Journal of Advanced Research in Applied Sciences and Engineering Technology 31, Issue 2 (2023) 234-244
Journal of Advanced Research in Applied
Sciences and Engineering Technology
Journal homepage:
https://semarakilmu.com.my/journals/index.php/applied_sciences_eng_tech/index
ISSN: 2462-1943
Preliminary Analysis on the Effect of Different Denoising Techniques
towards Texture Features of MRI Images of Alzheimer’s Disease
Muhammad Fathi Mohd Zain1, Wan Mahani Hafizah Wan Mahmud1,*, Hong-Seng Gan2
1
Department of Electronic Engineering, Faculty of Electrical and Electronics Engineering, Universiti Tun Hussein Onn Malaysia, Batu Pahat, 86400
Johor, Malaysia
2 Department of Data Science, Universiti Malaysia Kelantan, UMK City Campus, Pengkalan Chepa, 16100 Kelantan, Malaysia
ARTICLE INFO ABSTRACT
Article history: Early detection of Alzheimer’s disease (AD) has become one of the major research topics
Received 10 February 2023 nowadays. The utilization of the computerized system may help medical experts to
Received in revised form 25 June 2023 better understand and analyse the magnetic resonance imaging (MRI) images of AD
Accepted 1 July 2023 patients for early detection. One of the commonly steps taken for the analysis of the
Available online 17 July 2023
image is image denoising using certain filters. However, finding shows that previous
researchers use different approaches. This study aims to analyse the effect of different
denoising techniques towards detection of Alzheimer’s disease. Data of two different
groups (AD patients and Normal Control) were collected from Alzheimer's Disease
Neuroimaging Initiative (ADNI). Then brain extraction and skull stripping were
performed. Several image denoising techniques were implemented for both groups
namely median filter, Wiener filter, histogram equalization filter and Gaussian low pass
filter. After that, all images underwent texture feature extraction process and analysis
were made to see the effect of those denoising techniques towards the features of
Gray-Level Co-occurrence Matrix (GLCM) extracted which are the contract, correlation,
energy and homogeneity features. The result shows that the use of mentioned
denoising filters do not give effect to the extracted features. All values of contrast,
correlation, energy and homogeneity cannot clearly distinguish between AD and NC
groups. Without any filters on the other hand, contrast feature gives the best output in
Keywords: distinguishing between AD and NC groups with the normalized value of 0.1. The result
Alzheimer’s Disease; MRI; image from this study may help in thorough investigation of other features or hybrid features
denoising; feature extraction; GLCM that could be used for the purpose of detection and classification of AD.
1. Introduction
Alzheimer’s disease (AD) is a neurological disease affecting the brain function that may lead to
memory loss in patients and could affect the daily life of both patients and caretakers [1]. This disease
has no cure and the patients’ condition may worsen due to the progression of the disease, which may
in the end lead to death. Researchers all over the world try to come up with a simple and more
accurate approach in order to identify AD before the symptoms become visible but most often, AD
*
Corresponding author.
E-mail address:
https://doi.org/10.37934/araset.31.2.234244
234
, Journal of Advanced Research in Applied Sciences and Engineering Technology
Volume 31, Issue 2 (2023) 234-244
detection is not that accurate. Usually, accuracy will be improved once the patient has started to
show the signs of the disease. Therefore, detection of AD at its earliest stage is essential so that
proper treatment could be introduced thus slowing down or even preventing the patients from
further brain damage [2].
Over the past few decades, magnetic resonance imaging (MRI) has been progressively used in
exploring brain anatomy including for AD [3]. In the diagnosis of AD using medical imaging modalities,
specifically using MRI images, some lesion areas in the brain could be determined with the help of
expert radiographers. Image processing technology in addition can be used to for a better quality of
reconstruction and measurement process, especially for brain, soft tissues and lesions. With the help
of sophisticated and high-end computers, medical experts can analyze multiple areas of interest
qualitatively and quantitatively [4]. Current practice on detecting AD is by looking into the brain
scans, performing a clinical assessment, and eventually asking questions of the patient and their
relatives [5,6]. This is a very challenging process as identifying the parts of the brain that are affected
by AD by using the bare eyes is not an easy task. Moreover, one of the AD symptoms, such as brain
shrinkage, can be observed as well in healthy, elderly normal control (NC) groups [7].
Therefore, computer-aided diagnosis (CAD) systems which utilize computerized and digital image
analysis can be used as an alternative solution to the problem. CAD can be developed based on image
analysis techniques using deep learning methods or common traditional methods [8]. Utilization of
deep learning frameworks for detection of AD in MRI images can be seen as in previous studies
[2,4,8]. Even though the deep learning methods can give great output when dealing with brain MRI
analysis, finding a robust and more generic algorithm still remain as a challenge. Pre-processing stage
can also affect the performance of deep learning techniques [3]. On the other hand, the common
framework which uses the pipeline of pre-processing, segmentation, feature extraction and
classification can be utilized using supervised or unsupervised machine learning approaches such as
Support Vector Machine (SVM) [9,10]. Detecting AD using these algorithms has their own issues as
well including the low image quality, issues in the brain segmentation and pre-processing steps, and
the complexity of medical images [11]. Thus, detailed analysis is needed to thoroughly investigate
each step involved in the pipeline.
Image pre-processing is usually performed before any further image analysis to remove any
possible noise or unwanted data, as well as to highlight and enhance important features for feature
extraction process. Due to the complexity of medical images construction, the images may contain
noise and distortion which may generally cause by the variations in the detector sensitivity,
diminished illumination of the object, limitation of the images, and spontaneous variations in the
radiation signal [12,13]. Therefore, it is crucial to pre-process the data to enhance its quality or to
optimize its geometric and intensity patterns [12]. The pre-processing step will help the researchers
to improve the quality which will then highlight the essential information that is required in the
feature extraction and classification process. There are many available image pre-processing steps
including intensity normalization, contrast enhancement, image denoising, brain extraction, skull
stripping, and others. Some researchers performed the pre-processing steps using secondary
tools/software such as the Brain Extractor Tools (BET2), Freesurfer Statistical Parameter Mapping
(SPM12), and Computer Anatomy Toolbox (CAT12) [2,14-17]. The implementation of image denoising
step is also varies between researchers where there were previous reported studies which uses the
more common filters such as median filter, Wiener filter, histogram equalization, and Gaussian low
pass filter [9,10,18,19]. There was also a study that implemented other types of filters such as Lucy-
Richardson approach [20,21]. Besides, there were some past studies which did not even implement
any image denoising process [8,22-24]. Due to these stated inconsistencies in the process, the effect
235
Journal of Advanced Research in Applied
Sciences and Engineering Technology
Journal homepage:
https://semarakilmu.com.my/journals/index.php/applied_sciences_eng_tech/index
ISSN: 2462-1943
Preliminary Analysis on the Effect of Different Denoising Techniques
towards Texture Features of MRI Images of Alzheimer’s Disease
Muhammad Fathi Mohd Zain1, Wan Mahani Hafizah Wan Mahmud1,*, Hong-Seng Gan2
1
Department of Electronic Engineering, Faculty of Electrical and Electronics Engineering, Universiti Tun Hussein Onn Malaysia, Batu Pahat, 86400
Johor, Malaysia
2 Department of Data Science, Universiti Malaysia Kelantan, UMK City Campus, Pengkalan Chepa, 16100 Kelantan, Malaysia
ARTICLE INFO ABSTRACT
Article history: Early detection of Alzheimer’s disease (AD) has become one of the major research topics
Received 10 February 2023 nowadays. The utilization of the computerized system may help medical experts to
Received in revised form 25 June 2023 better understand and analyse the magnetic resonance imaging (MRI) images of AD
Accepted 1 July 2023 patients for early detection. One of the commonly steps taken for the analysis of the
Available online 17 July 2023
image is image denoising using certain filters. However, finding shows that previous
researchers use different approaches. This study aims to analyse the effect of different
denoising techniques towards detection of Alzheimer’s disease. Data of two different
groups (AD patients and Normal Control) were collected from Alzheimer's Disease
Neuroimaging Initiative (ADNI). Then brain extraction and skull stripping were
performed. Several image denoising techniques were implemented for both groups
namely median filter, Wiener filter, histogram equalization filter and Gaussian low pass
filter. After that, all images underwent texture feature extraction process and analysis
were made to see the effect of those denoising techniques towards the features of
Gray-Level Co-occurrence Matrix (GLCM) extracted which are the contract, correlation,
energy and homogeneity features. The result shows that the use of mentioned
denoising filters do not give effect to the extracted features. All values of contrast,
correlation, energy and homogeneity cannot clearly distinguish between AD and NC
groups. Without any filters on the other hand, contrast feature gives the best output in
Keywords: distinguishing between AD and NC groups with the normalized value of 0.1. The result
Alzheimer’s Disease; MRI; image from this study may help in thorough investigation of other features or hybrid features
denoising; feature extraction; GLCM that could be used for the purpose of detection and classification of AD.
1. Introduction
Alzheimer’s disease (AD) is a neurological disease affecting the brain function that may lead to
memory loss in patients and could affect the daily life of both patients and caretakers [1]. This disease
has no cure and the patients’ condition may worsen due to the progression of the disease, which may
in the end lead to death. Researchers all over the world try to come up with a simple and more
accurate approach in order to identify AD before the symptoms become visible but most often, AD
*
Corresponding author.
E-mail address:
https://doi.org/10.37934/araset.31.2.234244
234
, Journal of Advanced Research in Applied Sciences and Engineering Technology
Volume 31, Issue 2 (2023) 234-244
detection is not that accurate. Usually, accuracy will be improved once the patient has started to
show the signs of the disease. Therefore, detection of AD at its earliest stage is essential so that
proper treatment could be introduced thus slowing down or even preventing the patients from
further brain damage [2].
Over the past few decades, magnetic resonance imaging (MRI) has been progressively used in
exploring brain anatomy including for AD [3]. In the diagnosis of AD using medical imaging modalities,
specifically using MRI images, some lesion areas in the brain could be determined with the help of
expert radiographers. Image processing technology in addition can be used to for a better quality of
reconstruction and measurement process, especially for brain, soft tissues and lesions. With the help
of sophisticated and high-end computers, medical experts can analyze multiple areas of interest
qualitatively and quantitatively [4]. Current practice on detecting AD is by looking into the brain
scans, performing a clinical assessment, and eventually asking questions of the patient and their
relatives [5,6]. This is a very challenging process as identifying the parts of the brain that are affected
by AD by using the bare eyes is not an easy task. Moreover, one of the AD symptoms, such as brain
shrinkage, can be observed as well in healthy, elderly normal control (NC) groups [7].
Therefore, computer-aided diagnosis (CAD) systems which utilize computerized and digital image
analysis can be used as an alternative solution to the problem. CAD can be developed based on image
analysis techniques using deep learning methods or common traditional methods [8]. Utilization of
deep learning frameworks for detection of AD in MRI images can be seen as in previous studies
[2,4,8]. Even though the deep learning methods can give great output when dealing with brain MRI
analysis, finding a robust and more generic algorithm still remain as a challenge. Pre-processing stage
can also affect the performance of deep learning techniques [3]. On the other hand, the common
framework which uses the pipeline of pre-processing, segmentation, feature extraction and
classification can be utilized using supervised or unsupervised machine learning approaches such as
Support Vector Machine (SVM) [9,10]. Detecting AD using these algorithms has their own issues as
well including the low image quality, issues in the brain segmentation and pre-processing steps, and
the complexity of medical images [11]. Thus, detailed analysis is needed to thoroughly investigate
each step involved in the pipeline.
Image pre-processing is usually performed before any further image analysis to remove any
possible noise or unwanted data, as well as to highlight and enhance important features for feature
extraction process. Due to the complexity of medical images construction, the images may contain
noise and distortion which may generally cause by the variations in the detector sensitivity,
diminished illumination of the object, limitation of the images, and spontaneous variations in the
radiation signal [12,13]. Therefore, it is crucial to pre-process the data to enhance its quality or to
optimize its geometric and intensity patterns [12]. The pre-processing step will help the researchers
to improve the quality which will then highlight the essential information that is required in the
feature extraction and classification process. There are many available image pre-processing steps
including intensity normalization, contrast enhancement, image denoising, brain extraction, skull
stripping, and others. Some researchers performed the pre-processing steps using secondary
tools/software such as the Brain Extractor Tools (BET2), Freesurfer Statistical Parameter Mapping
(SPM12), and Computer Anatomy Toolbox (CAT12) [2,14-17]. The implementation of image denoising
step is also varies between researchers where there were previous reported studies which uses the
more common filters such as median filter, Wiener filter, histogram equalization, and Gaussian low
pass filter [9,10,18,19]. There was also a study that implemented other types of filters such as Lucy-
Richardson approach [20,21]. Besides, there were some past studies which did not even implement
any image denoising process [8,22-24]. Due to these stated inconsistencies in the process, the effect
235
