Parkinson Disease Classification

This is an implementation of early identification of Parkinson's disease from hand-drawn images with inception_v3 and densenet121 as feature descriptors and various classification algorithms to separate them as healthy or affected.

To get the .apk click this link.

Application UI

• Front

  

• Spiral healthy test

  

• Spiral parkinson test

  

• Wave healthy test

  

• Wave parkinson test

  

Sample Images

• Wave Dataset with DenseNet121

  • Original

    Fig: (a) and (b) represents the train set sample, (c) and (d) are the test set sample of wave dataset

  • Heatmap

    

  • Grad-Cam

    Fig: (b) and (d) are the grad-cam image of (a) and (c) respectively

  • Evaluation

    

    

    

    

    

• Spiral Dataset with Inception V3

  • Original

    Fig: (a) and (b) represents the train set sample, (c) and (d) are the test set sample of spiral dataset

  • Heatmap

    

  • Grad-Cam

    Fig: (b) and (d) are the grad-cam image of (a) and (c) respectively

  • Evaluation

    

    

    

    

    

Installation

• git clone https://github.com/rjarman/parkinson-disease-classification.git
• cd parkinson-disease-classification/notebooks
• pip install -r requirements.txt /media/rafsun/Working Drive/projects/parkinson-disease-classification/notebooks/images/densenet121_wave