pyNIBS 0.2024.8__py3-none-any.whl

pynibs/data/configuration_linear_MEP.yaml

@@ -0,0 +1,61 @@
+########################################################################################################################
+# This configuration file can be used to set values in pynibs.Element() that influence the regression.                 #
+# The values in this version are optimized for MEP data.                                                     #
+########################################################################################################################
+#
+#
+############################################## General regression concept ##############################################
+# While trying to find the best fit, many possible regression functions are calculated and the best one is picked.
+# Usually, that is the one that explains the most noise, meaning that the data points have a minimal distance to the
+# regression function. With linear regression, this function always looks like the following term, with varying
+# coefficients m and n:
+##     y = m*x + n
+##     (m is the slope)
+##     (n is the intercept with the y-axis)
+
+# The usual approach is to start with a function described by some initial values (init_vals) for m and n and try out
+# many other values within the value limits (limits), minimizing the distance between regression function and real data.
+# During refitting, different initial vales (calculated within the random_vals_init_range) are used to see whether a
+# better result can be achieved.
+
+############################################### Picking suitable values ################################################
+# Initial values for the regression coefficients:
+init_vals:
+  m: 0.3
+  n: -1
+# Strategy: Pick the values you may expect the result to have, or that are not far-fetched. Picking reasonable initial
+# values can speed up the fitting procedure.
+
+# Values that limit all possible regression coefficients:
+limits:
+  m:
+  - -100
+  - 100
+  n:
+  - -100
+  - 100
+# Strategy: Rather wide range recommended, since a few points could have very extreme values and therefore be
+# approximated by a function very different from the expected values. E.g. a multiple of the presented data range.
+# (Note: This should not be used to factor out outliers, since an approximation will still be calculated, but with too
+# narrow limits it will just be a very bad one.)
+
+# Value range for the calculation of new initial values for refits:
+random_vals_init_range:
+  m:
+  - 0
+  - 100
+  n:
+  - 0
+  - 0.3
+# Strategy: During refitting, new initial values are calculated by picking a random number between these lower and upper
+# bounds, so the range should be a lot smaller than 'limits' and somewhat symmetrical around the 'init_vals'.
+
+
+# Example usage:
+##
+##        configfile = configuration_linear_MEP.yaml
+##        with open(configfile, "r") as yamlfile:
+##            config = yaml.load(yamlfile, Loader=yaml.FullLoader)
+##
+##        pynibs.regress_data(...,
+##                           **configfile)

pynibs/data/configuration_linear_RT.yaml

@@ -0,0 +1,61 @@
+########################################################################################################################
+# This configuration file can be used to set values in pynibs.Element() that influence the regression.                 #
+# The values in this version are optimized for reaction time data.                                                     #
+########################################################################################################################
+#
+#
+############################################## General regression concept ##############################################
+# While trying to find the best fit, many possible regression functions are calculated and the best one is picked.
+# Usually, that is the one that explains the most noise, meaning that the data points have a minimal distance to the
+# regression function. With linear regression, this function always looks like the following term, with varying
+# coefficients m and n:
+##     y = m*x + n
+##     (m is the slope)
+##     (n is the intercept with the y-axis)
+
+# The usual approach is to start with a function described by some initial values (init_vals) for m and n and try out
+# many other values within the value limits (limits), minimizing the distance between regression function and real data.
+# During refitting, different initial vales (calculated within the random_vals_init_range) are used to see whether a
+# better result can be achieved.
+
+############################################### Picking suitable values ################################################
+# Initial values for the regression coefficients:
+init_vals:
+  m: 0.1
+  n: 300
+# Strategy: Pick the values you may expect the result to have, or that are not far-fetched. Picking reasonable initial
+# values can speed up the fitting procedure.
+
+# Values that limit all possible regression coefficients:
+limits:
+  m:
+  - -10000
+  - 10000
+  n:
+  - -30000
+  - 30000
+# Strategy: Rather wide range recommended, since a few points could have very extreme values and therefore be
+# approximated by a function very different from the expected values. E.g. a multiple of the presented data range.
+# (Note: This should not be used to factor out outliers, since an approximation will still be calculated, but with too
+# narrow limits it will just be a very bad one.)
+
+# Value range for the calculation of new initial values for refits:
+random_vals_init_range:
+  m:
+  - -22
+  - 22
+  n:
+  - -50
+  - 500
+# Strategy: During refitting, new initial values are calculated by picking a random number between these lower and upper
+# bounds, so the range should be a lot smaller than 'limits' and somewhat symmetrical around the 'init_vals'.
+
+
+# Example usage:
+##
+##        configfile = configuration_linear_RT.yaml
+##        with open(configfile, "r") as yamlfile:
+##            config = yaml.load(yamlfile, Loader=yaml.FullLoader)
+##
+##        pynibs.regress_data(...,
+##                           **configfile)

pynibs/data/configuration_sigmoid4.yaml

@@ -0,0 +1,68 @@
+########################################################################################################################
+# This configuration file can be used to set values in pynibs.Element() that influence the regression.                 #
+# The values in this version are optimized for MEP data.                                                     #
+########################################################################################################################
+#
+#
+############################################## General regression concept ##############################################
+# While trying to find the best fit, many possible regression functions are calculated and the best one is picked.
+# Usually, that is the one that explains the most noise, meaning that the data points have a minimal distance to the
+# regression function. The function sigmoid4 always looks like the following term, with varying coefficients:
+##     y = y_0 + \frac{amp - y_0}{1+e^{-r(x-x_0)}}
+##     (x0 : Horizontal shift along the abscissa)
+##     (r : Slope parameter (steepness))
+##     (amp : Maximum value the sigmoid converges to)
+##     (y0 : Offset value of the sigmoid)
+# Note: The fourth parameter y0 is the only difference between sigmoid and sigmoid4. It presents the possibility to
+# factor in an offset in y-direction.
+
+# The usual regression approach is to start with a function described by some initial values (init_vals) and try out
+# many other values within the value limits (limits), minimizing the distance between regression function and real data.
+# During refitting, different initial vales (calculated within the random_vals_init_range) are used to see whether a
+# better result can be achieved.
+
+element_params:
+  ############################################### Picking suitable values ################################################
+  # Values transforming the initial regression coefficients:
+  init_vals:
+    x0: # set initial x_0 to rate p of the x-data range
+      p: 0.75
+    r: # slope of tangent over middle of 1/p of the e-range (e.g. 25% = 0.25 = 1/4 -> set p=4)
+      p: 4
+    amp: # amp = y_max * m + t
+      m: 1
+      t: 0
+    y0: # y0 = y_min * m + t
+      m: 1
+      t: 0
+  # Strategy: Pick the values you may expect the result to have, or that are not far-fetched. Picking reasonable initial
+  # values can speed up the fitting procedure.
+
+  # Values that limit all possible regression coefficients:
+  limits:
+    limit_factor: # the initial values get multiplied with "limit_factor" to calculate the limits
+      100
+    y0: # y0 \in [c, y_max * m]
+      c: 1.0e-12
+      m: 1
+  # Strategy: Rather wide range recommended, since a few points could have very extreme values and therefore be
+  # approximated by a function very different from the expected values. E.g. a multiple of the presented data range.
+  # (Note: This should not be used to factor out outliers, since an approximation will still be calculated, but with too
+  # narrow limits it will just be a very bad one.)
+
+  # Value range for the calculation of new initial values for refits:
+  random_vals_init_range:
+    range_factor: # the initial values get multiplied with "range_factor" to calculate this range
+      3
+  # Strategy: During refitting, new initial values are calculated by picking a random number between these lower and upper
+  # bounds, so the range should be a lot smaller than 'limits' and somewhat symmetrical around the 'init_vals'.
+
+
+  # Example usage:
+  ##
+  ##        configfile = configuration_sigmoid4.yaml
+  ##        with open(configfile, "r") as yamlfile:
+  ##            config = yaml.load(yamlfile, Loader=yaml.FullLoader)
+  ##
+  ##        pynibs.regress_data(...,
+  ##                           **configfile)

pynibs/data/network mapping configuration/configuration guide.md

@@ -0,0 +1,238 @@
+# Network detection: Guide to using the configuration files
+
+All settings and parameters relevant in the network detection process can be adjusted via the configuration files.
+Here is a short explanation of the process and the configuration entries. 
+
+## Loading the config file
+
+Using the following code you can add the desired configuration file to your script:
+
+    import yaml
+    # choose file name and path
+    configfile = 'configuration_TEMPLATE.yaml'  
+    config_path = '/data/pt_01756/software/git/pynibs/data/network mapping configuration'
+    # read configfile
+    with open(f'{config_path}/{configfile}', "r") as yamlfile:
+        config = yaml.load(yamlfile, Loader=yaml.FullLoader)
+        print("-" * 64)
+        print("Using file '" + configfile + "' for current parameter setting.")
+        print("-" * 64)
+
+Aside from that, the NDA application and testing functions only need the experimental data (e_matrix, response_values, roi_surf, fn_geo)
+and the base_path (output folder).
+
+## Configuration Parameters Guide
+
+There are two different main functions: `network_detection_algorithm_testing()` and 
+`network_detection_algorithm_application()`. The configuration parameters are very similar, many entries apply to both. 
+They are presented separately in the following for simplicity.
+
+
+### Application (real data)
+
+<u> Binarization </u>
+
+Data binarization is a crucial preprocessing step since it determines what data the algorithm considers as affected.
+Carefully adjust to the data at hand. Note that this is only relevant for the scoring_method 'clf'.
+Default function is `binarize_response_data()`. For MEPs, you could also use `binarize_real_meps()`.
+
+- ``bin_method``: Method used to calculate the binarization threshold.
+  - Options: `'mean'`, `'slope'`, `'median'`.
+
+- ``bin_factor``: Multiplied with the value of `bin_method(response)` is used as binarization threshold. 
+(Above=1=affected, below=0=not affected).
+
+<u> Scoring </u>
+
+- ``scoring_method``: Select the method for the computation of hotspot scores per ROI element. 
+  - Options: `'clf'`, `'regression'`, `'regress_data'` (single hotspot), `'mi'` (single hotspot).
+  - Remark: 
+  Network detection:
+    - `'clf'`: Scores are based on the accuracies of Decision Tree Classifiers of pairwise element combinations. (recommended)
+    - `'regression'`: Scores are the R2-scores of multivariable regression.
+    - `'regress_data'`: Scores are the R2-score of a sigmoidal fit.
+    - `'mi'`: Scores are the mutual information scores of the e-fields and the response variables.
+
+- ``scoring_emag_thr``: Threshold above which elements are scored. 
+More specifically: Per element, the `e_mag_max` as the
+maximum e-field magnitude reached in all zaps indicates how accessible an element is to stimulation (gyral element have
+the highest `e_mag_max`). This parameter is the lower bound for the `e_mag_max` of all scored elements. 
+  - Remark: The higher the threshold, the higher the scored elements are on the cortical ROI.
+
+- ``scoring_interval``: Interval at which elements are scored, determines subsampling resolution.
+  - Remark: `1/scoring_interval` elements are scored. Adjust to balance between computational efficiency and scoring detail.
+
+<u>Detection</u>
+
+- ``acc_thr``: Lower bound for the accuracy (clf) or scores (other scoring methods) of hotspot elements. Below that, 
+an element will not be categorized as hotspot.
+  - Remark: See recommendation sheet``data/network mapping configuration/recommendations_for_accuracy_threshold.md``.
+  For the clf-method, there is a table included recommending accuracy thresholds based on sample size.
+
+- ``corr_thr``: Correlation threshold (upper bound) for hotspot pairs. 
+  - Remark: When a second hotspot candidate too highly 
+  correlates with the first one, it is ignored, to avoid double detection of a single hotspot. Preliminary analyses have 
+  shown that this threshold does not have a lot of impact.
+  - 1: no limitation to how correlated they are, 0: only one hotspot allowed)
+ 
+
+<u>Additional Info</u>
+
+- ``note``: Arbitrary string for additional comments or notes.
+
+- ``fn_flag``: String to be mentioned in the output files. (to distinguish)
+
+- ``subject_id``: ID of the subject.
+- ``response_spec``:  Specification (if any) to retrieve the response. (e.g. FDI, APB,...)
+- ``e_field``:  Specification (if any) to retrieve the e-field values. (E_tan, E_mag, E_norm)
+
+
+The remaining 8 entries are not saved in the results overview. Namely the following: 
+
+- ``save_files``:  (boolean) Whether to save hotspot scores and scoremaps as HDF5 files.
+
+- ``write_effect_map``: (boolean) Whether to create an effect map. See more under <u> Effect map </u>.
+
+- ``fn_results``: Filename for the CSV the results/evaluation should be saved in.
+To understand output, see ``output_documentation``.
+
+
+<u> Effect map </u>
+
+If the output should include an effect map (useful for validation), the following parameters should be included.
+
+- ``effect_full``:  The e-field magnitude that approximately needs to be reached to achieve the desired effect (in  µV).
+
+- ``effect_saturation``: The approx. e-field magnitude at which the effect is expected to saturate (in  µV).
+
+The effect map depicts the following: Assuming a network as detected was in effect, where can the effect
+be expected highest/lowest? I.e., where is the response expected to be affected most/least?
+This can be useful during experimental validation. 
+
+<u>Plotting</u> 
+
+- ``plot_std``: (boolean) Whether to plot the response data with color indicating response value, and the e-field values
+of the chosen hotspots on x- and y-axis.
+
+- ``plot_bin``: (boolean) Whether to plot the binarized response data with color indicating response category, 
+and the e-field values of the chosen hotspots on x- and y-axis.
+
+- ``plot_curves``: (boolean) Whether to create two response data scatter plots, one for each chosen hotspot. (e-field 
+magnitudes on x-axis, response on y-axis).
+
+
+### Testing (generated data)
+
+<u> Artificial Data Generation </u>
+
+The response will be generated as if `hotspot_elm0` and `hotspot_elm1` 
+form a network of type `network_type`. See function `create_response_data()` for the detailed process.
+
+- ``hotspot_elm0``:  `[0, n_elms-1]`
+Specify the index of the first hotspot element. 
+
+- ``hotspot_elm1``: `[0, n_elms-1]`
+Specify the index of the second hotspot element.
+
+- ``sample_size``: The desired number of samples. Bounded by the number of available e-fields. 
+If `sample_size` is smaller, e-fields are randomly chosen. 
+(See `create_dummy_binary_data.py` for more elaborate e-field choosing functions.)
+
+- ``rn_seed``: Seed for all processes that are in some way random, ensuring reproducibility.
+
+- ``distribution_type``: Defines the statistical distribution type for data generation.
+  - Options: `'normal'`, `'logistic'`.  
+  - Remark: `'ExGaussian'` is also an option but is not fully implemented. 
+
+- ``network_type``: Choose the network the two hotspots are forming. 
+  - Options: `['NO', 'AND', '1_INH_0', 'SH_0', '0_INH_1', 'SH_1', 'XOR', 'OR']`.
+  - Remark:
+    - (1) NO: No network ("pseudonetwork").
+    - (2) AND: Dual node network: Effect if elm0 AND elm1 are stimulated.
+    - (3) 1_INH_0: Dual node network: elm1 inhibits elm0, elm0 has an effect.
+    - (4) SH_0: Single hotspot: Only elm0 has an effect.
+    - (5) 0_INH_1: Dual node network: elm0 inhibits elm1, elm1 has an effect.
+    - (6) SH_1: Single hotspot: Only elm1 has an effect.
+    - (7) XOR: Dual node network: elm0 inhibits elm1, elm1 inhibits elm0. Both have an effect.
+    - (8) OR: Dual node network: Effect if either elm0 or elm1 is stimulated.
+
+- ``effect_full``: E-field magnitude at which the desired effect fully manifests.
+  - Remark: What value makes sense here strongly depends on the range of the e-field values. 
+  For normalized e-field values (1% stimulator intensity) across the motor cortex, values tend to range up to 1.6 µV. 
+  We chose to normalize the effect to 0.9 µV.
+
+- ``effect_saturation``: E-field magnitude at which the desired effect saturates.
+  - Remark: See `effect_full`. We chose an effect saturation at 1.2 µV.
+
+- ``jitter_ratio``: Percentage of data receiving random response values (uniformly distributed within the expected range).
+
+- ``jitter_scale``: Multiplied with the mean value to calculate distribution deviation. Note that a heteroscedastic 
+distribution is implemented at the moment -> the higher the effect, the higher the deviation. 
+
+
+<u> Binarization </u>
+
+Data binarization is a crucial preprocessing step since it determines what data the algorithm considers as affected.
+Carefully adjust to the data at hand.
+Default function is `binarize_response_data()`. For MEPs, you could also use `binarize_real_meps()`.
+
+- ``bin_method``: Method used to calculate the binarization threshold.
+  - Options: `'mean'`, `'slope'`, `'median'`.
+
+- ``bin_factor``: Multiplied with the value of `bin_method(response)` is used as binarization threshold. 
+(Above=1=affected, below=0=not affected).
+
+<u> Scoring </u>
+
+- ``scoring_method``: Select the method for the computation of hotspot scores per ROI element. 
+  - Options: `'clf'`, `'regression'`, `'regress_data'` (single hotspot), `'mi'` (single hotspot).
+  - Remark: 
+  Network detection:
+    - `'clf'`: Scores are based on the accuracies of Decision Tree Classifiers of pairwise element combinations. (recommended)
+    - `'regression'`: Scores are the R2-scores of multivariable regression.
+    - `'regress_data'`: Scores are the R2-score of a sigmoidal fit.
+    - `'mi'`: Scores are the mutual information scores of the e-fields and the response variables.
+
+- ``scoring_emag_thr``: Threshold above which elements are scored. 
+More specifically: Per element, the `e_mag_max` as the
+maximum e-field magnitude reached in all zaps indicates how accessible an element is to stimulation (gyral element have
+the highest `e_mag_max`). This parameter is the lower bound for the `e_mag_max` of all scored elements. 
+  - Remark: The higher the threshold, the higher the scored elements are on the cortical ROI.
+
+- ``scoring_interval``: Interval at which elements are scored, determines subsampling resolution.
+  - Remark: `1/scoring_interval` elements are scored. Adjust to balance between computational efficiency and scoring detail.
+
+<u>Detection</u>
+
+- ``acc_thr``: Lower bound for the accuracy (clf) or scores (other scoring methods) of hotspot elements. Below that, 
+an element will not be categorized as hotspot.
+  - Remark: Remark: See recommendation sheet``data/network mapping configuration/recommendations_for_accuracy_threshold.md``.
+  For the clf-method, there is a table included recommending accuracy thresholds based on sample size.
+
+- ``corr_thr``: Correlation threshold (upper bound) for hotspot pairs. 
+  - Remark: When a second hotspot candidate too highly 
+  correlates with the first one, it is ignored, to avoid double detection of a single hotspot. Preliminary analyses have 
+  shown that this threshold does not have a lot of impact.
+  - 1: no limitation to how correlated they are, 0: only one hotspot allowed)
+
+- ``note``: Arbitrary string for additional comments or notes.
+
+The remaining 5 entries are not saved in the results overview. Namely the following: 
+
+<u>Evaluation</u>
+
+- ``save_files``: (boolean) Whether to save hotspot scores and scoremaps as HDF5 files.
+
+- ``fn_results``: Filename for the CSV the results/evaluation should be saved in.
+To understand output, see ``output_documentation``.
+
+<u>Plotting</u> 
+
+- ``plot_std``: (boolean) Whether to plot the response data with color indicating response value, and the e-field values
+of the chosen hotspots on x- and y-axis.
+
+- ``plot_bin``: (boolean) Whether to plot the binarized response data with color indicating response category, 
+and the e-field values of the chosen hotspots on x- and y-axis.
+
+- ``plot_curves``: (boolean) Whether to create two response data scatter plots, one for each chosen hotspot. (e-field 
+magnitudes on x-axis, response on y-axis).

pynibs/data/network mapping configuration/configuration_TEMPLATE.yaml

@@ -0,0 +1,42 @@
+# This configuration file contains settings and parameters for the application of the network detection algorithm.
+
+# BINARIZATION
+# binarization threshold is calculated as follows: method(response) * bin_factor
+bin_method: 'mean'           # mean, median or slope
+bin_factor: 1.0
+
+# SCORING
+scoring_method: 'clf'         # dual network expected: 'clf' or 'regression', single hotspot expected: 'regress_data' or 'mi'
+scoring_emag_thr: 0           # elements with `max(e_mag) > scoring_emag_thr` are scored
+scoring_interval: 22          # subsampling resolution: `1/scoring_interval' of all elements are scored
+
+# DETECTION
+acc_thr: 0.70                  # accuracy threshold for hotspot elements (lower bound)
+corr_thr: 0.5                 # correlation(hot0, hot1) threshold for hotspot pairs (upper bound)
+
+# ADDITIONAL INFO
+note: 'MI_score mapping reasons'     # arbitrary string    # NO COMMAS
+fn_flag: 'testrun_11'              # mentioned in the file name
+subject_id: '35357.53'          #'09440.22' # '13061.30' 35334.2b
+response_spec: 'FDI'            # specification (if any) to retrieve the response
+efield_spec: 'E_mag'            # E_mag, E_tan, E_norm
+
+
+####################################################
+# The following is not saved in the evaluation sheet:
+####################################################
+
+save_files: True                # whether to save experimental data & hotspot score data as hdf5
+write_effect_map: True          # whether to create effect map which can be used during validation
+fn_results: 'evaluation_template'             # file name of the .csv to save the results in
+
+# EFFECT
+# parameters used to create effect map - depends on e-field data and does not influence mapping result
+effect_full: 42             # at what e-field magnitude the desired effect is expected to manifest fully
+effect_saturation: 70       # at what e-mag the desired effect should saturate
+
+# PLOTTING of the found data (not in evaluation sheet)
+plot_std: True                # plot e-mag hotspot0 x emag hotspot1 x response
+plot_bin: True                # plot e-mag hotspot0 x emag hotspot1 x binarized response
+plot_curves: True             # plot e-mag hotspot0 x response and e-mag hotspot1 x response
+

pynibs/data/network mapping configuration/configuration_for_testing.yaml

@@ -0,0 +1,43 @@
+# This configuration file contains settings and parameters for the testing of the network detection algorithm
+# by generating response data and then test the algorithm on it.
+
+# ARTIFICIAL DATA GENERATION
+hotspot_elm0: 8797            # hotspot element 0
+hotspot_elm1: 5334            # hotspot element 1
+sample_size: 300              # number of data manifestations (i.e. trials / coil configurations / ...)
+rn_seed: 44               # random seed used in multiple places
+distribution_type: 'normal'   # 'normal', 'logistic' are implemented
+network_type: 'SH_0'           # \in ['NO', 'AND', '1_INH_0', 'SH_0', '0_INH_1', 'SH_1', 'XOR', 'OR']
+effect_full: 0.9              # at what e-field magnitude the desired effect is expected to manifest fully
+effect_saturation: 1.2       # effect \in [0, effect_upper_bound) (at what e-mag the desired effect should saturate)
+jitter_ratio: 0.05            # percentage of data that gets assigned random response values within observed range
+jitter_scale: 0.2             # gets multiplied with the mean value to calculate distribution deviation:
+                              ##   scale = loc * jitter_scale (smaller -> less noise)
+
+# BINARIZATION
+# binarization threshold is calculated as follows: method(response) * bin_factor
+bin_method: 'mean'
+bin_factor: 1.0
+
+# SCORING
+scoring_method: 'clf'         # dual network expected: 'clf' or 'regression', single hotspot expected: 'regress_data' or 'mi'
+scoring_emag_thr: 0           # elements with max(e_mag) > scoring_emag_thr are scored
+scoring_interval: 22          # subsampling resolution: `1/scoring_interval' of all elements are scored
+
+# DETECTION
+acc_thr: 0.70                  # accuracy threshold for hotspot elements (lower bound)
+corr_thr: 0.75                 # correlation(hot0, hot1) threshold for hotspot pairs (upper bound)
+note: 'PLOTTING'                  # arbitrary string  # NO COMMAS
+
+
+####################################################
+# The following is not saved in the evaluation sheet:
+####################################################
+
+save_files: True               # whether to save hotspot scores and scoremaps as hdf5
+fn_results: 'eval_final_tests'    #  file name of the .csv to save the results and evaluation in
+
+# PLOTTING of the generated data (not in evaluation sheet)
+plot_std: True                # plot e-mag hotspot0 x emag hotspot1 x response
+plot_bin: True                # plot e-mag hotspot0 x emag hotspot1 x binarized response
+plot_curves: True             # plot e-mag hotspot0 x response and e-mag hotspot1 x response

pynibs/data/network mapping configuration/configuration_modelTMS.yaml

@@ -0,0 +1,43 @@
+# This configuration file contains settings and parameters for the application of the network detection algorithm
+#  on reaction time data.
+
+# BINARIZATION
+# binarization threshold is calculated as follows: method(response) * bin_factor
+bin_method: 'mean'           # mean, median or slope
+bin_factor: 1.3
+
+# SCORING
+scoring_method: 'clf'         # dual network expected: 'clf' or 'regression', single hotspot expected: 'regress_data' or 'mi'
+scoring_emag_thr: 0           # elements with `max(e_mag) > scoring_emag_thr` are scored
+scoring_interval: 22          # subsampling resolution: `1/scoring_interval' of all elements are scored
+
+# DETECTION
+acc_thr: 0.8                  # accuracy threshold for hotspot elements (lower bound)
+corr_thr: 0.75                 # correlation(hot0, hot1) threshold for hotspot pairs (upper bound)
+
+# ADDITIONAL INFO
+note: 'MI_score mapping reasons'     # arbitrary string    # NO COMMAS
+fn_flag: 'testrun_11'              # mentioned in the file name
+subject_id: '35357.53'          # '09440.22' # '13061.30' 35334.2b
+response_spec: 'rt'            # specification (if any) to retrieve the response
+efield_spec: 'E_mag'            # E_mag, E_tan, E_norm
+
+
+####################################################
+# The following is not saved in the evaluation sheet:
+####################################################
+
+save_files: True                # whether to save experimental data & hotspot score data as hdf5
+write_effect_map: True          # whether to create effect map which can be used during validation
+fn_results: 'eval_model_TMS'             # file name of the .csv to save the results in
+
+# EFFECT
+# parameters used to create effect map - depends on e-field data and does not influence mapping result
+effect_full: 42             # at what e-field magnitude the desired effect is expected to manifest fully
+effect_saturation: 55       # at what e-mag the desired effect should saturate
+
+# PLOTTING of the found data (not in evaluation sheet)
+plot_std: True                # plot e-mag hotspot0 x emag hotspot1 x response
+plot_bin: True                # plot e-mag hotspot0 x emag hotspot1 x binarized response
+plot_curves: True             # plot e-mag hotspot0 x response and e-mag hotspot1 x response
+

pynibs/data/network mapping configuration/configuration_reg_isi_05.yaml

@@ -0,0 +1,43 @@
+# This configuration file contains settings and parameters for the application of the network detection algorithm
+#  on MEP data. More details below.
+
+# BINARIZATION
+# binarization threshold is calculated as follows: method(response) * bin_factor
+bin_method: 'mean'           # mean, median or slope
+bin_factor: 1.0
+
+# SCORING
+scoring_method: 'clf'         # dual network expected: 'clf' or 'regression', single hotspot expected: 'regress_data' or 'mi'
+scoring_emag_thr: 0           # elements with `max(e_mag) > scoring_emag_thr` are scored
+scoring_interval: 22          # subsampling resolution: `1/scoring_interval' of all elements are scored
+
+# DETECTION
+acc_thr: 0.70                  # accuracy threshold for hotspot elements (lower bound)
+corr_thr: 0.5                 # correlation(hot0, hot1) threshold for hotspot pairs (upper bound)
+
+# ADDITIONAL INFO
+note: 'MI_score mapping reasons'     # arbitrary string    # NO COMMAS
+fn_flag: 'testrun_11'              # mentioned in the file name
+subject_id: '35357.53'          #'09440.22' # '13061.30' 35334.2b
+response_spec: 'FDI'            # FDI / ADM / APB
+efield_spec: 'E_mag'            # E_mag, E_tan, E_norm
+
+
+####################################################
+# The following is not saved in the evaluation sheet:
+####################################################
+
+save_files: True                # whether to save experimental data & hotspot score data as hdf5
+write_effect_map: True          # whether to create effect map which can be used during validation
+fn_results: 'evaluation_template'             # file name of the .csv to save the results in
+
+# EFFECT
+# parameters used to create effect map - depends on e-field data and does not influence mapping result
+effect_full: 42             # at what e-field magnitude the desired effect is expected to manifest fully
+effect_saturation: 70       # at what e-mag the desired effect should saturate
+
+# PLOTTING of the found data (not in evaluation sheet)
+plot_std: True                # plot e-mag hotspot0 x emag hotspot1 x response
+plot_bin: True                # plot e-mag hotspot0 x emag hotspot1 x binarized response
+plot_curves: True             # plot e-mag hotspot0 x response and e-mag hotspot1 x response
+