Siti Nurul Aqmariah Mohd Kanafiah
Thumbnail Image
Preferred name
Siti Nurul Aqmariah Mohd Kanafiah
Official Name
Siti Nurul Aqmariah, Mohd Kanafiah
Alternative Name
Kanafiah, S. N.Aqmariah
Kanafiah, S. N.A.M.
Kanafiah, Siti Nurul Aqmariah Mohd
Aqmariah Kanafiah, S. N.
Kanafiah, S. N. A. M
Main Affiliation
Scopus Author ID
55987982900
Researcher ID
HTR-1815-2023

Research Output

Filters

30
Reset filters

Settings
Now showing 1 - 10 of 30
  • Publication
    Color contrast enhancement on pap smear images using statistical analysis
    ( 2021-01-01)
    Nahrawi N.
    ;
    Wan Azani Wan Mustafa
    ;
    Siti Nurul Aqmariah Mohd Kanafiah
    ;
    Mashor M.Y.
    In the conventional cervix cancer diagnosis, the Pap smear sample images are taken by using a microscope,causing the cells to be hazy and afflicted by unwanted noise. The captured microscopic images of Pap smear may suffer from some defects such as blurring or low contrasts. These problems can hide and obscure the important cervical cell morphologies, leading to the risk of false diagnosis. The quality and contrast of the Pap smear images are the primary keys that could affect the diagnosis’ accuracy. The paper's main objective is to propose the best contrast enhancement to eliminate contrast problems in images and cor-rect them in color images to ensure smooth segmentation. In this paper, the med-ian and standard deviation are used for the image's global and local data where the problem region is normalized by using a special proposed formula. The expected resulting image shows only the object (nuclei and cytoplasm), and a background without any noise. The results were compared with CLAHE, HE, and Gray World, and the performance was evaluated based on PSNR, RMSE, and MAE. Proposed method shows higher PSNR and RMSE value while lower value for MAE compared to other methods. This paper's main impact will help doctors in identifying the patient's disease, such as cervical cancer, based on a Pap smear analysis, and increase the accuracy percentages as compared to the conventional method.
  • Publication
    Comparison between Support Vector Machine and K-Nearest Neighbor Algorithms for Leukemia Images Classification Using Shape Features
    ( 2021-01-01)
    Jusman Y.
    ;
    Hasanah A.N.
    ;
    Purwanto K.
    ;
    Siti Nurul Aqmariah Mohd Kanafiah
    ;
    Riyadi S.
    ;
    Hassan R.
    ;
    Mohamed Z.
    Leukemia occurs when the body produces abnormal white blood cells in amounts exceeding the normal limit, making them misfunctioning. It is highly influential on the human immune system. Currently, medical personnel require a long time to recognize leukemia, and it is difficult to distinguish between acute leukemia cells and normal cells. Hence, this study aims to build a system program using white blood cell images with image processing using feature extraction with the Hu moments invariant and the Support Machine Machine (SVM) and K-Nearest Neighbor (K-NN) classification methods. The samples used were 800 blood images divided into two classes, acute and normal, with each class consisting of 400 sample images. Based on the test results from comparing the average value of accuracy and training time in both methods, the highest accuracy value was in the SVM method, with an accuracy of 87.97% and the K-NN method of 83.96%. The fastest training time was in the K-NN method of 2.43 seconds and the SVM method of 3.73 seconds.
  • Publication
    Comparison of Malaria Parasite Image Segmentation Algorithm Using Thresholding and Watershed Method
    ( 2021-02-12)
    Jusman Y.
    ;
    Pusparini A.
    ;
    Nazilah Chamim A.N.
    ;
    Siti Nurul Aqmariah Mohd Kanafiah
    Malaria is an infectious disease caused by plasmodium that lives and breeds in the red blood cells, transmitted by the Anopheles mosquito. During this time, the paramedics to diagnose symptoms use any imagery that is done manually. In the identification analysis of the malaria parasite cell infection, there is a possibility of human error factor done by paramedics because of the number of samples analyzed. This case is because the human eye tends to be tired while working continuously, leading to misclassification and treatment that is not right. Therefore, it takes a computer-based system that facilitates image processing to paramedics or laboratory technicians to identify the parasite cells and reduce human error instances. This research conducted on identification of the thresholding and watershed of segmentation method for three types of plasmodium parasite, namely Plasmodium falciparum, Plasmodium malaria, and Plasmodium vivax. This study offered modifications thresholding and watershed algorithm. The results showed the success of the technique that can effectively segment on the three types of Plasmodium malaria, which has an accuracy rate above 90% as well as the results of the computation time between the thresholding method could segment imagery for 1-2 seconds and the watershed method intelligent segmented representation for 3-4 seconds.
  • Publication
    Classification of Parasite Malaria in Schizon Stage with GoogleNet and VGG-19 Pre-Trained Models
    ( 2023-01-01)
    Jusman Y.
    ;
    Aftal A.A.
    ;
    Tyassari W.
    ;
    Siti Nurul Aqmariah Mohd Kanafiah
    ;
    Hayati N.
    ;
    Mohamed Z.
    The development of artificial intelligence technology has currently given benefit for humans in various fields. In the medical field, artificial intelligence was developed to help medical experts to classify various diseases using medical images, including malaria. Early detection of malaria parasites is important to save the patients, thus this study developed a detection system for some malaria parasites (P. falciparum, P. vivax, and P. malariae) in the schizont stage. This system uses deep learning methods using GoogleNet and VGG 19 pre-trained models. This study performs accuracy, running time, and analysis based on the confusion matrix for testing result. The best training result is performed by the GoogleNet pre-trained model, with an average running time of 7 minutes 14 seconds and an average accuracy of 98.53% \pm 1.27\%. The best model for classifying malaria image in the blood is the GoogleNet model with an accuracy value of 97.41%, precision 100%, recall 93.75%, specificity 100% and f-score 99.53%.
  • Publication
    Development of life cycle classification system for Plasmodium knowlesi malaria species using deep learning
    (AIP Publishing, 2023)
    Muhd Syamir Azhar
    ;
    Mohd Yusoff Mashor
    ;
    Siti Nurul Aqmariah Mohd Kanafiah
    ;
    Zeehaida Mohamed
    In this paper, the performance of deep learning model for GoogleNet and AlexNet are analysed to classify plasmodium knowlesi life cycle stages. Plasmodium knowlesi images are taken from department of microbiology and parasitology in Hospital Universiti Sains Malaysia (HUSM) in this research work. The data images are enhanced using contrast stretching method. The enhanced image undergoes process of segmentation to extract parasite inside the effected red blood cells. The segmented images go through bounding box process according to their size input image for both deep learning models. There are 5940 data which it represents for four classes: artifact, trophozoite, schizont and gametocyte stage. These datasets are trained using GoogleNet and AlexNet to classify the life cycle stages of plasmodium knowlesi. The analysed performance of both models includes training, validation, and testing process. According to the result, both model able to reach 100% for training accuracy. For validation accuracy, AlexNet has higher accuracy with 93.4% compared to GoogleNet with 92.2%. For testing accuracy, Google has higher accuracy with 91.1% where AlexNet with 89.1%.
  • Publication
    Clasification of Malaria images in thropozoid stages using deep learning models
    (IEEE, 2024-03)
    Wikan Tyassari
    ;
    Yessi Jusman
    ;
    Novian Dwi Payana
    ;
    Siti Nurul Aqmariah Mohd Kanafiah
    ;
    Zeehaida Mohamed
    The risk of malaria infection is very high, especially for people living in eastern Indonesia, such as Papua, Maluku, and Nusa Tenggara. In Indonesia there are several types of malaria parasite infected, Plasmodium Falciparum, Plasmodium Vivax, and Plasmodium Malaria. Identifying malaria at an early stage is an important to reduce the risk of death and find suitable treatment. However, identifying and diagnosing malaria is time consuming. Therefore, it is necessary to apply technology in detecting the class of malaria parasites. This study classified images of malaria parasites Plasmodium Falciparum, Plasmodium Vivax, and Plasmodium Malarie at the trophozoite stage using the deep learning pre-trained models AlexNet and Inception-V3. According to accuracy of training, Inception-V3 is the best deep learning model. The performance analysis result of inception is accuracy 98.98% ± 0.71%, precision 98.83% ± 1.44%, recall 98.83% ± 1.38%, specificity 99.11% ± 1.09%, and F-score 98.82% ± 0.83%. However, despite having lower accuracy and performance AlexNet have faster in computational training time.
  • Publication
    Comparison of Multi Layered Percepton and Radial Basis Function Classification Performance of Lung Cancer Data
    ( 2020-03-10)
    Jusman Y.
    ;
    Indra Z.
    ;
    Salambue R.
    ;
    Siti Nurul Aqmariah Mohd Kanafiah
    ;
    Nurkholid M.A.F.
    Lung cancer was the most commonly diagnosed cancer as well as the leading cause of cancer death in males in 2008 globally. The way used to detect lung cancer are through examination chest X-ray, Computed Tomography (CT) scan, and Magnetic Resonance Imaging results. The accurate and efisien analysis of the imaging results are important to ensure the minimal time processing. A computed assisted diagnosis system is the crusial research which can conduct the analysis efficiently and efectively. This paper aimed to compare the classification performances of Multi Layered Perceptron (MLP) and Radial Basis Function (RBF) techniques. The public lung cancer datasets was used as training and testing data in the classfication techniques. Ten fold cross validation was used for dividing data before classifying techniques. The accuracy performances are compared to check a better technique for classification step.
      18  6
  • Publication
    Feature Extraction Performance to Differentiate Spinal Curvature Types using Gray Level Co-occurrence Matrix Algorithm
    ( 2020-11-24)
    Jusman Y.
    ;
    Lubis J.H.
    ;
    Chamim A.N.N.
    ;
    Siti Nurul Aqmariah Mohd Kanafiah
    Spinal curvature type can be detected from digital X-ray images. Experts diagnose spinal curvature for a long time to obtain accurate results. This research aims to analyze the use of image processing techniques to extract features in two types of spinal imagery, normal and abnormal (i.e., scoliosis), by applying the Gray Level Co-occurrence Matrix (GLCM) algorithm and Support Vector Machine (SVM) for the classification method. This study used 40 images divided into 4 data sets for analysis. Three distance parameters, 50, 75, and 100 pixels, and three parameters of quantization values, 8, 16, and 32, were utilized for analysis. The highest accuracy obtained from one of the specific data set was 100%, while the highest accuracy of the average of each value distance and quantization was 90%. The GLCM algorithm could differentiate the abnormality of spinal imagery.
      2  9
  • Publication
    Rebar Path Mapping using Ground Penetrating Radar
    ( 2023-01-01)
    Basri N.A.B.
    ;
    Hasimah Ali
    ;
    Ahmad M.R.
    ;
    Ahmad Firdaus Ahmad Zaidi
    ;
    Siti Nurul Aqmariah Mohd Kanafiah
    ;
    Jusman Y.
    Ground penetrating radar (GPR) is a non-destructive device that helps to determine the position and direction of underground utilities such as rebar while preventing any inaccurate excavation process. The direction of buried rebar is usually mapped using the X-Y grid scanning method, which requires a lot of manpower and time to complete. Therefore, this paper investigated the ability of parallel scanning of B-scan to imitate the result of C-scan produced by X-Y grid scanning. Parallel scanning has been emphasised to reduce the time consumption of the data acquisition process while delivering a quality output. To develop a rebar path mapping, a data processing step has been implemented on the B-scan data for seven parallel lines that correspond to the x-axis. Next, Kirchhoff migration has been applied along with stacking and interpolation techniques to map a two-dimensional (2-D) image of the buried rebar. The obtained result was then compared with the grid scanning data of C-scan to evaluate the correlation between them. The performance of the mapped rebar path using parallel B-scan data was evaluated based on the ability of the data to give an accurate depth calculation of the buried rebar. Ultimately, the results show that this proposed method for using parallel B-scan to do mapping is verified.
      1
  • Publication
    Recognition of different utility pipes size of ground penetrating radar images at different penetration depth
    ( 2024-02-08)
    Nasri M.I.S.
    ;
    Hasimah Ali
    ;
    Zaidi A.F.A.
    ;
    Shukor S.A.A.
    ;
    Ahmad M.R.
    ;
    Amran T.S.T.
    ;
    Siti Nurul Aqmariah Mohd Kanafiah
    ;
    Othman S.M.
    ;
    Elshaikh M.
    Ground Penetrating Radar (GPR) is a geophysical locating method that uses radio waves to capture images below the surface of the ground in a minimally invasive way. It also requires two main essential equipment which is a transmitter and a receiving antenna. To address the problem, this project proposed the hyperbolic recognition of different utility pipes of GPR images at different level of penetration depth. In this framework, the raw data of GPR images were firstly to be pre-processed. The grayscale images were cropped, resized, and enhanced to increase the contrast of the features of the image. Then, the pre-processed GPR images were extracted using the Histogram of Oriented Gradient (HOG) method with three different windows. The extracted HOG features were then used as input to the k-Nearest Neighbor classifier. A series of experiments has been conducted using 10-fold cross-validation technique for training and testing the GPR data. Based on the result obtained, it shows that at depth 20cm the average accuracy is about 99.87%, whereas at depth 40cm the average accuracy achieved 100%. Thus, the result shows that the extracted HOG features exhibit the significant information of hyperbolic signature of different pipe size with different depth of buried object. Therefore the results seem promising in recognizing the hyperbolic of utilities.
      18  4