HypoSubstrateTesting

Testing hypothesized neural substrates with lesions

ABOUT

This code has been developed by the High Dimensional Neurology group at the Queen Square Institute of Neurology UCL.

Authors:
Parashkev Nachev;
Patrik Bey

INPUT REQUIREMENTS

The framework assumes the following file structure:

DATA_DIR
    |_lesions
        |_lesion1.nii
        |_lesion2.nii
    |_masks
        |_mask1.nii
        |_mask2.nii

SOFTWARE REQUIREMENTS

1. Matlab
2. SPM
3. BASH (optional for automated run script usage)

STEPS

1. Computing lesion mask morphological properties

# single function call
run_step1_morphology

2. Computing lesion mask / target mask features

# single function call
run_step2_features 

3. Running simulations

# ---- true substrate regression ---- #
# example use case for true substrate regression model.

REGRESSION_TARGETS=(
    "targetfile"
)

BASE_IMAGES=(
    "MDN.nii")

TARGET_FILES=(
    "MDN.nii")

run_step3_simulations

# ---- false substrate regression ---- #
# example use case for MDN.nii base image and RF.nii showcasing the false substrate regression model.

REGRESSION_TARGETS=(
    "targetfile"
)

BASE_IMAGES=(
    "MDN.nii")

TARGET_FILES=(
    "RF.nii")

run_step3_simulations

4. Visualizations of significance ratios across different models

# exemplary call for true-substrate regression models
python create_semi_lolli.py /path/to/true-substrate/SignificanceRatios.csv --spoke-color 'darkgrey' --limit 105

REFERENCES

Patrik Bey, Joe Mole, James K. Ruffle, Amy Nelson, Edgar Chan, Parashkev Nachev, and Lisa Cipolotti, Testing hypothesized neural substrates with lesions, (2026), under review