search

fepegar / gestures-miccai-2021

Code for "Pérez-García et al. 2021, Transfer Learning of Deep Spatiotemporal Networks to Model Arbitrarily Long Videos of Seizures, MICCAI 2021".
https://doi.org/10.1007/978-3-030-87240-3_32
6 stars 3 forks source link

Dependencies #1

Closed HaukeCornell closed 2 years ago

HaukeCornell commented 2 years ago

Best Fernando, thanks for providing this repository alongside the feature dataset and the paper. This is essential work!

I have run into a few issues replicating it, though. What specific anaconda version were you using? I had to add protobuf~=3.19.0 and pillow<7 to the requirements to get your code to execute. Also, I created the slack.json for sacred and changed a few file paths in your code. However, it still fails on: train_features_lstm failed after 3 seconds with AttributeError: 'DataFrame' object has no attribute 'RatioGTCS'.

fepegar commented 2 years ago

Hi, @HaukeCornell. I'm glad you find this research useful!

I'm afraid the code is not fully reproducible as I'm not allowed to share the data. You can, however, adapt the code to your own needs.

I used a CSV file containing one row per seizure. The Duration column was the seizure duration in seconds. RatioGTCS is a number between 0 and 1 representing how far into the seizure the patient starts convulsing. I hope that helps!