Deep learning for detecting Parkinson's disease

Adrian Rosebrock of PyImageSearch recently released a brand new tutorial: Detecting Parkinson’s Disease with OpenCV, Computer Vision, and the Spiral/Wave Test which shows how to automatically detect Parkinson’s disease in hand-drawn images of spirals and waves. Adrian used classical computer vision techniques like Histogram of Oriented Gradients (HOG) for quantifying the features of the images and used them to train a Random Forest Classifier . He got an accuracy of 83.33% .

I decided to apply deep learning to this problem and see if I can push the score. It turns out that I was able to do it. You can find my notebook here.

3 Likes

Hi Sayak,
I could get 97.5% accuracy by bringing all the images from different folders together and using valid_pct=0.2…

1 Like

Would you mind doing it and share the results here? The researchers actually supplied the train and test sets separately. Hence, I decided not to break that arrangement.

2 Likes

i am really interesting about using deep learning for detecting Parkinson’disease. With the accuracy up to 97.5 %, can you do and share the results .