Browsing by Author "Naziullah, Shekh"

Now showing 1 - 4 of 4

A Hybrid Environment Control System Combining EMG and SSVEP Signal Based on Brain-computer Interface Technology
(SN Applied Sciences, 2021-08-23) Rashid, Mamunur; Bari, Bifta Sama; Sulaiman, Norizam; Mustafa, Mahfuzah; Hasan, Md Jahid; Islam, Md Nahidul; Naziullah, Shekh
The patients who are impaired with neurodegenerative disorders cannot command their muscles through the neural pathways. These patients are given an alternative from their neural path through Brain-Computer Interface (BCI) systems, which are the explicit use of brain impulses without any need for a computer's vocal muscle. Nowadays, the steady-state visual evoked potential (SSVEP) modality offers a robust communication pathway to introduce a non-invasive BCI. There are some crucial constituents, including window length of SSVEP response, the number of electrodes in the acquisition device and system accuracy, which are the critical performance components in any BCI system based on SSVEP signal. In this study, a real-time hybrid BCI system consists of SSVEP and EMG has been proposed for the environmental control system. The feature in terms of the common spatial pattern (CSP) has been extracted from four classes of SSVEP response, and extracted feature has been classified using K-nearest neighbors (k-NN) based classification algorithm. The obtained classification accuracy of eight participants was 97.41%. Finally, a control mechanism that aims to apply for the environmental control system has also been developed. The proposed system can identify 18 commands (i.e., 16 control commands using SSVEP and two commands using EMG). This result represents very encouraging performance to handle real-time SSVEP based BCI system consists of a small number of electrodes. The proposed framework can offer a convenient user interface and a reliable control method for realistic BCI technology.
A Novel Vision Transformer Approach for Driver Fatigue Detection from EEG Signal with Correlation-Based Channel Selection
(Scopus, 2024-12-19) Khan, Robiul; Roy, Dipon; Ali, Md. Wajed; Refat, Kawsar Ahmed; Naziullah, Shekh
Driver fatigue poses a critical threat to road safety, necessitating the development of robust detection methods to minimize traffic accidents and societal burdens. Deep neural networks have recently been effectively applied to Electroencephalography (EEG)-based driving fatigue detection. Nevertheless, most of the existing models, particularly those relying on extensive pooling, often struggle to capture long range dependencies within images. To address this issue, we propose a Correlation-based Channel Selection (CCS) with a Vision Transformer (ViT) approach for driver fatigue detection using an EEG. Our methodology integrates a pioneering Channel Selection (CS) block to extract discriminative channels via CCS. This mechanism systematically identifies the most informative EEG channels crucial for fatigue detection. Subsequently, we leverage Continuous Wavelet Transform (CWT) to convert the selected EEG channels into a time-frequency spectral image. Finally, the resulting time-frequency spectral images, encompassing both temporal and spectral information, are concatenated and fed into the Vision Transformer (ViT) model to classify them as either normal or fatigued. The proposed model is evaluated on a publicly available EEG dataset containing recordings from twelve subjects. The model achieved superior accuracy: 95.83% for the effective selected combined subject and 99.925% for the average accuracy of each subject, demonstrating its potential for robust driver fatigue detection.
Classification of EEG-Based Auditory Evoked Potentials Using Entropy-Based Features and Machine Learning Techniques
(IEEE, 2023-11-06) Islam, Thamina; Ahmed, Firoz; Ahmed, Nayem; Naziullah, Shekh; Islam, Md Nahidul; Rashid, Mamunur
Hearing loss is a prevalent impairment that disrupts interactions with others and individuals' learning abilities. Immediate and accurate diagnosis of hearing loss using Electroencephalogram (EEG) signals, particularly Auditory Evoked Potentials (AEP), is considered the most effective approach to address this issue. The AEP signals, generated in the cerebral cortex in response to auditory stimuli, serve as the most reliable method for diagnosing deafness. This study introduces a novel approach for detecting hearing ability through the classification of EEG-AEP signals. The current experiment makes use of a publicly available dataset that contains AEP responses from 16 people who responded to auditory stimuli on either the left or right side. Sample Entropy is employed to extract the feature, capturing the complex temporal dynamics of the EEG signals. Four popular machine learning-based classifiers, namely Support Vector Machines (SVM), K-Nearest Neighbors (KNN), Random Forest (RF), and Logistic Regression (LR), are utilized for classification purposes. The results indicate that SVM achieves the highest classification accuracy of 99.37% with subject-4 and the average accuracy of 90.74% is achieved with all subjects. This finding shows the effectiveness of Sample Entropy as a feature extraction technique for characterizing AEPs and highlights the potential of SVM as a robust classifier for the accurate identification of auditory stimuli localization. The accuracy achieved in this study indicates a promising direction for the development of reliable and non-invasive methods for hearing-related diagnoses.
Lung Cancer Detection using Deep Learning with Hybrid Preprocessing Pipeline
(Daffodil International University, 2025-01-12) Naziullah, Shekh; Mosfiq, M.Mukit
Lung cancer remains one of the leading causes of death globally, with millions of lives lost each year. It is one of the most prevalent non-communicable diseases, responsible for approximately 6% of all deaths. Symptoms of lung cancer are diverse and may include chest and bone pain, wheezing, persistent coughs, unexplained weight loss, fatigue, shortness of breath, and hemoptysis, among others. Risk factors include long-term smoking, exposure to secondhand smoke, asbestos, radon gas, radiation therapy to the chest, and a family history of lung cancer. While CT scans are commonly used for detection, they have limitations, particularly in early-stage diagnosis, due to high false positive rates, and can be uncomfortable for patients. An alternative approach, incorporating machine learning and deep learning, offers the potential for improved early detection, increasing survival rates and reducing unnecessary follow-up tests and treatments. This study focuses on detecting lung cancer using CT scan images, applying a multiclass classification system to differentiate between malignant, benign and normal images. The proposed system has been designed for use in hospitals to aid in the diagnosis and treatment of lung cancer. To obtain accurate results after applying our model, we use an online dataset, for which high and low quality CT images are presented here for this online dataset. In this study, to process the quality CT images we use a hybrid preprocessing pipeline where we use stratified sampling and SMOTE oversampling method to do sampling and to increase image quality we use gaussian blur method. Experimental results show that while not all models achieved high accuracy, most surpassed 96%. Notably, models such as CNN (96%), ResNet50 (91%) and VGG16 (88%) demonstrated superior performance in accurately identifying lung cancer in CT scans.