Class: InferenceSettings

One thresholding and correction procedure. This is the canonical storage home for thresholding, correction, and cluster-inference metadata. The test statistic itself belongs to the tested effect and lives in Effect.statistic.

One record may be referenced by several Analyses, so every field must hold for all of them; an analysis thresholded or corrected differently takes its own record. The test is mechanical: two analyses share a record when the threshold, its type, the correction method, the scope corrected over and the alpha are all the same. A paper that states one correction scheme in Methods and applies it throughout has one record.

Inference attaches to the analysis, not to the fit: one model estimated once is often thresholded several ways -- an ROI test Bonferroni-corrected over four regions and a searchlight over the same fit corrected by FDR. See storage-schema-design-notes.md for the reasoning behind this shape.

URI: neurostudy:InferenceSettings

 classDiagram
    class InferenceSettings
    click InferenceSettings href "../InferenceSettings/"
      InferenceSettings : alpha_level

      InferenceSettings : cluster_extent_threshold

      InferenceSettings : clusterwise_threshold_value

      InferenceSettings : correction_regions





        InferenceSettings --> "*" Region : correction_regions
        click Region href "../Region/"



      InferenceSettings : correction_scope

      InferenceSettings : height_threshold_type

      InferenceSettings : height_threshold_value

      InferenceSettings : id

      InferenceSettings : inference_level

      InferenceSettings : multiple_comparison_method

      InferenceSettings : neighborhood_definition

      InferenceSettings : number_of_tests

      InferenceSettings : permutation_count

      InferenceSettings : search_volume

      InferenceSettings : tfce_parameters

      InferenceSettings : tfce_used

Slots

Name	Cardinality and Range	Description	Inheritance
id	1 String	Stable identifier used by Analysis foreign-key references	direct
inference_level	0..1 String	Level at which inference is made, such as voxel, vertex, cluster, region, net...	direct
height_threshold_value	0..1 Float	The threshold applied to each element's statistic, before correction and befo...	direct
height_threshold_type	0..1 String	Meaning of height_threshold_value, such as p, q, t, z, or F	direct
cluster_extent_threshold	0..1 Integer	Minimum surviving cluster extent in voxels, vertices, or the unit stated in t...	direct
clusterwise_threshold_value	0..1 Float	Numeric cluster-level threshold for retaining clusters	direct
multiple_comparison_method	0..1 String	Correction method, such as FWE, FDR, Bonferroni, permutation, Monte Carlo, or...	direct
correction_scope	0..1 String or  SpatialScope	Whether the multiple-comparison correction searched the whole acquired volume...	direct
correction_regions	* Region	The regions correction was restricted to -- what correction_scope: roi name...	direct
search_volume	0..1 String	Reported search volume, mask size, or number of vertices or voxels used for i...	direct
neighborhood_definition	0..1 String	Voxel or vertex adjacency or connectivity rule used for cluster formation	direct
tfce_used	0..1 Boolean	Whether threshold-free cluster enhancement was used	direct
tfce_parameters	0..1 String	TFCE parameters such as height and extent exponents when reported	direct
permutation_count	0..1 Integer	Number of permutations, randomizations, or simulations used for inference	direct
alpha_level	0..1 Float	Family-wise, false-discovery, or nominal alpha threshold when reported	direct
number_of_tests	0..1 Integer	Number of statistical tests considered by the multiple-comparison procedure	direct

Usages

used by	used in	type	used
Analysis	inference_settings	range	InferenceSettings
Study	inference_settings	range	InferenceSettings

Rules

Rule Applied	Preconditions	Postconditions	Elseconditions
slot_conditions	{'correction_scope': {'any_of': [{'equals_string': 'whole_brain'}, {'equals_string': 'searchlight'}]}}	{'correction_regions': {'value_presence': 'ABSENT'}}

Rule Applied	Preconditions	Postconditions	Elseconditions
slot_conditions	{'correction_scope': {'equals_string': 'roi'}}	{'correction_regions': {'value_presence': 'PRESENT'}}

Identifier and Mapping Information

Schema Source

• from schema: https://neurostuff.org/schemas/neuroimaging-study-storage

Mappings

Mapping Type	Mapped Value
self	neurostudy:InferenceSettings
native	neurostudy:InferenceSettings

LinkML Source

Direct

name: InferenceSettings
description: 'One thresholding and correction procedure. This is the canonical storage
  home for thresholding, correction, and cluster-inference metadata. The test statistic
  itself belongs to the tested effect and lives in Effect.statistic.

  One record may be referenced by several Analyses, so every field must hold for all
  of them; an analysis thresholded or corrected differently takes its own record.
  The test is mechanical: two analyses share a record when the threshold, its type,
  the correction method, the scope corrected over and the alpha are all the same.
  A paper that states one correction scheme in Methods and applies it throughout has
  one record.

  Inference attaches to the analysis, not to the fit: one model estimated once is
  often thresholded several ways -- an ROI test Bonferroni-corrected over four regions
  and a searchlight over the same fit corrected by FDR. See storage-schema-design-notes.md
  for the reasoning behind this shape.'
from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage
rank: 1000
attributes:
  id:
    name: id
    description: Stable identifier used by Analysis foreign-key references.
    in_subset:
    - deterministic
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    identifier: true
    domain_of:
    - Arm
    - Timepoint
    - Group
    - Task
    - Condition
    - Assessment
    - Region
    - Acquisition
    - Device
    - Analysis
    - Measure
    - Preprocessing
    - ModelEstimation
    - ModelTerm
    - InferenceSettings
    - StatisticalMap
    - Table
    - CoordinateSet
    - Study
    range: string
    required: true
  inference_level:
    name: inference_level
    description: Level at which inference is made, such as voxel, vertex, cluster,
      region, network or other.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: string
  height_threshold_value:
    name: height_threshold_value
    description: 'The threshold applied to each element''s statistic, before correction
      and before any cluster-level test. Papers name it for whichever of its two jobs
      they are describing -- the *voxelwise* (or vertexwise) threshold when elements
      are tested individually, the *cluster-forming* threshold when surviving elements
      are grouped into clusters -- and it is one number either way, applied once to
      the map. Record it once here, whichever word the source uses.

      Not the cluster-level threshold: `clusterwise_threshold_value` is the p or q
      a whole cluster had to beat, and `cluster_extent_threshold` the size it had
      to reach. A paper reporting "p < 0.001 uncorrected, cluster-level p < 0.05 FWE"
      states this field and `clusterwise_threshold_value`, not this field twice.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: float
    minimum_value: 0
  height_threshold_type:
    name: height_threshold_type
    description: Meaning of height_threshold_value, such as p, q, t, z, or F.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: string
  cluster_extent_threshold:
    name: cluster_extent_threshold
    description: Minimum surviving cluster extent in voxels, vertices, or the unit
      stated in the source.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: integer
    minimum_value: 0
  clusterwise_threshold_value:
    name: clusterwise_threshold_value
    description: Numeric cluster-level threshold for retaining clusters.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: float
    minimum_value: 0
  multiple_comparison_method:
    name: multiple_comparison_method
    description: Correction method, such as FWE, FDR, Bonferroni, permutation, Monte
      Carlo, or none.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: string
  correction_scope:
    name: correction_scope
    description: 'Whether the multiple-comparison correction searched the whole acquired
      volume or was restricted to regions. The correction''s own domain, not the analysis''s:
      a whole-brain voxelwise model whose significance was corrected inside an anatomical
      mask has `Analysis.spatial_scope: whole_brain` and `correction_scope: roi`,
      and the two differing is the ordinary case rather than a contradiction.

      A domain the enum cannot name is written here as free text. An inference level
      such as "cluster level" is not a scope and belongs in `inference_level`.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    any_of:
    - range: SpatialScope
    - range: string
  correction_regions:
    name: correction_regions
    description: 'The regions correction was restricted to -- what `correction_scope:
      roi` names, as `Analysis.regions` names what `spatial_scope: roi` restricts
      the search space to. Empty for a whole-brain correction.

      Not the analysis''s search space. Naming the correction mask on the analysis
      instead says the model was fitted within it, which is a different claim about
      what the paper did.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: Region
    multivalued: true
    inlined: false
  search_volume:
    name: search_volume
    description: Reported search volume, mask size, or number of vertices or voxels
      used for inference.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: string
  neighborhood_definition:
    name: neighborhood_definition
    description: Voxel or vertex adjacency or connectivity rule used for cluster formation.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: string
  tfce_used:
    name: tfce_used
    description: Whether threshold-free cluster enhancement was used.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: boolean
  tfce_parameters:
    name: tfce_parameters
    description: TFCE parameters such as height and extent exponents when reported.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: string
  permutation_count:
    name: permutation_count
    description: Number of permutations, randomizations, or simulations used for inference.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: integer
    minimum_value: 0
  alpha_level:
    name: alpha_level
    description: Family-wise, false-discovery, or nominal alpha threshold when reported.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: float
    minimum_value: 0
    maximum_value: 1
  number_of_tests:
    name: number_of_tests
    description: Number of statistical tests considered by the multiple-comparison
      procedure.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    domain_of:
    - InferenceSettings
    range: integer
    minimum_value: 0
rules:
- preconditions:
    slot_conditions:
      correction_scope:
        name: correction_scope
        any_of:
        - equals_string: whole_brain
        - equals_string: searchlight
  postconditions:
    slot_conditions:
      correction_regions:
        name: correction_regions
        value_presence: ABSENT
  description: Whole-brain and searchlight correction restricts to no region.
- preconditions:
    slot_conditions:
      correction_scope:
        name: correction_scope
        equals_string: roi
  postconditions:
    slot_conditions:
      correction_regions:
        name: correction_regions
        value_presence: PRESENT
  description: Correction over an ROI must name the regions it was restricted to.

Induced

name: InferenceSettings
description: 'One thresholding and correction procedure. This is the canonical storage
  home for thresholding, correction, and cluster-inference metadata. The test statistic
  itself belongs to the tested effect and lives in Effect.statistic.

  One record may be referenced by several Analyses, so every field must hold for all
  of them; an analysis thresholded or corrected differently takes its own record.
  The test is mechanical: two analyses share a record when the threshold, its type,
  the correction method, the scope corrected over and the alpha are all the same.
  A paper that states one correction scheme in Methods and applies it throughout has
  one record.

  Inference attaches to the analysis, not to the fit: one model estimated once is
  often thresholded several ways -- an ROI test Bonferroni-corrected over four regions
  and a searchlight over the same fit corrected by FDR. See storage-schema-design-notes.md
  for the reasoning behind this shape.'
from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage
rank: 1000
attributes:
  id:
    name: id
    description: Stable identifier used by Analysis foreign-key references.
    in_subset:
    - deterministic
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    identifier: true
    owner: InferenceSettings
    domain_of:
    - Arm
    - Timepoint
    - Group
    - Task
    - Condition
    - Assessment
    - Region
    - Acquisition
    - Device
    - Analysis
    - Measure
    - Preprocessing
    - ModelEstimation
    - ModelTerm
    - InferenceSettings
    - StatisticalMap
    - Table
    - CoordinateSet
    - Study
    range: string
    required: true
  inference_level:
    name: inference_level
    description: Level at which inference is made, such as voxel, vertex, cluster,
      region, network or other.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: string
  height_threshold_value:
    name: height_threshold_value
    description: 'The threshold applied to each element''s statistic, before correction
      and before any cluster-level test. Papers name it for whichever of its two jobs
      they are describing -- the *voxelwise* (or vertexwise) threshold when elements
      are tested individually, the *cluster-forming* threshold when surviving elements
      are grouped into clusters -- and it is one number either way, applied once to
      the map. Record it once here, whichever word the source uses.

      Not the cluster-level threshold: `clusterwise_threshold_value` is the p or q
      a whole cluster had to beat, and `cluster_extent_threshold` the size it had
      to reach. A paper reporting "p < 0.001 uncorrected, cluster-level p < 0.05 FWE"
      states this field and `clusterwise_threshold_value`, not this field twice.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: float
    minimum_value: 0
  height_threshold_type:
    name: height_threshold_type
    description: Meaning of height_threshold_value, such as p, q, t, z, or F.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: string
  cluster_extent_threshold:
    name: cluster_extent_threshold
    description: Minimum surviving cluster extent in voxels, vertices, or the unit
      stated in the source.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: integer
    minimum_value: 0
  clusterwise_threshold_value:
    name: clusterwise_threshold_value
    description: Numeric cluster-level threshold for retaining clusters.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: float
    minimum_value: 0
  multiple_comparison_method:
    name: multiple_comparison_method
    description: Correction method, such as FWE, FDR, Bonferroni, permutation, Monte
      Carlo, or none.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: string
  correction_scope:
    name: correction_scope
    description: 'Whether the multiple-comparison correction searched the whole acquired
      volume or was restricted to regions. The correction''s own domain, not the analysis''s:
      a whole-brain voxelwise model whose significance was corrected inside an anatomical
      mask has `Analysis.spatial_scope: whole_brain` and `correction_scope: roi`,
      and the two differing is the ordinary case rather than a contradiction.

      A domain the enum cannot name is written here as free text. An inference level
      such as "cluster level" is not a scope and belongs in `inference_level`.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: string
    any_of:
    - range: SpatialScope
    - range: string
  correction_regions:
    name: correction_regions
    description: 'The regions correction was restricted to -- what `correction_scope:
      roi` names, as `Analysis.regions` names what `spatial_scope: roi` restricts
      the search space to. Empty for a whole-brain correction.

      Not the analysis''s search space. Naming the correction mask on the analysis
      instead says the model was fitted within it, which is a different claim about
      what the paper did.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: Region
    multivalued: true
    inlined: false
  search_volume:
    name: search_volume
    description: Reported search volume, mask size, or number of vertices or voxels
      used for inference.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: string
  neighborhood_definition:
    name: neighborhood_definition
    description: Voxel or vertex adjacency or connectivity rule used for cluster formation.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: string
  tfce_used:
    name: tfce_used
    description: Whether threshold-free cluster enhancement was used.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: boolean
  tfce_parameters:
    name: tfce_parameters
    description: TFCE parameters such as height and extent exponents when reported.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: string
  permutation_count:
    name: permutation_count
    description: Number of permutations, randomizations, or simulations used for inference.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: integer
    minimum_value: 0
  alpha_level:
    name: alpha_level
    description: Family-wise, false-discovery, or nominal alpha threshold when reported.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: float
    minimum_value: 0
    maximum_value: 1
  number_of_tests:
    name: number_of_tests
    description: Number of statistical tests considered by the multiple-comparison
      procedure.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis
    rank: 1000
    owner: InferenceSettings
    domain_of:
    - InferenceSettings
    range: integer
    minimum_value: 0
rules:
- preconditions:
    slot_conditions:
      correction_scope:
        name: correction_scope
        any_of:
        - equals_string: whole_brain
        - equals_string: searchlight
  postconditions:
    slot_conditions:
      correction_regions:
        name: correction_regions
        value_presence: ABSENT
  description: Whole-brain and searchlight correction restricts to no region.
- preconditions:
    slot_conditions:
      correction_scope:
        name: correction_scope
        equals_string: roi
  postconditions:
    slot_conditions:
      correction_regions:
        name: correction_regions
        value_presence: PRESENT
  description: Correction over an ROI must name the regions it was restricted to.