Overview / Usage

Epilepsy is a neurological disorder that affects millions of people worldwide, and is characterized by recurrent seizures. The detection of epilepsy and changes in mental state is important for the diagnosis and management of these conditions. One effective way to achieve this is by analysing electroencephalography (EEG) signals using deep learning algorithms. In particular, the use of 1D convolutional neural networks (CNN) and long short-term memory (LSTM) algorithms has been shown to be effective in detecting epilepsy and changes in mental state from EEG signals. To implement the CNN and LSTM algorithms, a system-on-chip (SoC) can be used to process the data from the EEG sensors. The SoC is a single chip that integrates all the components needed to perform data processing and analysis. It can include sensors for measuring EEG signals, as well as processors for analysing and interpreting the data.

Methodology / Approach

The EEG signals are first undergone through pre-processing to remove noise and artifacts, and then segmented into smaller sections for analysis. The 1D CNN is applied to the segmented EEG signals to learn relevant features and identify patterns that are associated with epilepsy or changes in mental state. The output from the CNN is then fed into the LSTM model, which is designed to capture the temporal dependencies of the EEG signals and predict whether the segment is associated with epilepsy or a change in mental state.

The CNN and LSTM models can be trained on labelled EEG data to improve their accuracy and performance. The SoC can be designed to be portable and wearable, making it useful for real-world monitoring and diagnosis of these conditions. It can also be connected to other devices, such as smartphones or cloud-based servers, for data storage and analysis. Overall, the use of 1D CNN and LSTM algorithms with an SoC provides a powerful tool for analysing EEG signals and detecting epilepsy and changes in mental state. This technology has the potential to improve the diagnosis and management of these conditions, and can be used in a variety of medical and research settings.

STAGES OF IMPLEMENTATION:

Stage-1: Building a deep learning model using 1D CNN and LSTM with good accuracy.

Stage-2: Dumping the model into the hardware (Arduino board etc) and collecting EEG signals from the sensors.

Stage-3: Developing a System On Chip for processing data with low latency and high speed, which intern helps in making it a wearable device by reducing the size.

Stage-4: Transferring processed data into cloud servers and visualizing the retrieved data using the mobile application

Technologies Used

1.Intel® AI Analytics Toolkit (AI Kit)

2.Deep learning

3.Machine learning

Documents and Presentations

Repository

https://github.com/Vishalece/AUTOMATED-DIAGNOSIS-OF-EPILEPSY-USING-EEG-SIGNALS