To implement and enhance EEG disease classification with CM-Attention Mechanism using Deep CNN-Bi-LSTM
Python, TensorFlow, Keras, Deep CNN, BiLSTM, Attention Mechanism, NumPy, Pandas, OpenCV, PIL, Matplotlib
- Designed a new architecture of Deep CNN integrated with BiLSTM which has a customized mixed attention mechanism for classifying brain diseases from EEG-MRI images achieving 99.45% accuracy.
- Merged, resized, normalized, data augmented, and prepared two datasets found on Kaggle containing brain MRI images (over 5000 images) to improve generalization performance.
- Deep CNN was used to extract spatial features while Bi-LSTM captured temporal features. The attention layer improved relevant feature focus, thus enhancing classification accuracy.
View Project