Wan Azani Wan Mustafa
Thumbnail Image
Preferred name
Wan Azani Wan Mustafa
Official Name
Wan Azani, Wan Mustafa
Alternative Name
Mustafa, W.
Azani Mustafa, Wan
Mustaffa, Wan Azani
Wan Mustafa, Wan Azani
Main Affiliation
Scopus Author ID
57219421621
Researcher ID
J-4603-2014

Research Output

Filters

37
12
40
Reset filters

Settings
Now showing 1 - 10 of 40
Thumbnail Image
Publication

Automated Detection of Corneal Ulcer Using Combination Image Processing and Deep Learning

2022-12-01 , Qasmieh I.A. , Alquran H. , Zyout A. , Al-Issa Y. , Wan Azani Wan Mustafa , Alsalatie M.

A corneal ulcers are one of the most common eye diseases. They come from various infections, such as bacteria, viruses, or parasites. They may lead to ocular morbidity and visual disability. Therefore, early detection can reduce the probability of reaching the visually impaired. One of the most common techniques exploited for corneal ulcer screening is slit-lamp images. This paper proposes two highly accurate automated systems to localize the corneal ulcer region. The designed approaches are image processing techniques with Hough transform and deep learning approaches. The two methods are validated and tested on the publicly available SUSTech-SYSU database. The accuracy is evaluated and compared between both systems. Both systems achieve an accuracy of more than 90%. However, the deep learning approach is more accurate than the traditional image processing techniques. It reaches 98.9% accuracy and Dice similarity 99.3%. However, the first method does not require parameters to optimize an explicit training model. The two approaches can perform well in the medical field. Moreover, the first model has more leverage than the deep learning model because the last one needs a large training dataset to build reliable software in clinics. Both proposed methods help physicians in corneal ulcer level assessment and improve treatment efficiency.

View
Thumbnail Image
Publication

EOG Based Eye Movements and Blinks Classification Using Irisgram and CNN-SVM Classifier

2023-01-01 , Zyout A. , Alquraan O. , Alsalatie M. , Alquran H. , Alqudah A.M. , Wan Azani Wan Mustafa , Mohammed F.F. , Alkhayyat A.

The classification of eye movements and blinks is an important task in various fields, including ophthalmology, psychology, and human-computer interaction. In recent years, the use of EOG signals and convolutional neural networks (CNNs) has shown promising results in accurately classifying different types of eye movements and blinks. The Irisgram, which is a two-dimensional representation of the short-time Fourier transform in the shape of a human iris, has been used as a feature for distinguishing between different types of eye movements and blinks. Additionally, CNNs have been utilized to learn the features automatically from Irisgrams and classify the eye movements and blinks based on these learned features. In this paper, we provide a methodology to classify blinks and four eye movements by employing Irisgram as input to the CNN-SVM classifier which achieved test accuracy of 96.2% in the testing dataset.

View
Thumbnail Image
Publication

Enhancement Cervical Whole Slice Images Using Histogram Techniques

2024-05-10 , Khreast S. , Al Quraan O. , Wan Azani Wan Mustafa , Badarneh A. , Alquran H.

Cervical cancer is a major cause of mortality among women worldwide, and early detection is crucial for successful treatment. However, the interpretation of cervical whole slice images can be challenging due to poor image quality. This paper presents a study on the use of histogram techniques to enhance the quality of cervical whole slice images. The aim of the study is to improve the visibility of important structures in the image, such as blood vessels and cell nuclei, for more accurate diagnosis and treatment of cervical cancer. The study used histogram equalization and stretching techniques to enhance the contrast and brightness of cervical whole slice images. Experiments were conducted to test the effectiveness of these techniques in improving the image quality. The results show that the enhanced images are of higher quality and are easier to interpret than the original images. The histogram equalization technique improved the visibility of structures in the image by increasing the contrast, while the histogram stretching technique improved the brightness and color balance. In conclusion, this study demonstrates the effectiveness of histogram techniques in enhancing the quality of cervical whole slice images for better diagnosis and treatment of cervical cancer. The use of these techniques can greatly improve the visibility of important structures in the image and lead to more accurate diagnosis and treatment. These findings can have important implications for the development of more effective screening and diagnostic methods for cervical cancer.

View
Thumbnail Image
Publication

Pap Smear Image Analysis Based on Nucleus Segmentation and Deep Learning – A Recent Review

2023-02-01 , Alias N.A. , Wan Azani Wan Mustafa , Mohd Aminudin Jamlos , Ismail S. , Alquran H. , Mohamad Nur Khairul Hafizi Rohani

Cervical cancer refers to a dangerous and common illness that impacts women worldwide. Moreover, this cancer affects over 300,000 people each year, with one woman diagnosed every minute. It affects over 0.5 million women annually, leading to over 0.3 million deaths. Recently, considerable literature has grown around developing technologies to detect cervical cancer cells in women. Previously, a cervical cancer diagnosis was made manually, which may result in a false positive or negative. Automated detection of cervical cancer and analysis method of the Papanicolaou (Pap) smear images are still debated among researchers. Thus, this paper reviewed several studies related to the detection method of Pap smear images focusing on Nuclei Segmentation and Deep Learning (DL) from the publication year of 2020, 2021, and 2022. Training, validation, and testing stages have all been the subject of study. However, there are still inadequacies in the current methodologies that have caused limitations to the proposed approaches by researchers. This study may inspire other researchers to view the proposed methods' potential and provide a decent foundation for developing and implementing new solutions.

View
Thumbnail Image
Publication

Cervical Cancer Detection Techniques: A Chronological Review

2023-05-01 , Wan Azani Wan Mustafa , Ismail S. , Mokhtar F.S. , Alquran H. , Al-Issa Y.

Cervical cancer is known as a major health problem globally, with high mortality as well as incidence rates. Over the years, there have been significant advancements in cervical cancer detection techniques, leading to improved accuracy, sensitivity, and specificity. This article provides a chronological review of cervical cancer detection techniques, from the traditional Pap smear test to the latest computer-aided detection (CAD) systems. The traditional method for cervical cancer screening is the Pap smear test. It consists of examining cervical cells under a microscope for abnormalities. However, this method is subjective and may miss precancerous lesions, leading to false negatives and a delayed diagnosis. Therefore, a growing interest has been in shown developing CAD methods to enhance cervical cancer screening. However, the effectiveness and reliability of CAD systems are still being evaluated. A systematic review of the literature was performed using the Scopus database to identify relevant studies on cervical cancer detection techniques published between 1996 and 2022. The search terms used included “(cervix OR cervical) AND (cancer OR tumor) AND (detect* OR diagnosis)”. Studies were included if they reported on the development or evaluation of cervical cancer detection techniques, including traditional methods and CAD systems. The results of the review showed that CAD technology for cervical cancer detection has come a long way since it was introduced in the 1990s. Early CAD systems utilized image processing and pattern recognition techniques to analyze digital images of cervical cells, with limited success due to low sensitivity and specificity. In the early 2000s, machine learning (ML) algorithms were introduced to the CAD field for cervical cancer detection, allowing for more accurate and automated analysis of digital images of cervical cells. ML-based CAD systems have shown promise in several studies, with improved sensitivity and specificity reported compared to traditional screening methods. In summary, this chronological review of cervical cancer detection techniques highlights the significant advancements made in this field over the past few decades. ML-based CAD systems have shown promise for improving the accuracy and sensitivity of cervical cancer detection. The Hybrid Intelligent System for Cervical Cancer Diagnosis (HISCCD) and the Automated Cervical Screening System (ACSS) are two of the most promising CAD systems. Still, deeper validation and research are required before being broadly accepted. Continued innovation and collaboration in this field may help enhance cervical cancer detection as well as ultimately reduce the disease’s burden on women worldwide.

View
Thumbnail Image
Publication

Automated Classification of Skin Lesions Using Different Classifiers

2023-01-01 , Al-Tawalbeh J. , Alshargawi B. , Al-Daraghmeh M. , Alquran H. , Wan Azani Wan Mustafa , Al-Dolaimy F. , Alkhayyat A.

Human skin cancer is the most common death. Skin cancer is defined as the abnormal growth of skin cells that most commonly occurs in areas of the body that are exposed to sunlight, but it can occur anywhere on the body. In their early stages, the majority of skin cancers are curable. As a result, detecting skin cancer early and quickly can save a patient's life. The incidence of malignant melanoma, the most dangerous type of skin cancer, rises year after year. Detecting skin cancer from a skin lesion is difficult due to artifacts, low contrast, mole, scar, etc. Due to the new technological advancements, early detection of skin cancer is now possible. This paper uses K-nearest neigbour (KNN), Artificial neural network (ANN) and support vector machine (SVM) classifiers for segmented and non-segmented groups and shows 95.8% overall accuracy for all classes, with the sensitivity of 97%, 91.4% and 99.7% for Benign, melanoma, seborrheic keratosis, respectively as well a precision of 92.4%, 96.6% and 99.7%, respectively. With all automatically extracted features, the accuracy is better in a non-segmented case. This paper could be extended and further processed to meet an everyday demand of how the lesions are classified or if there are any cancers.

View
Thumbnail Image
Publication

Pneumonianet: Automated detection and classification of pediatric pneumonia using chest x-ray images and cnn approach

2021-12-01 , Alsharif R. , Al-Issa Y. , Alqudah A.M. , Qasmieh I.A. , Wan Azani Wan Mustafa , Alquran H.

Pneumonia is an inflammation of the lung parenchyma that is caused by a variety of infectious microorganisms and non-infective agents. All age groups can be affected; however, in most cases, fragile groups are more susceptible than others. Radiological images such as Chest X-ray (CXR) images provide early detection and prompt action, where typical CXR for such a disease is characterized by radiopaque appearance or seemingly solid segment at the affected parts of the lung due to inflammatory exudate formation replacing the air in the alveoli. The early and accurate detection of pneumonia is crucial to avoid fatal ramifications, particularly in children and seniors. In this paper, we propose a novel 50 layers Convolutional Neural Network (CNN)-based architecture that outperforms the state-of-the-art models. The suggested framework is trained using 5852 CXR images and statistically tested using five-fold cross-validation. The model can distinguish between three classes: viz viral, bacterial, and normal; with 99.7% ± 0.2 accuracy, 99.74% ± 0.1 sensitivity, and 0.9812 Area Under the Curve (AUC). The results are promising, and the new architecture can be used to recognize pneumonia early with cost-effectiveness and high accuracy, especially in remote areas that lack proper access to expert radiologists, and therefore, reduces pneumonia-caused mortality rates.

View
Thumbnail Image
Publication

Detection of Polycystic Ovary Syndrome (PCOS) Using Machine Learning Algorithms

2022-01-01 , Hdaib D. , Almajali N. , Alquran H. , Wan Azani Wan Mustafa , Al-Azzawi W. , Alkhayyat A.

One of the most common diseases in women of reproductive age is Polycystic Ovary Syndrome (PCOS). PCOS diagnosis can be tricky, because not everyone with PCOS has polycystic ovaries (PCO), nor does everyone with ovarian cysts have PCOS, hence the pelvic ultrasound as a stand-alone diagnosis is not sufficient. The full diagnostic plan is mainly a combination of a pelvic ultrasound besides blood tests of specific parameters that indicate the presence of PCOS. Since PCOS is a hard-to-diagnose widespread hormonal disorder, blood tests, symptoms, and other parameters with the help of a computer can form a new and easy method to diagnose it. Therefore, we had successfully built a high performing diagnostic model using MATLAB. The data was obtained from the website Kaggle, and the dataset is called Polycystic Ovary Syndrome. In this paper various machine algorithms were employed by utilizing seven classifiers. Results demonstrated that Linear Discriminant classifier exhibits the best performance in terms of accuracy, while in terms of sensitivity, the KNN classifier had the best result. Also, a comparison with four other research papers that exploited the same PCOS dataset was done in terms of implementation platforms, evaluation methods, classifiers, classes, accuracy, and precision of each classifier. Our research excelled among all in terms of accuracy and varied in precedence with precision. MATLAB had shown substantial results and a great model fitting embedded approaches, scoring a high accuracy and precision outcome compared to other studies. Other improvements on the overall PCOS prediction can involve employing preprocessed ultrasound images with the features presented in the dataset.

View
Thumbnail Image
Publication

Counting Non-Overlapping Abnormal Cervical Cells in Whole Slide Images

2023-01-01 , Badarneh A. , Alzuet A. , Wan Azani Wan Mustafa , Alquran H. , Alsalatie M. , Mohammed F.F. , Alkhayyat A.

Cervical cancer is one of the most common cancer among women globally. The Pap smear test has been widely used to detect cervical cancers according to the morphological characteristics of the cell nuclei on the micrograph. The aim of this paper is to count the non-overlapping abnormal cervical cells in whole slide images automatically by employing various image techniques. The proposed approach consists of four main steps; image enhancement, transform the extended minima, remove small pixels, and count the number of abnormal cells in the image. The proposed system used 250 cervical pap smear images where the overlap between cells is minimal. The performance of the proposed system is evaluated based on comparing the manual counting and automating counting over whole images. Therefore, the accuracy is evaluated mainly on the difference between manual and automated, and it is 92.5%. The proposed method can be used in laboratory to decrease the false positive rates in counting abnormal cells.

View
Thumbnail Image
Publication

Nucleus Detection using Bradley Algorithm Modification on Pap Smear Cell Image

2024-05-10 , Wan Azani Wan Mustafa , Arifin Z. , Azir K.N.F.K. , Yaakob N. , Syed Zulkarnain Syed Idrus , Alquran H.

According to a 2010 study, cervical cancer affects women all over the world and is the fourth most common disease among women after lung, colorectal, and breast cancer. Numerous studies are being conducted to improve pre-cancer screening methods. It is challenging for doctors to collect correct data due to inconsistent and unsystematic analysis techniques. Making a determination just from eye inspection is challenging. It is probably erroneous based on the sample of microscopic images studied. This is because to the blurring, noise, shadow, lighting, and artefact issues that can damage a microscopic image. This study has modified Bradley’s binarization method in order to improvise nucleus detection on pap smear image. The modification method has performance that overcome other methods which is Bernsen, Bradley, Feng, Niblack, Nick, Otsu, Sauvola and Wolf. The analysis indicate modification method has value 75.03% on Fmeasure, 92.78% on accuracy, 13.61 on PSNR and 0.086 on NRM. This study helps the modification method inspired by Bradley shows an improvement in segmenting nucleus region.

View