Chastine Fatichah

A commixed modified Gram-Schmidt and region growing mechanism for white blood cell image segmentation

A modified Gram-Schmidt orthogonalisation method has been commixed with a region growing method for efficient white blood cell image segmentation. The modified Gram-Schmidt method is used to segment the nucleus of white blood cells, while the region growing method is employed to segment the cytoplasm of white blood cells. To evaluate the performance of WBC image segmentation, the 100 samples of the microscopic WBC images is used. The segmentation results from the proposed mechanism are compared with manually segmented images, which are considered to be the correct segmentation result. Accuracy of the results reaches 95.21% and 92.31% for the Lymphocyte and Neutrophil cell types for the modified Gram-Schmidt and region growing methods, respectively. The results suggest that the proposed mechanism could be used for WBC classification in other applications such as cancer diagnosis.

A Combined AdaBoost and NEWFM Technique for Medical Data Classification

A hybrid technique combining the AdaBoost ensemble method with the neural network with fuzzy membership function (NEWFM) method is proposed for medical data classification and disease diagnosis. Combining the Adaboost, a general method used to improve the performance of learning methods, with the ‘standard’ NEWFM, which uses as base classifiers, ensures better accuracy in medical data classification tasks and diagnosis of diseases. To validate the proposal, four medical datasets related to epileptic seizure detection, Parkinson, cardiovascular (heart), and hepatitis disease diagnoses were used. The results show an average classification accuracy of 95.8% (made up of best accuracy of 99.5% for epileptic seizure, 87.9% for Parkinson, 97.4% for cardiovascular (heart) disease, and 98.7% for Hepatitis dataset classifications), which suggests that the proposed technique is capable of efficient medical data classification and potential applications in disease diagnosis and treatment.

Nuclei segmentation of microscopic breast cancer image using Gram-Schmidt and cluster validation algorithm

A combination of Gram-Schmidt method and cluster validation algorithm based Bayesian is proposed for nuclei segmentation on microscopic breast cancer image. Gram-Schmidt is applied to identify the cell nuclei on a microscopic breast cancer image and the cluster validation algorithm based Bayesian method is used for separating the touching nuclei. The microscopic image of the breast cancer cells are used as dataset. The segmented cell nuclei results on microscopic breast cancer images using Gram-Schmidt method shows that the most of MSE values are below 0.1 and the average MSE of segmented cell nuclei results is 0.08. The average accuracy of separated cell nuclei counting using cluster validation algorithm is 73% compares with the manual counting.

Evidence Accumulation Clustering with Possibilitic Fuzzy C-Means base clustering approach to disease diagnosis

Traditionally, supervised machine learning methods are the first choice for tasks involving classification of data. This study provides a non-conventional hybrid alternative technique (pEAC) that blends the Possibilistic Fuzzy C- Means (PFCM) as base cluster generating algorithm into the ‘standard’ Evidence Accumulation Clustering (EAC) clustering method. The PFCM coalesces the separate properties of the Possibilistic C-Means (PCM) and Fuzzy C-Means (FCM) algorithms into a sophisticated clustering algorithm. Notwithstanding the tremendous capabilities offered by this hybrid technique, in terms of structure, it resembles the hEAC and fEAC ensemble clustering techniques that are realised by integrating the K-Means and FCM clustering algorithms into the EAC technique. To validate the new techniques effectiveness, its performance on both synthetic and real medical datasets was evaluated alongside individual runs of well-known clustering methods, other unsupervised ensemble clustering techniques and some supervised machine learning methods. Our results show that the proposed pEAC technique outperformed the individual runs of the clustering methods and other unsupervised ensemble techniques in terms accuracy for the diagnosis of hepatitis, cardiovascular, breast cancer, and diabetes ailments that were used in the experiments. Remarkably, compared alongside selected supervised machine learning classification models, our proposed pEAC ensemble technique exhibits better diagnosing accuracy for the two breast cancer datasets that were used, which suggests that even at the cost of none labelling of data, the proposed technique offers efficient medical data classification.

Spline and color representation for batik design modification

A computer based system for batik design modification using spline and color representation is proposed to create a new batik design from the existing batik design image. The system can be used to support artisan/designer works for drawing motifs and applying color composition, so the design can be seen visually on the screen without having to make a real design. To evaluate the performance of the system, five batik images are used as the input of the system. The experimental results show that the system generates the similar batik design using spline and color representation from the existing batik design image and the user can edit the spline and its color to create new batik design.

Pencarian Question-Answer Menggunakan Convolutional Neural Network Pada Topik Agama Berbahasa Indonesia

Community-based question answering (CQA) is formed to help people who search information that they need through a community. One condition that may occurs in CQA is when people cannot obtain the information that they need, thus they will post a new question. This condition can cause CQA archive increased because of duplicated questions. Therefore, it becomes important problems to find semantically similar questions from CQA archive towards a new question. In this study, we use convolutional neural network methods for semantic modeling of sentence to obtain words that they represent the content of documents and new question. The result for the process of finding the same question semantically to a new question (query) from the question-answer documents archive using the convolutional neural network method, obtained the mean average precision value is 0,422. Whereas by using vector space model, as a comparison, obtained mean average precision value is 0,282. Index Terms—community-based question answering, convolutional neural network, question retrieval

Determining the number of clusters for nuclei segmentation in breast cancer image

Clustering is commonly technique for image segmentation, however determining an appropriate number of clusters is still challenging. Due to nuclei variation of size and shape in breast cancer image, an automatic determining number of clusters for segmenting the nuclei breast cancer is proposed. The phase of nuclei segmentation in breast cancer image are nuclei detection, touched nuclei detection, and touched nuclei separation. We use the Gram-Schmidt for nuclei cell detection, the geometry feature for touched nuclei detection, and combining of watershed and spatial k-Means clustering for separating the touched nuclei in breast cancer image. The spatial k-Means clustering is employed for separating the touched nuclei, however automatically determine the number of clusters is difficult due to the variation of size and shape of single cell breast cancer. To overcome this problem, first we apply watershed algorithm to separate the touched nuclei and then we calculate the distance among centroids in order to solve the over-segmentation. We merge two centroids that have the distance below threshold. And the new of number centroid as input to segment the nuclei cell using spatial k- Means algorithm. Experiment show that, the proposed scheme can improve the accuracy of nuclei cell counting.

Alcoholism classification based on EEG data using Independent Component Analysis (ICA), Wavelet de-noising and Probabilistic Neural Network (PNN)

Alcoholism is a clinical symptom characterized by a tendency to drink more alcohol than planned or commonly called alcoholics. Alcoholics will suffer the damage in some parts of the body, including the brain. One way to detect alcoholics from the brain is to record the electrical activity of the brain through the scalp or called electroencephalography (EEG). EEG records are often disturbed by noise such as muscle movements, eye blinking and heartbeat. Therefore, this research suggests Independent Component Analysis (ICA), as noise removal, Stationary Wavelet Transform (SWT) as a feature extraction method and are classified into two classes, namely alcoholism and normal using Probabilistic Neural Network (PNN). In this research, the result obtained from the ICA noise removal, signal decomposition using Daubechies SWT at level 6 and Probabilistic Neural Network (PNN) is considered effective to extract features and classify the 64 channels alcoholism data. The data come from Neurodynamics Laboratory, State University of New York Health Center. The result of this research generate an accuracy of 85.00% from 100 random data trial using ICA, SWT decomposition level 6, Wavelet Daubechies type 4 and PNN deviation value of 0.6.

CBE: Corpus-based of emotion for emotion detection in text document

Emotion Detection is a part of Natural Language Processing (NLP) that still evolve. Emotional Corpus that had been widely used are Wordnet Affect Emotion (WNA) and ANEW (Affective Norms for English Words). There are two ways to analyze the text based Emotion Detection: Categorical and Dimensional Model. Each model has different advantages and disadvantages. And each model has a different concept to predict emotion. The contribution of this research is forming automatic emotional corpus with merging two computational model. It called Corpus-Based of Emotion (CBE). CBE developed from ANEW and WNA with term similarity measure and distance of node approach. Latent Dirichlet Allocation (LDA) is used too for automatically expand CBE. The CBE attributes are a score of Valence (V), Arousal (A), Dominance (D) and categorical label emotion. Categorical label emotion based on six basic emotion of Ekman. Based on experiment results, it is known that CBE is able to improve the accuracy in detection of emotions. F-Measure using WNA+ANEW is 0.50 and F-Measure using CBE with expanding is 0.61.

Feature extraction using statistical moments of wavelet transform for iris recognition

Iris is unique for each person, so that it can be used as one alternative solution for human identification. In this study, an iris recognition system is developed to automatically identify a person by using eye image data. Firstly, iris area of eye image is detected using Canny Edge Detection and Hough Transform methods. Secondly, texture feature of iris image is extracted using statistical moments of Wavelet Transform. Furthermore, the texture feature representation is recognized using Support Vector Machine classifier method. Experiment on CASIA eye image dataset gives good recognition rate, that is 93.5%.