IAES International Journal of Artificial Intelligence (IJ-AI)

Vol. 14, No. 2, April 2025, pp. 1183~1191

ISSN: 2252-8938, DOI: 10.11591/ijai.v14.i2.pp1183-1191  1183

Optimizing the gallstone detection process with feature selection

statistical analysis algorithm

Musli Yanto1, Yuhandri1, Muhammad Tajuddin2, Vina Tri Septiana3

1
Department of Information Technology, Faculty of Computer Science, Universitas Putra Indonesia YPTK, Padang, Indonesia
2
Department of Computer Science, Faculty of Engineering, Universitas Bumi Gora, Mataram, Indonesia
3
Department of Radiology, Faculty of Medicine, Baiturrahmah University, Padang, Indonesia

Article Info ABSTRACT

Article history: Early detection is one form of early anticipation in treating gallstone disease
patients using medical images. However, the problem that exists is that there
Received Mar 16, 2024 are still many shortcomings in medical images, such as noise in the image that
Revised Oct 31, 2024 causes the detection process to not run optimally. Based on this, this study
Accepted Nov 14, 2024 aims to carry out the process of detecting gallstone objects in magnetic
resonance cholangiopancreatography (MRCP) images by optimizing the
performance of extraction techniques for feature selection. Optimization of
Keywords: extraction techniques in feature selection is carried out using the performance
of the feature selection statistics analysis (FSSA) algorithm. The performance
Detection of the FSSA algorithm can provide improvements in the feature selection
FSSA process by excelling in the performance of classification methods such as k-
Gallstones nearest neighbor (KNN), support vector machine (SVM), and artificial neural
MRCP network (ANN), and the Pearson correlation (PC) method. Based on the tests
Optimization that have been carried out, the performance of the FSSA algorithm in the
detection process provides an accuracy level of 95.69%, a sensitivity of
89.65%, and a specificity of 98.43%. Overall, this study can contribute to the
development of extraction and provide a significant technical impact on
optimizing the gallstone detection process.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Musli Yanto
Department of Informatics Engineering, Faculty of Computer Science, Universitas Putra Indonesia YPTK
Padang, 25145, Indonesia
Email: [email protected]

1. INTRODUCTION
Gallstones are a disease that can attack the bile organs due to excessive cholesterol content [1].
The effects of excess cholesterol can have a negative impact by forming stone particles in the gallbladder area
[2]. Not only that, this disease also causes pain for quite a long time so early prevention needs to be done [3].
Previous research stated that gallstones are a disease that can threaten human health [4]. Based on this, an early
detection process is really needed by utilizing developments in medical image technology [5].
Medical images are a technology developed in the world of health that plays a role in supporting the
diagnosis process [6]. Medical images or known as biomedical processing technology can contribute to
decision making [7]. The active role of medical images in several previous studies has also had a significant
impact on progress in the health sector [8]. One form of medical image can be seen from magnetic resonance
cholangiopancreatography (MRCP). The implementation of MRCP can be used to assist medical personnel in
the process of diagnosing disease [9]. Previous research explains that MRCP images have been widely used in
the post-operative detection process [10]. Other research also states that MRCP images are a tool in the medical
world that plays an active role in several disease detection processes [11].

Journal homepage: http://ijai.iaescore.com

1184  ISSN: 2252-8938

MRCP images in detection cases have also been used to describe the location of gallstone objects
[12]. However, the fact is that there are still visible shortcomings in MRCP images such as speckle noise
[13], [14]. Based on this, the image processing process is needed to provide an optimal role in the detection
process [15]. The concept of image processing is a technique for analyzing objects contained in an image [16].
Image processing can also be said to be a method that can manipulate images using color, shape, and texture
[17]. Previous research explains that image processing can provide quite good image output in several
processes such as detection [18], [19]. The image processing performance in the previous case has provided an
identification accuracy level of [20]. The same research also reports that the role of image processing can
contribute effectively to the object identification process [21]. Similar research has also proven that the
implementation of image processing in the identification process provides quite good performance [22].
One technique that can be adopted in image processing can be seen based on the performance of
segmentation and extraction techniques. Segmentation is a technique in image processing that is capable of
solving problems in the disease diagnosis process [23]. Segmentation techniques are able to carry out the process
of separating objects in an image quite well [24]. Previous research reported that image segmentation performance
was able to identify gallstone image objects with an accuracy rate of 91% [25]. Furthermore, the same study has
also reported that embed stone objects can be detected well with average precision, recall, and deviation ratio
values of 94.56%, 96.56%, and 98.92% respectively [26]. Segmentation techniques have also been developed in
accordance with the need for solving problems such as identification with fairly good results [27]–[29].
Extraction techniques can also play an important role in digital image processing. This technique is
also experiencing development along with increasing performance in the identification process [30]. Previous
research explains that extraction techniques can provide optimal results in the object classification process [31].
The development of extraction techniques is also presented in model form to provide increased accuracy values
in the identification process [32]. The use of feature selection algorithms in the image extraction process is also
able to make an active contribution to the image processing process [33]. Previous research reported that feature
selection in the image extraction process was able to provide an object selection process by utilizing the
attribute values of an image [34].
Based on previous research, the process of detecting gallstone objects in MRCP images needs to be
optimized by developing a feature selection process in extraction techniques. This optimization is aimed at
maximizing the detection process and results that will be carried out in the development of the feature selection
process using the feature selection statistical analysis (FSSA) algorithm. The performance of the FSSA
algorithm provides an analysis process in determining optimal characteristic patterns involving classification
methods such as k-nearest neighbor (KNN), support vector machine (SVM), and artificial neural network
(ANN) as well as the Pearson correlation (PC) method. The development of the FSSA algorithm in the feature
selection process can also provide novelty in image extraction techniques used in detecting gallstone content.
The performance of FSSA is also expected to be able to present optimal characteristic patterns from previously
detected image objects. Overall, this research can contribute to helping medical parties in the process of
diagnosing patients who have identified gallstone disease.

2. RESEARCH METHOD
The process of detecting gallstone objects in MRCP images is carried out by developing extraction
techniques in the feature selection process based on the performance of the FSSA algorithm. The performance
of the developed FSSA is aimed at providing accurate detection results for gallstone objects on MRCP images.
The performance of the FSSA algorithm is presented in several stages starting from the preprocessing,
segmentation and extraction stages. Based on these stages, the performance of FSSA in the process of
optimizing the gallstone object detection process in feature selection will be able to present new extraction
techniques in image processing. The performance of the FSSA algorithm in the gallstone object detection
process is presented in the research framework in Figure 1.
Figure 1 is an illustration of the performance of the FSSA algorithm in detecting gallstone objects.
The detection process based on the performance of the FSSA algorithm begins with the performance of the
segmentation process which is based on the performance of the multistage segmentation algorithm (MSA) in
Figure 1(a). The MSA algorithm involves k-means cluster-based segmentation (CBS) combined with
morphological segmentation. The output of the segmentation results will later become input for the extraction
process using the performance of the FSSA algorithm in Figure 1(b). The FSSA algorithm presents the
development of the extraction process in feature selection by involving the performance of the KNN, SVM,
and ANN classification methods as well as the PC method in measuring the correlation of each feature pattern
produced. The performance results of the FSSA algorithm are able to provide optimization of the detection
process for gallstone content objects.

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1183-1191

Int J Artif Intell ISSN: 2252-8938  1185

(a)

(b)

Figure 1. Research framework of (a) segmentation process and (b) extraction with feature selection based on
FSSA algorithm performance

Optimizing the gallstone detection process with feature selection statistical analysis algorithm (Musli Yanto)
1186  ISSN: 2252-8938

Next, the performance stages of the FSSA algorithm can be explained as follows:
‒ Image preprocessing stage: this stage is the stage used for the process of improving the input image. Image
improvement includes several processes including gray image transformation, image adjustment, and
filtering. The output of the preprocessing image will later become the input image at the segmentation
process stage.
‒ Image segmentation stage: the segmentation process is an advanced stage in FSSA performance in gallstone
object detection. The segmentation process adopts the performance of the MSA algorithm by playing the
role of the CBS method which is optimized with the elbow method in separating objects. The results of the
segmentation process will later become input in the image extraction process.
‒ Image extraction stage: the extraction process using feature selection was developed by using statistical
analysis methods on the performance of PC. The performance of the PC method can provide an optimal
role in presenting optimal characteristic patterns of gallstone objects. Overall, the improvement in the
feature selection process based on FSSA performance can provide novelty in the process of detecting
gallstone objects.

2.1. Research dataset

The research dataset uses MRCP images sourced from patients at Santa Maria Hospital Pekanbaru
and Siti Rahmah Hospital Padang. The dataset consists of 2371 images from 32 patients with indications of
gallstone disease. The sample research dataset can be presented in Figure 2.
Figure 2 is a sample of MRCP image results used as a research dataset in the gallstone object detection
process. Figure 2(a) depicts an MRCP image of a gallstone patient on image slice 64, Figure 2(b) is also an
MRCP image of a gallstone patient on image slice 65, and Figure 2(c) is also one of the MRCP images of a
gallstone patient on image slice 66. The dataset on the MRCP image (.*jpg format) with an image resolution
of 512×512 pixels. The dataset will later be divided into 1921 training data and 450 data for testing.

(a) (b) (c)

Figure 2. MRCP image dataset: (a) Slice 64, (b) Slice 65, and (c) Slice 66

2.2. Multistage segmentation algorithm

The segmentation process with the MSA algorithm adopts the performance of CBS combined with
morphological segmentation. CBS segmentation involves the k-means cluster which is optimized using the
elbow method. The performance of the elbow method can present optimal cluster (k) values based on the
calculation of the sum square error (SSE) [35]. The equations used in the performance of the MSA algorithm
are presented in (1)-(3) [36], [37].
In (1) is the formula used to measure distance using Euclidian distance in the k-means cluster process.
Distance measurements are calculated on image intensity to group image objects. In (2) is the formula used to
calculate the SSE in the elbow method. The formula is used as a form of optimizing the cluster process in CBS
in separating image objects. The combination of (1) and (2) produces (3) in determining the optimal k value
used in the CBS-based cluster process.

(𝑥, 𝑦) = √∑𝑘1 (𝑥𝑖 − 𝑦𝑖 )2 ; 𝑘 = 1,2,3, … . , 𝑛 (1)

𝑆𝑆𝐸 = ∑𝑛𝑘=1‖𝑥𝑖 − 𝑐𝑘 ‖2 (2)

d(x,y)
K_Elbow= ∑nk=1 ‖√∑k1(xi − yi )2 − ∑k d(x,y) ‖ (3)
1

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1183-1191

Int J Artif Intell ISSN: 2252-8938  1187

2.3. Feature selection

Feature selection in image extraction is used as a detection process by utilizing the attribute values for
each image feature. The feature selection process adopts the performance of the data classification process by
taking the level of accuracy as a selection parameter [38]. The feature selection process uses several methods
such as KNN, SVM, and ANN. KNN is an algorithm concept in supervised learning that considers the use of
many k comparable patterns in the training [39]. SVM has contributed greatly to handling classification
problems based on hyperplanes [40]. The performance of ANN has also been proven to provide maximum
results such as identification, classification, and prediction problems [41], [42].

2.4. Person correlation

PC is a statistical analysis concept that is capable of measuring correlation [43]. PC methods can be
combined to improve analysis performance with quite good output [44]. The PC method has also been used to
review the accuracy of a model analysis [45]. Regarding PC performance, it can be seen in (4) and (5) [46].
In (4) and (5) is the formula used to calculate the value of cov (X, Y) which is the covariance between
X and Y. The value of X, Y can be interpreted as a standard deviation value of the variables X and Y [47]. PC
performance has contributed to maximizing the image processing process in a model that has been designed [48].
cov (X,Y) E ((cov (X−µX)(Y−µY))
𝑃 𝑋, 𝑌 = ( )=( ) (4)
𝜎𝑋.𝜎𝑌 𝜎𝑋.𝜎𝑌

cov (X,Y) E (XY)− E (X)E (Y)

𝑃 𝑋, 𝑌 = ( )= (5)
𝜎𝑋.𝜎𝑌 √E (X.X)−𝐸.𝐸(𝑋) √𝐵 (𝑌.𝑌)−𝐸.𝐸 (𝑌)

3. RESULTS AND DISCUSSION

Optimizing the gallstone object detection process by developing feature selection using FSSA is the
main topic of research. The development of feature selection based on the performance of the FSSA algorithm in
the extraction process is a novelty to provide improvements in the detection of gallstone objects. The performance
of the FSSA results that have been developed will be able to provide a detection model that provides accuracy.

3.1. Image preprocessing

The preprocessing stage is the initial stage of the gallstone object detection process. This process plays
a role in providing an increase in the quality of the input image used. Preprocessing results involve several
processes including gray image transformation, image adjustment, and filtering. The preprocessing results can
be presented in Figure 3.
Figure 3 is the output of the preprocessing stage using the gray image transformation process, image
adjustment, and filtering. Figure 3(a) is the input image in the detection process. Figure 3(b) is the gray image
transformation process which is the beginning of preprocessing. Figure 3(c) is the result of continued
preprocessing involving the image adjustment process. Figure 3(d) is the final stage of preprocessing, which
presents the image output from the filtering process. In the preprocessing stage can be seen that there is an
improvement in the quality of the input image used previously. The output of the preprocessing image will be
the input in the detection process.

(a) (b) (c) (d)

Figure 3. Preprocessing result (a) input image, (b) gray transformation, (c) image adjustment, and
(d) results filtering

3.2. Multistage segmentation algorithm

The segmentation process using the MSA algorithm is part of the gallstone object detection process.
The performance of the MSA algorithm adopts a CBS approach and operational morphology to guarantee the
accuracy of segmented objects with precise and accurate results. The performance results of the MSA algorithm
in segmentation can be presented in Figure 4.
Optimizing the gallstone detection process with feature selection statistical analysis algorithm (Musli Yanto)
1188  ISSN: 2252-8938

Figure 4 is the result of the segmentation process of the MSA algorithm performance in detecting
gallstone disease. Figure 4(a) is the result of the previous preprocessing image which is used as the input image
in the segmentation process with the MSA algorithm. Figure 4(b) is the result of CBS segmentation which is
one part of the segmentation process in the MSA algorithm. Figure 4(c) is the result of morphological
segmentation which is the next stage in the segmentation process in the MSA algorithm. Figure 4(d) is the final
output of the MSA algorithm segmentation process in detecting gallstone objects. The performance of the MSA
algorithm has been able to describe the object of gallstone disease quite well. The results of the MSA algorithm
segmentation will be used as input in the image extraction process.

(a) (b) (c) (d)

Figure 4. Results of the MSA algorithm segmentation process (a) preprocessing result, (b) CBS results,
(c) morphology results, and (d) MSA results

3.3. Feature selection statistical analysis algorithm extraction

Development of feature selection in the delay image extraction process using FSSA for the process of
increasing detection. The FSSA performance process plays the role of the classification method and correlation
examiners found in the image characteristic pattern. Classification methods such as KNN, SVM, and ANN are
involved in finding optimal feature patterns. The performance results of the FSSA algorithm in the feature
selection process can be presented in Table 1. Based on Table 1, it can be seen that the feature selection process
by adopting the performance of the FSSA algorithm has gone well in determining image characteristic patterns.
The results of the feature selection process which involves several classification methods with PC statistical
analysis methods have been able to illustrate a more effective feature selection process. The performance test
results of the FSSA algorithm can be presented in Table 2.
Table 1 is the result of measuring the performance of the FSSA algorithm in detecting gallstone objects.
Based on this table, it can be seen that the texture feature outperforms other features in presenting optimal image
characteristic patterns. This characteristic pattern can be seen based on the total performance accuracy of the KNN
classification method of 84.61%, SVM 90.38%, and ANN 99.95%. The results of correlation measurements using
the PC method have presented a correlation level of 95.90%. Based on these results, the performance of the FSSA
method is quite good, providing an improvement in the detection process for gallstone objects.
Improving the gallstone object detection process using the FSSA algorithm can provide novelty in the
feature selection process in extraction techniques. Testing the performance of FSSA in increasing detection
has been proven to be quite good with an accuracy of 95.83%, sensitivity of 96.96%, and specificity of 95.23%.
Testing based on several previous studies is also a process of proving the performance of the FSSA algorithm
in improving the process of detecting gallstone objects presented in Table 3.
Table 3 is a form of testing the performance results of the FSSA algorithm with previous research in
the detection process. Based on these results, it can be stated that the FSSA algorithm can provide accurate
image characteristic patterns in the gallstone object detection process. Overall, this research has been quite
successful and can contribute to optimizing the process of diagnosing gallstone disease.

Table 1. The performance results of the FSSA algorithm in the feature selection process
Shape features Texture features Combination of shapes and textures
Area 100 119 418 - 100 119 21
Perimeter 38.117 40.466 86.540 - 38.117 40.466 12.756
Metric 0.864913 0.913222 0.701378 - 0.864913 0.913222 1.6261
Eccentricity 0.833286 0.847956 0.693967 - 0.833286 0.847956 N/A
Y 1 1 1 1 1 1 1 1 1
Contrast - 0.003256 0.002554 0.002527 0.003256 0.002554 0.0009
Correlation - 0.627183 0.704434 0.849754 0.627183 0.704434 0.569403
Energy - 0.99835 0.998167 0.995909 0.99835 0.998167 0.9199
Homogeneity - 0.999722 0.999732 0.999661 0.999722 0.999732 0.9999
K-NN (%) 80.77 84.62 82.69
SVM (%) 88.40 88.46 86.54
ANN (%) 99.98 99.98 99.97

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1183-1191

Int J Artif Intell ISSN: 2252-8938  1189

Table 2. The performance test results of the FSSA algorithm (%)

K-NN accuracy SVM accuracy ANN accuracy
Analysis
Features Total Total Total
Training Testing Training Testing Training Testing (PC)
accuracy accuracy accuracy
Shape
61.53 100 80.76 69.23 100 84.61 99.46 99.99 99.73 80.04
features
Texture
69.23 100 84.61 80.76 100 90.38 99.93 99.97 99.95 95.90
features
Combination
of shapes 65.38 100 82.69 69.23 100 84.62 99.90 99.94 99.92 87.96
and textures

Table 3. Comparison of the performance of the FSSA algorithm in improving the process of detecting
gallstone objects with previous research
No Previous research results Performance results of the FSSA algorithm
1 The results of the tests that have been carried out show that the performance Presents the development of the image
accuracy level of the multiclass classification method combined with deep extraction process with the FSSA feature in
learning reaches 93.4% and 94.36% [48]. detecting gallstone objects. Performance
2 The CNN classification method with feature extraction and feature selection in testing of FSSA performance in increasing
the detection process provides output with an accuracy level of 0.6536% and detection provided an accuracy rate of
0.8942% for the training and testing [38]. 95.83%, sensitivity of 96.96%, and
3 The MSA algorithm uses coarse network segmentation with shape operations in specificity of 95.23%. Based on these
object detection providing a detection accuracy rate of 90% [49]. results, the FSSA algorithm can present a
4 MSA algorithm development was carried out using the Enhancement technique novelty in the form of an effective and
using the feature improvement pyramid network (multi-stage FEPN) providing efficient algorithm for improving the
an accuracy rate of 92.1% [50]. gallstone object detection process.

4. CONCLUSION
Improving the gallstone object detection process by developing feature selection with the FSSA
algorithm has provided maximum output results. These results are based on the results of FSSA performance
testing with an accuracy rate of 95.83%, sensitivity of 96.96%, and specificity of 95.23%. Based on these
results, it can be proven that the FSSA algorithm can provide optimal results in the process of detecting
gallstone objects. The overall performance of the FSSA algorithm can be used as a novelty in extraction
techniques, especially in the feature selection process.

ACKNOWLEDGEMENTS
The author would like to express his gratitude to Dr. Hj. Zerni Melmusi, SE, MM, Ak, CA, as the
Chairperson of the YPTK Padang Computer College Foundation who has provided support in this research
based on letter no. 004/PSDTI/UPIYPTK/SB/V/2023, regarding the appointment of experts in this research.

REFERENCES
[1] D. A. Goldman, “Gallbladder, gallstones, and diseases of the gallbladder in children,” Pediatrics in Review, vol. 41, no. 12,
pp. 623–629, Dec. 2020, doi: 10.1542/pir.2019-0077.
[2] A. M. Madden, D. Trivedi, N. C. Smeeton, and A. Culkin, “Modified dietary fat intake for treatment of gallstone disease,” Cochrane
Database of Systematic Reviews, vol. 2021, no. 6, Jun. 2021, doi: 10.1002/14651858.CD012608.pub2.
[3] P. Cianci and E. Restini, “Management of cholelithiasis with choledocholithiasis: endoscopic and surgical approaches,” World
Journal of Gastroenterology, vol. 27, no. 28, pp. 4536–4554, Jul. 2021, doi: 10.3748/wjg.v27.i28.4536.
[4] L. Pan et al., “The treatment of cholecystitis and cholelithiasis by tibetan medicine,” Evidence-based Complementary and
Alternative Medicine, vol. 2021, pp. 1–21, Sep. 2021, doi: 10.1155/2021/9502609.
[5] J. Wang, H. Zhu, S. H. Wang, and Y. D. Zhang, “A review of deep learning on medical image analysis,” Mobile Networks and
Applications, vol. 26, no. 1, pp. 351–380, 2021, doi: 10.1007/s11036-020-01672-7.
[6] S. Suganyadevi, V. Seethalakshmi, and K. Balasamy, “A review on deep learning in medical image analysis,” International Journal
of Multimedia Information Retrieval, vol. 11, no. 1, pp. 19–38, Mar. 2022, doi: 10.1007/s13735-021-00218-1.
[7] X. Liu, L. Song, S. Liu, and Y. Zhang, “A review of deep-learning-based medical image segmentation methods,” Sustainability,
vol. 13, no. 3, Jan. 2021, doi: 10.3390/su13031224.
[8] W. Zhou, H. Wang, and Z. Wan, “Ore image classification based on improved CNN,” Computers and Electrical Engineering,
vol. 99, Apr. 2022, doi: 10.1016/j.compeleceng.2022.107819.
[9] M. A. S. Ai et al., “Real-time facemask detection for preventing covid-19 spread using transfer learning based deep neural network,”
Electronics, vol. 11, no. 14, Jul. 2022, doi: 10.3390/electronics11142250.
[10] L. Wei, S. L. Zhang, N. Xie, C. M. Li, and B. M. Fu, “The value of MRCP in children with biliary symptomatology–an essential
adjunct for safe cholecystectomy,” South African Journal of Surgery, vol. 60, no. 1, pp. 67–69, 2022, doi: 10.17159/2078-
5151/2022/v60n1a3430.
[11] A. Javaid, R. Mahmood, H. Ulah, M. Shafiq, N. Dildar, and G. Abbas, “Diagnostic accuracy of magnetic resonance
cholangiopancreatography in the detection of choledocholith, taking post-operative findings as the gold standard,” Pakistan Armed
Forces Medical Journal, vol. 73, no. 2, pp. 394–397, Apr. 2023, doi: 10.51253/pafmj.v73i2.7015.

Optimizing the gallstone detection process with feature selection statistical analysis algorithm (Musli Yanto)
1190  ISSN: 2252-8938

[12] M. Higashi, M. Tanabe, K. Ihara, E. Iida, M. Furukawa, and K. Ito, “Bile flow dynamics in patients with cholelithiasis: an evaluation
with cine-dynamic magnetic resonance cholangiopancreatography using a spatially selective inversion-recovery pulse,”
Tomography, vol. 8, no. 2, pp. 815–823, Mar. 2022, doi: 10.3390/tomography8020067.
[13] S. Rajput, A. Singh, and R. Gupta, “Automated kidney stone detection using image processing techniques,” in 2021 9th
International Conference on Reliability, Infocom Technologies and Optimization (Trends and Future Directions), ICRITO 2021,
IEEE, Sep. 2021, pp. 1–5, doi: 10.1109/ICRITO51393.2021.9596175.
[14] E. C. Yepez, L. M. L. Carrillo, G. Botella, and I. Amerini, “Introduction to the special section on image processing in security
applications (VSI-IPSA),” Computers and Electrical Engineering, vol. 89, 2021, doi: 10.1016/j.compeleceng.2020.106935.
[15] T. D. Shashikala, B. L. Sunitha, S. Basavarajappa, and J. P. Davim, “Some studies on measurement of worn surface by digital image
processing,” International Journal of Image and Graphics, vol. 23, no. 2, Mar. 2023, doi: 10.1142/S021946782350016X.
[16] J. Grühn, M. Vogel, and N. Kockmann, “Digital image processing of gas‐liquid reactions in coiled capillaries,” Chemie-Ingenieur-
Technik, vol. 93, no. 5, pp. 825–829, May 2021, doi: 10.1002/cite.202000240.
[17] M. Bakthavatchalam et al., “Morphological and elemental mapping of gallstones using synchrotron microtomography and
synchrotron x‐ray fluorescence spectroscopy,” JGH Open, vol. 3, no. 5, pp. 381–387, 2019, doi: 10.1002/jgh3.12171.
[18] I. R. I. Haque and J. Neubert, “Deep learning approaches to biomedical image segmentation,” Informatics in Medicine Unlocked,
vol. 18, 2020, doi: 10.1016/j.imu.2020.100297.
[19] M. Ji and Z. Wu, “Automatic detection and severity analysis of grape black measles disease based on deep learning and fuzzy
logic,” Computers and Electronics in Agriculture, vol. 193, Feb. 2022, doi: 10.1016/j.compag.2022.106718.
[20] K. Buschbacher, D. Ahrens, M. Espeland, and V. Steinhage, “Image-based species identification of wild bees using convolutional
neural networks,” Ecological Informatics, vol. 55, Jan. 2020, doi: 10.1016/j.ecoinf.2019.101017.
[21] A. R. Javed and Z. Jalil, “Byte-level object identification for forensic investigation of digital images,” in 1st Annual International
Conference on Cyber Warfare and Security, ICCWS 2020-Proceedings, IEEE, Oct. 2020, pp. 1–4, doi:
10.1109/ICCWS48432.2020.9292387.
[22] S. Tavakoli, A. Ghaffari, Z. M. Kouzehkanan, and R. Hosseini, “New segmentation and feature extraction algorithm for
classification of white blood cells in peripheral smear images,” Scientific Reports, vol. 11, no. 1, Sep. 2021, doi: 10.1038/s41598-
021-98599-0.
[23] X. Chen et al., “Diverse data augmentation for learning image segmentation with cross-modality annotations,” Medical Image
Analysis, vol. 71, Jul. 2021, doi: 10.1016/j.media.2021.102060.
[24] X. Lei, X. Yu, J. Chi, Y. Wang, J. Zhang, and C. Wu, “Brain tumor segmentation in mr images using a sparse constrained level set
algorithm,” Expert Systems with Applications, vol. 168, Apr. 2021, doi: 10.1016/j.eswa.2020.114262.
[25] S. Pang et al., “A novel yolov3-arch model for identifying cholelithiasis and classifying gallstones on CT images,” PLoS ONE,
vol. 14, no. 6, Jun. 2019, doi: 10.1371/journal.pone.0217647.
[26] D. Li, B. Du, Y. Shen, and L. Ge, “Artificial intelligence‐assisted visual sensing technology under duodenoscopy of gallbladder
stones,” Journal of Sensors, vol. 2021, no. 1, 2021, doi: 10.1155/2021/5158577.
[27] A. G. Molsosa, H. A. Orengo, D. Lawrence, G. Philip, K. Hopper, and C. A. Petrie, “Potential of deep learning segmentation for
the extraction of archaeological features from historical map series,” Archaeological Prospection, vol. 28, no. 2, pp. 187–199, Apr.
2021, doi: 10.1002/arp.1807.
[28] H. Wang, W. Zhang, S. Li, Y. Fan, M. Feng, and R. Wang, “Development and evaluation of deep learning-based automated
segmentation of pituitary adenoma in clinical task,” The Journal of Clinical Endocrinology & Metabolism, vol. 106, no. 9,
pp. 2535–2546, Aug. 2021, doi: 10.1210/clinem/dgab371.
[29] C. P. Jayapandian et al., “Development and evaluation of deep learning–based segmentation of histologic structures in the kidney
cortex with multiple histologic stains,” Kidney International, vol. 99, no. 1, pp. 86–101, Jan. 2021, doi: 10.1016/j.kint.2020.07.044.
[30] S. Latha, P. Muthu, S. Dhanalakshmi, R. Kumar, K. W. Lai, and X. Wu, “Emerging feature extraction techniques for machine
learning-based classification of carotid artery ultrasound images,” Computational Intelligence and Neuroscience, vol. 2022, pp. 1–
14, May 2022, doi: 10.1155/2022/1847981.
[31] G. Latif, “DeepTumor: framework for brain mr image classification, segmentation and tumor detection,” Diagnostics, vol. 12,
no. 11, Nov. 2022, doi: 10.3390/diagnostics12112888.
[32] R. Bygari, K. Rithesh, S. Ambesange, and S. G. Koolagudi, “Prostate cancer grading using multistage deep neural networks,” in
Machine Learning, Image Processing, Network Security and Data Sciences, pp. 271–283, 2023, doi: 10.1007/978-981-19-5868-7_21.
[33] T. S. Priya, “Resnet based feature extraction with decision tree classifier for classificaton of mammogram images,” Turkish Journal of
Computer and Mathematics Education (TURCOMAT), vol. 12, no. 2, pp. 1147–1153, Apr. 2021, doi: 10.17762/turcomat.v12i2.1136.
[34] Y. K. Ganeeva and E. V. Myasnikov, “Identifying persons from iris images using neural networks for image segmentation and
feature extraction,” Computer Optics, vol. 46, no. 2, pp. 308–316, 2022, doi: 10.18287/2412-6179-CO-1023.
[35] A. B. Habib, “Elbow method vs silhouette co-efficient in determining the number of clusters,” Research Gate, pp. 1-7, 2021, doi:
10.13140/RG.2.2.27982.79688.
[36] S. S. Yu, S. W. Chu, C. M. Wang, Y. K. Chan, and T. C. Chang, “Two improved k-means algorithms,” Applied Soft Computing
Journal, vol. 68, pp. 747–755, Jul. 2018, doi: 10.1016/j.asoc.2017.08.032.
[37] M. Cui, “Introduction to the k-means clustering algorithm based on the elbow method,” Accounting, Auditing and Finance, vol. 1,
pp. 5–8, 2020.
[38] B. S. Riza, J. Na’am, and S. Sumijan, “Convolutional neural network as an image processing technique for classification of bacilli
tuberculosis extra pulmonary (TBEP) disease,” TEM Journal, vol. 11, no. 3, pp. 1331–1340, Aug. 2022, doi: 10.18421/TEM113-43.
[39] H. Salem, M. Y. Shams, O. M. Elzeki, M. A. Elfattah, J. F. Al‐amri, and S. Elnazer, “Fine-tuning fuzzy knn classifier based on
uncertainty membership for the medical diagnosis of diabetes,” Applied Sciences, vol. 12, no. 3, Jan. 2022, doi:
10.3390/app12030950.
[40] T. S. Reddy and J. Harikiran, “Hyperspectral image classification using support vector machines,” IAES International Journal of
Artificial Intelligence, vol. 9, no. 4, pp. 684–690, 2020, doi: 10.11591/ijai.v9.i4.pp684-690.
[41] G. Shobana and N. Priya, “Cancer drug classification using artificial neural network with feature selection,” in Proceedings of the
3rd International Conference on Intelligent Communication Technologies and Virtual Mobile Networks, ICICV 2021, IEEE, Feb.
2021, pp. 1250–1255, doi: 10.1109/ICICV50876.2021.9388542.
[42] M. Yanto, S. Sanjaya, Yulasmi, D. Guswandi, and S. Arlis, “Implementation multiple linear regresion in neural network predict
gold price,” Indonesian Journal of Electrical Engineering and Computer Science, vol. 22, no. 3, pp. 1635–1642, Jun. 2021, doi:
10.11591/ijeecs.v22.i3.pp1635-1642.
[43] H. Pan, X. You, S. Liu, and D. Zhang, “Pearson correlation coefficient-based pheromone refactoring mechanism for multi-colony
ant colony optimization,” Applied Intelligence, vol. 51, no. 2, pp. 752–774, Feb. 2021, doi: 10.1007/s10489-020-01841-x.

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1183-1191

Int J Artif Intell ISSN: 2252-8938  1191

[44] P. Waldmann, “On the use of the pearson correlation coefficient for model evaluation in genome-wide prediction,” Frontiers in
Genetics, vol. 10, Sep. 2019, doi: 10.3389/fgene.2019.00899.
[45] J. Cai, M. H. Zhang, Y. T. Zhu, and Y. H. Liu, “Model of freight vehicle energy consumption based on pearson correlation
coefficient,” Jiaotong Yunshu Xitong Gongcheng Yu Xinxi/Journal of Transportation Systems Engineering and Information
Technology, vol. 18, no. 5, pp. 241–246, 2018, doi: 10.16097/j.cnki.1009-6744.2018.05.035.
[46] H. Zhu, X. You, and S. Liu, “Multiple ant colony optimization based on pearson correlation coefficient,” IEEE Access, vol. 7,
pp. 61628–61638, 2019, doi: 10.1109/ACCESS.2019.2915673.
[47] A. Bahrami, A. Karimian, E. Fatemizadeh, H. Arabi, and H. Zaidi, “A new deep convolutional neural network design with efficient
learning capability: application to ct image synthesis from mri,” Medical Physics, vol. 47, no. 10, pp. 5158–5171, Oct. 2020, doi:
10.1002/mp.14418.
[48] Z. Wang, E. Wang, and Y. Zhu, “Image segmentation evaluation: a survey of methods,” Artificial Intelligence Review, vol. 53,
no. 8, pp. 5637–5674, Dec. 2020, doi: 10.1007/s10462-020-09830-9.
[49] T. Ali et al., “Multistage segmentation of prostate cancer tissues using sample entropy texture analysis,” Entropy, vol. 22, no. 12,
2020, doi: 10.3390/e22121370.
[50] K. Zhang and H. Shen, “Multi-stage feature enhancement pyramid network for detecting objects in optical remote sensing images,”
Remote Sensing, vol. 14, no. 3, 2022, doi: 10.3390/rs14030579.

BIOGRAPHIES OF AUTHORS

Musli Yanto was born in Jakarta on July 7, 1989. Currently, he is a lecturer at the
Putra Indonesia University YPTK Padang. The educational history of the Informatics
Engineering Undergraduate Program was completed in 2012 and the Informatics Engineering
Masters Program in 2014. His areas of expertise include artificial intelligence (AI), expert
systems (ES), data mining (DM), and decision support systems (DSS). He can be contacted at
email: [email protected].

Prof. Dr. Yuhandri, S. Kom., M. Kom. was born in Tanjung Alam on May 15,
1973. He is an Assistant Professor in Faculty of Computer Science, Universitas Putra Indonesia
YPTK. He received the bachelor’s degree in informatics management and master’s degree in
information tecnology in 1992 and 2006 from Universitas Putra Indonesia YPTK. Moreover,
he completed his Doctor of Information Technology as Informatics Medical Image Expertise
from Gunadarma University in April 2017. He is a lecturer at the Faculty of Computer Science,
Universitas Putra Indonesia YPTK. He can be contacted at email: [email protected].

Muhammad Tajuddin is a lecturer at Bumigora University. He serves as a

Professor at the Faculty of Computer Science. He obtained a Bachelor's degree at Matrama
University in 1986. Then a Master of Science degree in 2000 at Brawijaya University. He also
achieved a doctorate degree in 2012 at Brawijaya University. Teaching history includes
research methods, management information systems, decision support systems, software
engineering, and information technology research. He can be contacted at email:
[email protected].

Vina Tri Septiana, Sp.Rad., dr. S. Ked. is a radiology doctor and also works as
a lecturer at Baiturrahmah University, Padang City, West Sumatra Province. History of medical
studies graduated in 2006 then completed professional studies in 2008 at Andalas University.
In 2018 he also completed Sp-1 studies at the University of Indonesia. Skills and expertise
include diagnostic imaging, diagnostic radiology, computed tomography, magnetic resonance,
imaging, medical imaging, radiography, clinical imaging, ultrasound imaging, and
neuroimaging. She can be contacted at email: [email protected].

Optimizing the gallstone detection process with feature selection statistical analysis algorithm (Musli Yanto)