Datasets
Models
95 lines (60 with data), 5.0 kB
CinC2023
Predicting Neurological Recovery from Coma After Cardiac Arrest: The George B. Moody PhysioNet Challenge 2023
- The Conference
- Description of the files/folders(modules)
- Distributions of the EEG data against clinical information of the patients
- External Resources Used
- SQI (Signal Quality Index) Calculation
The Conference
Click to view the conference poster
Conference paper: GitHub | IEEE Xplore | [CinC Papers On-line](https://cinc.org/archives/2023/pdf/CinC2023-060.pdf)
👉 Back to TOC
Description of the files/folders(modules)
Files
Click to view the details
- [README.md](README.md): this file, serves as the documentation of the project. - [cfg_models.py](cfg_models.py), [cfg.py](cfg.py): configuration files (the former for configuration of models, the latter for configuration of the whole project) - [data_reader.py](data_reader.py): data reader, including data downloading, file listing, data loading, etc. - [dataset.py](dataset.py): dataset class, which feeds data to the models. - [Dockerfile](Dockerfile): docker file for building the docker image for submissions. - [requirements.txt](requirements.txt), [requirements-docker.txt](requirements-docker.txt), [requirements-no-torch.txt](requirements-no-torch.txt): requirements files for different purposes. - [trainer.py](trainer.py): trainer class, which trains the models.Folders(Modules)
Click to view the details
- [models](models): folder for model definitions, including [CRNN models](models/crnn.py), and [traditional ML models](models/ml.py). The latter serves as a minimal garantee model using patient metadata only, which is used when no (EEG) data is available. It is indeed a wrapper containing model construction, training, hyperparameter tuning via grid search, model saving/loading, and end-to-end inference (from raw input to the form of output that the challenge requires). - [utils](utils): various utility functions, as well as some intermediate data files (e.g. train-val split files, etc.). SQI computation code, as mentioned in the unofficial phase (and also the [v1 version of the I-CARE database](https://physionet.org/content/i-care/1.0/)). This will be described in detail in the [External Resources Used](#external-resources-used) section.👉 Back to TOC
Distributions of the EEG data against clinical information of the patients
  Â
  Â
👉 Back to TOC
External Resources Used
SQI (Signal Quality Index) Calculation
Source Code integrated from bdsp-core/icare-dl.
As stated in the Artfiact Screening (Signal Quality) subsection of the Data Description section of the
I-CARE database version 1.0 hosted at PhysioNet, the SQI is calculated as follows:
...This artifact score is based on how many 10-second epochs within a 5-minute EEG window are contaminated by artifacts. Each 10-second epoch was scored for the presence of the following artifacts including: 1) flat signal, 2) extreme high or low values, 3) muscle artifact, 4) non-physiological spectra, and 5) implausibly fast rising or decreasing signal amplitude...
Precomputed SQI (5min window (epoch), 1min step length) for all EEGs: Google Drive | Alternative
Distribution of SQI for all 5min windows (epochs):
👉 Back to TOC