Faculty of Computing

Permanent URI for this communityhttps://repository.nileuniversity.edu.ng/handle/123456789/13

Browse

Author: search.filters.author.Steve AdeshinaStart date: 2020

Search Results

Now showing 1 - 7 of 7
  • No Thumbnail Available
    Item
    Hepatitis C Stage Classification with hybridization of GA and Chi2 Feature Selection
    (IJCSNS International Journal of Computer Science and Network Security, 2022-02-02) Steve Adeshina; Moussa Mahamat Boukar
    In metaheuristic algorithms such as Genetic Algorithm (GA), initial population has a significant impact as it affects the time such algorithm takes to obtain an optimal solution to the given problem. In addition, it may influence the quality of the solution obtained. In the machine learning field, feature selection is an important process to attaining a good performance model; Genetic algorithm has been utilized for this purpose by scientists. However, the characteristics of Genetic algorithm, namely random initial population generation from a vector of feature elements, may influence solution and execution time. In this paper, the use of a statistical algorithm has been introduced (Chi2) for feature relevant checks where p-values of conditional independence were considered. Features with low p-values were discarded and subject relevant subset of features to Genetic Algorithm. This is to gain a level of certainty of the fitness of features randomly selected. An ensembled-based learning model for Hepatitis has been developed for Hepatitis C stage classification. 1385 samples were used using Egyptian-dataset obtained from UCI repository. The comparative evaluation confirms decreased in execution time and an increase in model performance accuracy from 56% to 63%.
  • No Thumbnail Available
    Item
    A Few-shot custom CNN Model for Retinal Nerve Fibre Layer Thickness Measurement in OCT Images of Epilepsy
    (Proc. of International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (, 2024-02-01) Ruqayya Muhammad; Moussa Mahamat Boukar; Steve Adeshina; Senol Dane
    This study aims to assess the effectiveness of employing deep learning models for measuring retinal nerve fiber layer (RNFL) thickness in optical coherence tomography (OCT) scans of epilepsy patients. Conventional OCT scan segmentation methods typically rely on supervised learning, demanding substantial data for training and assuming fixed network weights post-training. To mitigate these challenges, we explore the applicability of few-shot learning (FSL) in CNN architectures, allowing dynamic fine-tuning of network weights with minimal additional data. Experimental results demonstrate enhanced segmentation accuracy, with the proposed Few shot Custom CNN achieving a notable 91% accuracy, surpassing both the Custom CNN (86%) and the OCT machine data. This suggests the superiority of the few-shot Custom CNN model in segmentation performance compared to OCT scans.
  • No Thumbnail Available
    Item
    Analysis of Prostate Cancer DNA Sequences Using Bi-direction Long Short Term Memory Model
    (IEEE, 2021-02-02) Yusuf Aleshinloye Abass; Steve Adeshina; Nwojo Nnana Agwu; Moussa Mahamat Boukar
    Machine and deep learning-based models are the emerging techniques in addressing prediction problems in biomedical data analysis. DNA sequence prediction is a critical problem that requires huge attention in the biomedical domain. These techniques have been shown to provide better accurate results when compared to the conventional regression-based models. Prediction of the gene sequence that leads to cancerous diseases such as prostate cancer is very crucial. Identifying the most important features in a gene sequence is one of the most challenging tasks and extracting the components of the gene sequence that can give an insight into the kind of mutation in the gene is very important, it will lead to effective drug design and promote the new concept of personalized medicine. In this work we have extracted the exons in the various prostate gene sequence that was used in the experiment, we built a bi-LSTM model using a k-mer encoding for the DNA sequence and one- hot encoding for the class label. The bi-LSTM model was evaluated on different classification metrics. Our experimental results show that the model prediction offers a training accuracy and validation accuracy of 95 percent and 91 percent respectively.
  • No Thumbnail Available
    Item
    Comparison of Transfer Learning Model Accuracy for Osteoporosis Classification on Knee Radiograph
    (IEEE, 2022-02-02) Moussa Mahamat Boukar; Steve Adeshina
    In terms of financial costs and human suffering, osteoporosis poses a serious public health burden. Reduced bone mass, degeneration of the micro architecture of bone tissue, and an increased risk of fracture are its main skeletal symptoms. Osteoporosis is caused not just by low bone mineral density, but also by other factors such as age, weight, height, and lifestyle. Recent advancement in Artificial Intelligence (AI) has led to successful applications of expert systems that use Deep Learning techniques for osteoporosis diagnosis based on some modalities such as dental radiographs amongst others. This study uses a dataset of knee radiographs (i.e., knee-Xray images) to apply and compare three robust transfer learning model algorithms: GoogLeNet, VGG-16, and ResNet50 to classify osteoporosis. From the statistical analysis and scikit learn python analysis, the accuracy of the GoogLeNet model was 90%, the accuracy of the VGG-16 model was 87% and lastly, the accuracy of the ResNet- 50 model was 83%.
  • No Thumbnail Available
    Item
    Age Estimation from Facial Images Using Custom Convolutional Neural Network (CNN)
    (International Conference on Frontiers in Academic Research, 2023-02-23) Gilbert George; Steve Adeshina; Moussa Mahamat Boukar
    Given that aging is influenced by a variety of factors, including gender, ethnicity, environment, and others, automatic age assessment of facial images is a difficult challenge in computer vision and image analysis. Additionally, a significant amount of data and a laborious training phase are needed to estimate age from facial photos with near accuracy. In this study, we present a custom convolutional neural network-based age estimator that can almost precisely predict age from facial photos. We use the UTK facial image dataset using about 17475 images. We train the model to group the facial images into three groups which are; Child, Teenager and Adult. Compared to similar efforts, our method uses less training data while maintaining a high accuracy of 95%.
  • No Thumbnail Available
    Item
    Transfer Learning Model Training Time Comparison for Osteoporosis Classification on Knee Radiograph of RGB and Grayscale Images
    (WSEAS TRANSACTIONS on ELECTRONICS, 2022-09-13) Moussa Mahamat Boukar; Steve Adeshina; Senol Dane
    In terms of financial costs and human suffering, osteoporosis poses a serious public health burden. Reduced bone mass, degeneration of the microarchitecture of bone tissue, and an increased risk of fracture are its main skeletal symptoms. Osteoporosis is caused not just by low bone mineral density, but also by other factors such as age, weight, height, and lifestyle. Recent advancement in Artificial Intelligence (AI) has led to successful applications of expert systems that use Deep Learning techniques for osteoporosis diagnosis based on some modalities such as dental radiographs amongst others. This study uses a dataset of knee radiographs (i.e., knee-Xray images) to apply and compare the training time of two robust transfer learning model algorithms: GoogLeNet, VGG-16, and ResNet50 to classify osteoporosis. The dataset was split into two subcategories using python opencv library: Grayscale Images and Red Green Blue (RGB) images. From the scikit learn python analysis, the training time of the GoogLeNet model on grayscale images and RGB images was 42minutes and 50 minutes respectively. The VGG-16 model training time on grayscale images and RGB images was 37 minutes and 44 minutes respectively. In addition, to compare the diagnostic performance of the two models, several state-of-the-art neural networks metric was used.
  • No Thumbnail Available
    Item
    Evaluation of Parameter Fine-Tuning with Transfer Learning for Osteoporosis Classification in Knee Radiograph
    ((IJACSA) International Journal of Advanced Computer Science and Applications, 2022-02-02) Moussa Mahamat Boukar; Steve Adeshina
    Osteoporosis is a bone disease that raises the risk of fracture due to the density of the bone mineral being low and the decline of the structure of bone tissue. Among other techniques, such as Dual-Energy X-ray Absorptiometry (DXA), 2D x-ray pictures of the bone can be used to detect osteoporosis. This study aims to evaluate deep convolutional neural networks (CNNs), applied with transfer learning techniques, to categorize specific osteoporosis features in knee radiographs. For objective labeling, we obtained a selection of patient knee x-ray images. The study makes use of the Visual Geometry Group Deep (VGG-16), and VGG-16 with fine-tuning. In this work, the deployed CNNs were assessed using state-of-the-art metrics such as accuracy, sensitivity, and specificity. The evaluation shows that fine-tuning enhanced the VGG-16 CNN's effectiveness for detecting osteoporosis in radiographs of the knee. The accuracy of the VGG-16 with parameter fine-tuning was 88% overall, while the accuracy of the VGG-16 without parameter fine-tuning was 80%.