Wavelet-Transformed EEG Data for Alcohol Detection

This dataset contains wavelet-transformed EEG signal images (scalograms) designed for binary classification tasks, distinguishing individuals as either “Alcoholic” or “Control” groups. It provides high-quality preprocessed data for advanced machine learning and deep learning applications in biomedical research.

Key Features:

• Source Data: Derived from the publicly available , originally sampled at 256 Hz from 61 electrodes over the scalp during 1-second intervals.

• Preprocessing:

  • Applied Continuous Wavelet Transform (CWT) to convert raw EEG signals into scalograms, showcasing detailed time-frequency characteristics.
  • Images resized to 200×172 pixels, optimized for deep learning model compatibility.

• Dataset Composition:

  • Total Images: 924
    • Alcoholic Group: 468 images
    • Control Group: 456 images
  • File Naming Convention:
    • Format: subjectX_channelY.jpg
      • subjectX: Unique identifier for the subject.
      • channelY: EEG channel (1–61).

Applications:

1. Alcoholism Detection: Train and benchmark machine learning models for classifying EEG signals as alcoholic or control.
2. Biomedical Research: Explore time-frequency characteristics of EEG signals for alcohol dependency studies.
3. Deep Learning Tasks: Utilize the dataset for evaluating CNNs and other architectures on biomedical image classification problems.