NeuroTin is a clinical trial under the supervision of principal investigator Prof. Dr. Pascal Senn (HUG, Geneva) and supported by the Wyss Center. NeuroTin aims to compare tinnitus reduction after 3 different therapeutic approaches:
- Cognitive Behavioral Therapy (CBT), the current gold-standard treatment
- Neurofeedback with electroencephalography (EEG)
- Neurofeedback with functional magnetic resonance imaging (fMRI)
This repository contains the python implementation of the Neurofeedback paradigm using electroencephalography. Each session is articulated around 3 main steps: calibration, model, and neurofeedback.
- The calibration uses an auditory stimuli to elicit an N1-P2 evoked response.
- A model applies weights between 0 and 1 to each electrode based on the N1-P2 evoked response.
- Neurofeedback runs alternate between phases of non-regulation (also called rest) lasting 8 seconds, and phases of regulation lasting 16 seconds. During phases of regulation, participants attempt to up-regulate the ratio of alpha-band power over delta-band power displayed in real-time.
The implementation of the neurofeedback paradigm using EEG can be found on this repository.
The raw dataset folder structure is defined as:
> Data
> └─ 001
> └─ Session 1
> └─ Session 2
> └─ Calibration
> └─ Model
> └─ Online
> └─ Plots
> └─ RestingState
> └─ bads.txt
> └─ logs.txt
> └─ ...
> └─ Session 15
> └─ 002
> └─ ...
4 .csv
files are used to log different variables for every participant and
session:
mml_logs.csv
: results of the Minimum Masking Level test repeated at every session.sound_stimulus_logs.csv
: auditory stimulus settings used for calibration.model_var_logs.csv
: helmet size (54, 56, 58), model normalization variables and bad channels.scores_logs.csv
: neurofeedback scores displayed.
Many analysis function produces either (pre)processed data files or pandas
DataFrame. A list of functions accessible via the CLI can be displayed with the
command neurotin
. Help for those functions can be obtained with the --help
flag.