Quantitative mapping of cerebrovascular reactivity amplitude and delay with breath-hold BOLD fMRI when end-tidal CO₂ quality is low

This analysis code is being shared to accompany the manuscript available here: https://www.biorxiv.org/content/10.1101/2024.11.18.624159v1

Code

Check_CO2_quality.py contains functions that can be used to assess the quality of an exhaled CO₂ timeseries and better understand a participant's compliance during a breath hold task. We recommend using the functions in the following order:

1. Use get_BH_locations() to identify the start and end locations of each breath-hold in an exhaled CO₂ timeseries and plot the timeseries with marked breath-hold periods for manual verification.
2. After manually verifying that the outputted start and end locations are correct, use get_BH_CO2_changes() to calculate the CO₂ increase associated with each breath hold.
3. Decide on a threshold to classify breath holds as "high-quality," and then use check_BH_quality() to output a binary array that is as long as the number of breath holds, with 1s corresponding to each high-quality breath hold and 0s elsewhere.
4. Use calculate_usable_segments() to identify the lengths of usable "data segments" consisting or 2 or more high-quality, consecutive breath holds.

Preprocess_Signals.py can be used to pre-process P_ETCO₂ and RVT timeseries.

• delay_correction_with_limit() delay corrects P_ETCO₂ to maximize its negative correlation with RVT. This code is based on code developed by Agrawal et al.
• standardize() z-normalizes a timeseries so that it has a mean of 0 and a standard deviation of 1

The Model folder contains code for developing a 1D-FCN to predict end-tidal CO₂ from RVT. This code is heavily based on code developed by Agrawal et al. (see citation below)

• models.py contains the code for the FCN models with 1, 2, 4, 6, 8, 10, 12, and 14 convolutional layers
• Run_Cross_Validation.py contains code to determine optimal model hyperparameters. This script will output a .csv file containing the average and standard deviation error terms for each hyperparameter combination across the 5 folds
• train_utils.py contains functions to train and test the model
• utils.py contains a peak detection function
• eval_metrics.py contains functions for calculating evaluation metrics (source)

References

Agrawal, V., Zhong, X. Z., & Chen, J. J. (2023). Generating dynamic carbon-dioxide traces from respiration-belt recordings: Feasibility using neural networks and application in functional magnetic resonance imaging. Frontiers in Neuroimaging, 2. https://doi.org/10.3389/fnimg.2023.1119539