
existing approaches ((DCNN)[21],(DCNN)[22],   
(DCNN)[48],  (DCNN)[49] and (DCNN)[50]) in    
compared. Estimated values for accuracy show that the 
suggested methodology yields better results than do the 
existing methods. 
 
The performance of the suggested methodology is good, 
with an accuracy value of around 0.99.79%, this statistics 
demonstrate an increase in system efficiency when 
compared to those of the traditional methodologies. 
 
6 Conclusion  
     When creating diagnostic medical imaging systems, 
an approximation approach for defining FEL criteria is 
helpful, and breast cancer, brain magnetic resonance 
imaging (MRI), and visual skin therapy are the three 
categories of pictures used in this work. In terms of the 
differences between coefficients (average 93.82%) the 
Cocharge Supply had a coefficient of 97.57%, while the 
Jacquard and Dice achieved 98.65% and Coquette 
achieved 98.91%. The recommended procedure for 
evaluating images pixel-by-pixel thus demonstrates 
reliability in terms of identifying the cancer data in each 
image utilised in computational suggestion. Due to the 
area's sufficiency, computational expenses may be 
disregarded, while with regard to cancer diagnostics, the 
success rates were 99.08%, 99.87%, 97.94%; 97.08%, 
98.3%, 93.38%, 99.92%, and 99.97%; the success rates 
of the in particular were 99.99%, 98.23%, 99.53%, and 
99.98%, virtually with the ground state. Using this 
system, various forms of cancer may be categorised 
based on how they can be diagnosed medically. To better 
identify the metastasis of cancers other than those of the 
brain, however, the current algorithm may require further 
refinement in the near future. 
Acknowledgement  
This work is supported by Natural Science Foundation of 
Top Talent of SZTU (Grant No. 20211061010016). 
References 
[1] Xiangbin Liu,et,al (2021).” A Review of Deep-
Learning-Based Medical Image Segmentation 
Methods”. Sustainability, 13, 1224. doi: 
https://doi.org/10.3390/su13031224. 
[2] Corbat, Lisa, Mohammad Nauval, Julien Henriet, 
and Jean-Christophe Lapayre (2020). "A fusion 
method based on deep learning and case-based 
reasoning which improves the resulting medical 
image segmentations". Expert Systems with 
Applications 147,113200.doi: 
https://doi.org/10.1016/j.eswa.2020.113200. 
[3] dong Hu1, Yi Zhong, Shuxuan Xie, Haibin Lv and 
Zhihan Lv (2021).” fuzzy system based medical 
image processing for brain disease prediction”. 
Front. Neuroscience, vol. 15,714318. 
doi: https://doi.org/10.3389/fnins.2021.714318. 
[4] Zhang, Ziyang, Chenrui Duan, Tao Lin, Shoujun 
Zhou, Yuanquan Wang, and Xuedong Gao. (2020). 
"GVFOM: a novel external force for active contour 
based image segmentation". Information 
Sciences 506, 1-18. doi: 
https://doi.org/10.1016/j.ins.2019.08.003. 
[5] Koh, D.M. et al. (2022),” Artificial Intelligence and 
machine learning in Cancer Imaging”. Nature 
News. doi: https://doi.org/10.1038/s43856-022-
00199-0. 
[6] Tan, Shin Hwei, Jooyong Yi, Sergey Mechtaev, and 
Abhik Roychoudhury (2017). "Codeflaws: a 
programming competition benchmark for evaluating 
automated program repair tools". In 2017 
IEEE/ACM 39th International Conference on 
Software Engineering Companion (ICSE-C), pp. 
180-182. IEEE. doi: 10.1109/ICSE-C.2017.76. 
[7] Zhang, Jianpeng, Yutong Xie, Pingping Zhang, Hao 
Chen, Yong Xia, and Chunhua Shen (2019). "Light-
Weight Hybrid Convolutional Network for Liver 
Tumor Segmentation". In IJCAI, vol. 19, pp. 4271-
4277. doi: https://doi.org/10.24963/ijcai.2019/593. 
[8] Ganesh, M., M. Naresh, and C. Arvind (2017). 
"MRI brain image segmentation using enhanced 
adaptive fuzzy K-means algorithm". Intelligent 
Automation & Soft Computing 23, no. 2, 325-330. 
doi: 
https://doi.org/10.1080/10798587.2016.1231472. 
[9] Pandey, Anjali, Rekha Gupta, and Rahul Dubey 
(2018). "Improved brain tumor detection using 
fuzzy rules with image filtering for tumor 
identification". Int Res J Eng Technol (IRJET) 5, 
no. 9, 1-6. 
[10] Sheela, C., and G. J. M. T. Suganthi (2020). 
"Morphological edge detection and brain tumor 
segmentation in Magnetic Resonance (MR) images 
based on region growing and performance 
evaluation of modified Fuzzy C-Means (FCM) 
algorithm". Multimedia Tools and Applications 79, 
no. 25, 17483-17496. doi: 
https://doi.org/10.1007/s11042-020-08636-9. 
[11] Dey, Nilanjan, Venkatesan Rajinikanth, Amira S. 
Ashour, and João Manuel RS Tavares (2018). 
"Social group optimization supported segmentation 
and evaluation of skin melanoma 
images". Symmetry 10, no. 2, 51. doi: 
https://doi.org/10.3390/sym10020051. 
[12] Ghrabat, M.J., Ma, G., Avila, P.L.P., Jassim, M.J. 
and Jassim, S.J. (2019). “Content-based Image 
Retrieval of Color, Shape and Texture by Using 
Novel multi-SVM Classifier”. International Journal 
of Machine Learning and Computing, 9(4), pp.483-
489. doi: 10.18178/ijmlc.2019.9.4.830. 
[13] Rodríguez-Esparza, Erick, Laura A. Zanella-
Calzada, Diego Oliva, Ali Asghar Heidari, Daniel 
Zaldivar, Marco Pérez-Cisneros, and Loke Kok 
Foong (2020)."An efficient Harris hawks-inspired 
image segmentation method". Expert Systems with 
Applications 155,113428. doi: 

86   Informatica 47 (2023) 77–88                                                                                   H.M. Jasim et al. 
https://doi.org/10.1016/j.eswa.2020.113428. 
[14] Wu, Gen-Han, Chen-Yang Cheng, Hong-In Yang, 
and Chung-Te Chena (2018)."An improved water 
flow-like algorithm for order acceptance and 
scheduling with identical parallel machines". 
Applied Soft Computing 71, 1072-1084. doi: 
https://doi.org/10.1016/j.asoc.2017.10.015. 
[15] Abd Elaziz, Mohamed, Ahmed A. Ewees, Dalia 
Yousri, Husein S. Naji Alwerfali, Qamar A. Awad, 
Songfeng Lu, and Mohammed AA Al-Qaness 
(2020)." An improved Marine Predators algorithm 
with fuzzy entropy for multi-level thresholding: 
Real world example of COVID-19 CT image 
segmentation". Ieee Access 8, 125306-125330. doi: 
10.1109/ACCESS.2020.3007928. 
[16] Alam, Md Zahangir, M. Saifur Rahman, and M. 
Sohel Rahman (2019). "A Random Forest based 
predictor for medical data classification using 
feature ranking". Informatics in Medicine Unlocked 
15, 100180. doi: 
https://doi.org/10.1016/j.imu.2019.100180. 
[17] W. Li, P. Cao, D. Zhao, and J. Wang, “Pulmonary 
nodule classification with deep convolutional neural 
networks on computed tomography images,” 
Computational and Mathematical Methods in 
Medicine, vol. 2016, Article ID 6215085, 7 pages, 
2016. doi: 10.1155/2016/6215085. 
[18] J. Kuruvilla and K. Gunavathi, “Lung cancer 
classification using neural networks for CTimages,” 
Computer Methods andPrograms in Biomedicine, 
vol. 113, no. 1, pp. 202–209, 2014. doi: 
https://doi.org/10.1016/j.cmpb.2013.10.011. 
[19] W.-J. Choi and T.-S. Choi, “Automated pulmonary 
noduledetection system in computed tomography 
images: a hierarchical block classification 
approach,” Entropy, vol. 15, no. 2, pp. 507–523, 
2013. doi: https://doi.org/10.3390/e15020507. 
[20] SecS, MahalakshmiR. “An efficient strategy over 
brain tumour analysis using image processing 
technique”. Int J Pharm Technol 2017;9:30687–
96.  ISSN:0975-766X. 
[21] MallickPK, RyuSH, SatapathySK, MishraS, 
NguyenGN, TiwariP. “Brain MRI image 
classification for cancer detection using deep 
wavelet autoencoder-based deep neural network”. 
IEEE Access 2019;7:46278–87. doi: 
10.1109/ACCESS.2019.2902252. 
[22] SeethaJ, RajaSS. “Brain tumor classification using 
convolutional neural net-works”. Biomed 
Pharmacol J 2018;11:1457. doi: 
https://dx.doi.org/10.13005/bpj/1511. 
[23] Abdel-Basset, Mohamed, Victor Chang, and Reda 
Mohamed (2020). "HSMA_WOA: A hybrid novel 
Slime mould algorithm with whale optimization 
algorithm for tackling the image segmentation 
problem of chest X-ray images". Applied soft 
computing 95, 106642. doi: 
https://doi.org/10.1016/j.asoc.2020.106642. 
[24] Ghrabat, Mudhafar Jalil Jassim, Guangzhi Ma 
Abduljabbar Z. A.,  Mustafa A. Al Sibahee, and 
Safa Jalil Jassim (2019). "Greedy learning of deep 
Boltzmann machine (GDBM)’s variance and search 
algorithm for efficient image retrieval". IEEE 
Access 7, 169142-169159. doi: 
10.1109/ACCESS.2019.2948266. 
[25] Chouhan, Siddharth Singh, Ajay Kaul, and Uday 
Pratap Singh (2019). "Image segmentation using 
computational intelligence techniques". Archives of 
Computational Methods in Engineering 26, no. 3, 
533-596. doi: https://doi.org/10.1007/s11831-018-
9257-4. 
[26] Pham T. X., Siarry P., Oulhadj H (2018). 
“Integrating fuzzy entropy clustering with an 
improved PSO for MRI brain image segmentation”. 
Applied Soft Computing., 65(1), 230-242. doi: 
https://doi.org/10.1016/j.asoc.2018.01.003. 
[27] Li, Bing Nan, Chee Kong Chui, Stephen Chang, and 
Sim Heng Ong (2011). "Integrating spatial fuzzy 
clustering with level set methods for automated 
medical image segmentation". Computers in 
biology and medicine 41, no. 1, 1-10. doi: 
https://doi.org/10.1016/j.compbiomed.2010.10.007. 
[28] Nilakant, Rupika, Hema P. Menon, and K. Vikram 
(2017). "A survey on advanced segmentation 
techniques for brain MRI image segmentation". 
International Journal on Advanced Science, 
Engineering and Information Technology 7, no. 4, 
1448-1456. doi: 10.18517/ijaseit.7.4.1271. 
[29] Roy, Sudipta, Debnath Bhattacharyya, Samir 
Kumar Bandyopadhyay, and Tai-Hoon Kim (2017). 
"An effective method for computerized prediction 
and segmentation of multiple sclerosis lesions in 
brain MRI". Computer methods and programs in 
biomedicine 140, 307-320. doi: 
https://doi.org/10.1016/j.cmpb.2017.01.003. 
[30] Chen, Shengcong, Changxing Ding, and Minfeng 
Liu (2019). "Dual-force convolutional neural 
networks for accurate brain tumor segmentation". 
Pattern Recognition 88, 90-100. doi:  
https://doi.org/10.1016/j.patcog.2018.11.009. 
[31] Xu, Yueting, Huiling Chen, Jie Luo, Qian Zhang, 
Shan Jiao, and Xiaoqin Zhang (2019). "Enhanced 
Moth-flame optimizer with mutation strategy for 
global optimization". Information Sciences 492, 
181-203. doi: 
https://doi.org/10.1016/j.ins.2019.04.022. 
[32] Hao, S., Huang, C., Heidari, A. A., Shao, Q., & 
Chen, H. (2023). Performance optimization of 
hunger games search for multi-threshold COVID-19 
image segmentation. Multimedia Tools and 
Applications. doi: https://doi.org/10.1007/s11042-
023-16116-z. 
[33] Ghrabat, Mudhafar Jalil Jassim, Guangzhi Ma, and 
Chih Cheng (2018). "Towards Efficient for 
Learning Model Image Retrieval". In 2018 14th 
International Conference on Semantics, Knowledge 
and Grids (SKG), pp. 92-99. IEEE. doi:  
10.1109/SKG.2018.00020. 
[34] He, Lifang, and Songwei Huang (2017). "Modified 
firefly algorithm based multilevel thresholding for 
color image segmentation". Neurocomputing 240, 
152-174. doi:  

Provably Efficient Multi-Cancer Image Segmentation Based…                                                Informatica 47 (2023) 77–88 87 
https://doi.org/10.1016/j.neucom.2017.02.040. 
[35] Zhao, Dong, Lei Liu, Fanhua Yu, Ali Asghar 
Heidari, Mingjing Wang, Guoxi Liang, Khan 
Muhammad, and Huiling Chen (2021). "Chaotic 
random spare ant colony optimization for multi-
threshold image segmentation of 2D Kapur 
entropy". Knowledge-Based Systems 216, 106510. 
doi: https://doi.org/10.1016/j.knosys.2020.106510. 
[36] Magurran, Anne E. (2021). "Measuring biological 
diversity". Current Biology 31, no. 19, R1174-
R1177. doi: 10.1016/j.cub.2021.07.049. 
[37] Lefcheck, Jonathan S. (2016). "piecewiseSEM: 
Piecewise structural equation modelling in r for 
ecology, evolution, and systematics". Methods in 
Ecology and Evolution 7, no. 5, 573-579. doi: 
https://doi.org/10.1111/2041-210X.12512. 
[38] Maolood, Ismail Yaqub, Yahya Eneid Abdulridha 
Al-Salhi, and Songfeng Lu (2018). "Thresholding 
for medical image segmentation for cancer using 
fuzzy entropy with level set algorithm". Open 
Medicine 13, no. 1, 374-383. doi: 
https://doi.org/10.1515/med-2018-0056. 
[39] Chahal, Prabhjot Kaur, Shreelekha Pandey, and 
Shivani Goel (2020). "A survey on brain tumor 
detection techniques for MR images". Multimedia 
Tools and Applications 79, no. 29, 21771-21814. 
doi: https://doi.org/10.1007/s11042-020-08898-3. 
[40] Ilhan, Umit, and Ahmet Ilhan (2017). "Brain tumor 
segmentation based on a new threshold approach". 
Procedia computer science 120, 580-587. doi: 
https://doi.org/10.1016/j.procs.2017.11.282. 
[41] Benchara, Fatéma Zahra, and Mohamed Youssfi 
(2020). "A new distributed type-2 fuzzy logic 
method for efficient data science models of medical 
informatics". Advances in Fuzzy Systems 2020. 
doi:  https://doi.org/10.1155/2020/6539123. 
[42] Anam, Marrium, Muzammil Hussain, Muhammad 
Waqas Nadeem, Mazhar Javed Awan, Hock Guan 
Goh, and Sadia Qadeer (2021). "Osteoporosis 
prediction for trabecular bone using machine 
learning: a review". Computers, Materials & 
Continua (CMC) 67, no. 1. doi: 
https://doi.org/10.32604/cmc.2021.013159. 
[43] Wang, Aichen, Wen Zhang, and Xinhua Wei 
(2019). "A review on weed detection using ground-
based machine vision and image processing 
techniques". Computers and electronics in 
agriculture 158, 226-240. doi: 
https://doi.org/10.1016/j.compag.2019.02.005. 
[44] Erkan, Uğur, Levent Gökrem, and Serdar Enginoğlu 
(2018). "Different applied median filter in salt and 
pepper noise". Computers & Electrical Engineering 
70, 789-798.  doi: 
https://doi.org/10.1016/j.compeleceng.2018.01.019. 
[45] Ghrabat MJ, Hussien ZA, Khalefa MS, Abduljabbar 
Z. A., Nyangaresi VO, Al Sibahee MA, Abood EW. 
(2022). “Fully automated model on breast cancer 
classification using deep learning classifiers”. 
Indonesian Journal of Electrical Engineering and 
Computer Science, 28(1):183-91. doi: 
10.11591/ijeecs.v28.i1.pp183-191. 
[46] Khalefa, Mustafa S., Zaid Amin Abduljabar, Huda 
Ameer Zeki and Shatt Al-arab. “FINGERPRINT 
IMAGE ENHANCEMENT BY DEVELOP 
MEHTRE TECHNIQUE.” Advanced Computing: 
An International Journal 2 (2011): 171-182. doi: 
10.5121/ACIJ.2011.2616 
[47] Abduljabar, Zaid Amin. “Simulation and Design of 
Fuzzy Temperature Control for Heating and 
Cooling Water System.” International Journal of 
Advancements in Computing Technology 3 (2011): 
41-48. doi: 10.4156/IJACT.VOL3.ISSUE4.5. 
[48]  HemanthDJ, AnithaJ, NaajiA, GemanO, 
PopescuDE, et al. “A modified deep convolutional 
neural network for abnormal brain image 
classification”. IEEE Access 2018;7:4275–83.  doi: 
10.1109/ACCESS.2018.2885639 
[49] AnithaV, MurugavalliS. “Brain tumour 
classification using two-tier classifier with adaptive 
segmentation technique”. IET Comput Vis 
2016;10(1):9–17. doi: https://doi.org/10.1049/iet-
cvi.2014.0193 
[50] Ghrabat, M.J.J., Ma, G., Maolood, I.Y., Alresheedi 
S. S., Abduljabbar Z. A., “An effective image 
retrieval based on optimized genetic algorithm 
utilized a novel SVM-based convolutional neural 
network classifier”. Hum. Cent. Comput. Inf. Sci. 9, 
31 (2019). doi: https://doi.org/10.1186/s13673-019-
0191-8. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

88   Informatica 47 (2023) 77–88                                                                                   H.M. Jasim et al.