Journal of Advancement in Electronics Signal Processing

Heart Disease Detection Using ECG Waveforms

Afeefa Askar — 2024-07-02

Cardiovascular diseases, also known as CVDs, are still the main reason for sickness and death worldwide, with around 17.5 million deaths linked to them in 2012. The Electrocardiogram (ECG) signal shows the Heart's electrical activity on the body's surface, providing critical information about heart function. It is often used to spot any irregularities in heart rhythm and structure. Over the years, many techniques have been created and researched to classify and detect abnormalities in ECG signals, showing potential for use in medical settings. Current research frequently needs to provide thorough comparisons of different heart abnormalities. Some studies focus on specific conditions, such as atrial fibrillation, while others look at ST changes. This study introduces a new method using deep convolutional neural networks to classify heartbeats and accurately detect five types of arrhythmias. Our technique involves training a Convolutional Neural Network (CNN) on a meticulously selected dataset, thoroughly validating it, and fine-tuning it with specific parameters and epochs. By feeding ECG images into the model, users can quickly determine whether the cardiac condition is normal or abnormal.

EOG Signal Analysis for Efficient Human-Computer Interaction

Pratham R. Kale — 2024-08-01

Electrooculography (EOG) is an essential technique in bio-signal processing, capturing eye movement and position through electric potentials around the eyes. This technology has gained significant attention for its application in Human-Computer Interaction (HCI), particularly for individuals with disabilities. This paper explores the methodologies of EOG signal analysis, its applications in HCI, and strategies for optimizing signal processing to enhance interaction efficiency. We review various EOG signal acquisition methods, pre-processing techniques, feature extraction, and classification algorithms. Furthermore, we discuss the challenges and future directions for EOG-based HCI systems, aiming to improve accessibility and user experience.

Brain-Computer Interfaces (BCIs) based on EOG have significantly impacted daily life, gaming, physical medicine, and aviation. These systems capture user intentions, perceptions, and motor choices by translating physiological signals into commands for external devices, enabling the execution of user-intended functions. EOG signals can recognize and classify eye movements through active or passive engagement, facilitating the control of output devices. Research in the aviation industry explores EOG-BCIs as alternatives to manual commands and as tools for streamlining user intentions.

This study reviews the last two decades of EOG-based BCI experiments, presenting current systems and inspiring future developments. We first discuss fundamental aspects of EOG-BCI research, including signal capture, device specificity, feature extraction, translation algorithms, and interface instructions. Additionally, we summarize EOG-based BCI applications in both real and virtual environments, beyond aviation. We conclude by addressing the current limitations of EOG devices and offering recommendations for future design investigations.

Classification of Fractured Bones Using Machine Learning

M. Nagaraju Naik — 2024-08-13

This study uses advanced machine learning approaches to create an effective system for classifying damaged bones. The primary method utilized is supervised learning, specifically the Random Forest algorithm. This algorithm is applied to detect and categorize bone fractures using the Musculoskeletal Radiographs (MURA) dataset, which includes various X-ray images of different human bone categories. The proposed method involves several key steps: gathering and uploading datasets, extracting relevant features, dividing the data into training and testing sets, creating and training the Random Forest model, and finally, uploading test images for bone classification. The project's primary focus is to enhance the accuracy of bone fracture identification. The project has specific hardware and software requirements and is implemented using Python and various supporting libraries. This approach aims to significantly improve the efficiency and reliability of diagnosing bone fractures, ultimately contributing to better patient outcomes in medical practice.

Development and Implementation of Drowsiness-Fatigue Detection System For Increasing Road Security

Manjunath S — 2024-08-17

Driving fatigue and drowsiness are the main elements contributing to the increase in accidents, so addressing them well regarding avenue protection is crucial. Offers a modern generation-based, totally wearable Drowsiness-Fatigue Detection (DFD) tool to address those risks. They account for a massive percentage of road deaths and injuries. The system dynamically detects driver weariness and drowsiness in real-time by utilizing wearable technology and brilliant glasses with specialized sensors. For a Deep-Cascaded Convolutional Neural Network (DCCNN) for real-time video analysis, the technique first detects facial regions before using the Dlib toolbox to extract eye landmarks. These landmarks help determine whether the drivers' eyes are open or closed by computing the Eyes Aspect Ratio (EAR). To train the system offline, each driver must receive sets of EAR data corresponding to eyes open and closed conditions.

Improvements in high-speed rail alertness detection, including a tiredness warning system that depends on tracking train drivers' attentiveness using wearable, wireless EEG-gathering equipment. The system evaluates the attention levels of high-speed train operators using EEG data. This is done through an 8-channel wireless Brain-Computer Interface (BCI) combined with Support Vector Machine (SVM) classification. The EEG Power Spectrum Density (PSD) is extracted using the Fast Fourier Transform (FFT). Advancement in road safety through wearable technology provides a non-intrusive and easily navigable means of addressing the enduring risk of driver drowsiness.

Robust Speaker Recognition using Spectrogram and CNN Against Replay Attacks

A. Revathi — 2024-08-31

Speaker recognition systems benefit various security applications, including access control and authentication. However, these systems are vulnerable to replay attacks, in which an opponent captures and replicates previously recorded speech to trick the system. This study aims to improve the robustness of speaker recognition systems against replay attacks by combining spectrogram analysis and Convolutional Neural Networks (CNN). This process begins by converting speech signals into spectrograms, representing the time-frequency representation of audio signals. Spectrograms capture essential features of the speech signal for speaker identification. Then, CNN architecture is used to extract discriminative features from spectrogram images. The CNN is trained on a dataset of genuine speech samples. The suggested system's performance is evaluated experimentally using benchmark datasets and a range of replay attack scenarios. The findings show that the spectrogram-based technique, when paired with CNNs, effectively mitigates the impact of replay attacks on speaker recognition systems. The genuine speaker recognition system has provided 85.3% average accuracy for the training data taken as test data. The genuine speaker recognition system has provided 76% average accuracy for the independent test data. This system gives an 85% rejection rate for testing genuine models against replay attacks. The new system demonstrates improved accuracy and resilience in real-world circumstances, making it a promising solution for secure and dependable speaker recognition applications in the face of rising security threats.