EEG-ML

Appearance

The background section provides an overview of the evolution of biomedical and neural signal detection and its intersection with technology, highlighting the pioneering work of the late 19th century and the milestones that have shaped the field's trajectory.

  • Adolph Beck's Doctoral Thesis (1890): Detailed mapping of the spine, medulla, and cerebellar hemispheres was published. The oscillatory behavior of the brain was observed, which was confirmed to be unrelated to the pulse and breathing.
  • String Galvanometer (1901): Introduced by Wilhelm Einthoven, it had a frequency response of 200 Hz, marking a historical point where humans were able to acquire electrical signals from an organ, from the heart.
  • First Photographic Recordings (1912): Published by Vladimir Pravdich-Neminsky, he used a moving photographic paper combined with an Einthoven string galvanometer. He clearly described alpha and beta waves and blocking behavior based on his studies on dogs.
  • 1929: Hans Berger records the first human electroencephalogram (EEG).
  • 1950s: The development of the cognitive revolution, challenging behaviorism and emphasizing the study of the mind.
  • 1960s: The discovery of neurotransmitters and their role in brain function.
  • 1970s: Introduction of computed tomography (CT) scans, allowing for non-invasive imaging of the brain.
  • 1980s: Magnetic resonance imaging (MRI) enhances brain imaging capabilities.
  • 1990s: The Decade of the Brain, leading to significant funding and research in neuroscience.
  • 2000s: Advances in neurogenetics with the completion of the Human Genome Project.
  • 2010s: The BRAIN Initiative and the Human Brain Project aim to map the brain's activity in unprecedented detail.
  • 2020s: Growth in the field of connectomics, aiming to map the brain's neural connections.
  • 2024: Current trends include the integration of artificial intelligence with neuroscience, furthering the development of BMIs and personalized medicine.
    • Machine Learning Approaches: A wide range of ML approaches can be used. The most common are Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Long Short-Term Memory (LSTM) networks, and Support Vector Machines (SVMs).
    • Feature Extraction Algorithms: A wide range of feature extraction algorithms have been tested for EEG feature classification. Some of the most common are Common Spatial Patterns (CSP), Filter Bank Common Spatial Patterns (FBCSP), and Riemannian Geometry.