Skip to content

Class: Annotations

A set of annotations for an AI-ready dataset

URI: bia_mifa_models:Annotations

classDiagram class Annotations AuthorCollection <|-- Annotations FileLevelMetadataCollection <|-- Annotations Annotations : annotation_confidence_level Annotations : annotation_coverage Annotations : annotation_criteria Annotations : annotation_method Annotations : annotation_overview Annotations : annotation_type Annotations --> AnnotationType : annotation_type Annotations : authors Annotations --> Author : authors Annotations : file_metadata Annotations --> FileLevelMetadata : file_metadata

Inheritance

Slots

Name Cardinality and Range Description Inheritance
annotation_overview 1..1
String		 Short descriptive summary indicating the type of annotation and how it was ge... direct
annotation_type 1..*
AnnotationType		 Annotation type, for example class labels, segmentation masks, counts direct
annotation_method 1..1
String		 Description of how the annotations were created direct
annotation_criteria 0..1
String		 Rules used to generate annotations direct
annotation_coverage 0..1
String		 Which images from the dataset were annotated, and what percentage of the data... direct
annotation_confidence_level 0..1
String		 Confidence on annotation accuracy direct
authors 0..*
Author		 a collection of the authors of a study AuthorCollection
file_metadata 0..*
FileLevelMetadata		 a collection of the file level metadata for each annotation FileLevelMetadataCollection

Identifier and Mapping Information

Schema Source

  • from schema: https://w3id.org/BioImage-Archive/bia-mifa-models

Mappings

Mapping Type Mapped Value
self bia_mifa_models:Annotations
native bia_mifa_models:Annotations

LinkML Source

Direct

name: Annotations
description: A set of annotations for an AI-ready dataset
from_schema: https://w3id.org/BioImage-Archive/bia-mifa-models
mixins:
- AuthorCollection
- FileLevelMetadataCollection
slots:
- annotation_overview
- annotation_type
- annotation_method
- annotation_criteria
- annotation_coverage
- annotation_confidence_level
tree_root: true

Induced

name: Annotations
description: A set of annotations for an AI-ready dataset
from_schema: https://w3id.org/BioImage-Archive/bia-mifa-models
mixins:
- AuthorCollection
- FileLevelMetadataCollection
attributes:
  annotation_overview:
    name: annotation_overview
    description: Short descriptive summary indicating the type of annotation and how
      it was generated
    examples:
    - value: Segmentation masks of human cell nuclei curated by experts from a model
        prediction
    - value: Manual quantification of the number of cells in mouse embryos performed
        by experts
    from_schema: https://w3id.org/BioImage-Archive/bia-mifa-models
    rank: 1000
    alias: annotation_overview
    owner: Annotations
    domain_of:
    - Annotations
    range: string
    required: true
  annotation_type:
    name: annotation_type
    description: Annotation type, for example class labels, segmentation masks, counts...
    examples:
    - value: '[''segmentation_mask'', ''class_labels'']'
    - value: counts
    from_schema: https://w3id.org/BioImage-Archive/bia-mifa-models
    rank: 1000
    multivalued: true
    alias: annotation_type
    owner: Annotations
    domain_of:
    - FileLevelMetadata
    - Annotations
    range: AnnotationType
    required: true
  annotation_method:
    name: annotation_method
    description: Description of how the annotations were created. For example, were
      the annotations crowdsourced or expertly annotated,  produced by  a human or
      software, what software was used, when were the annotations created,  protocols
      used for consensus and quality assurance
    examples:
    - value: A machine learning-based framework was utilized to produce a first coarse
        annotation of nuclear contours.  Then, a disease expert curated all annotations
        by editing, adding, or removing polygons.  Finally, an expert pathologist
        revised all image annotations and a final version of the annotations was curated.
    - value: Cell counts were performed manually by expert biologists. For cases difficult
        to quantify,  a vote including all experts was taken and annotations were
        corrected according to the majority view.
    from_schema: https://w3id.org/BioImage-Archive/bia-mifa-models
    rank: 1000
    alias: annotation_method
    owner: Annotations
    domain_of:
    - Annotations
    range: string
    required: true
  annotation_criteria:
    name: annotation_criteria
    description: Rules used to generate annotations
    examples:
    - value: Only nuclei in focus were annotated, if parts of a nucleus were out of
        focus, only the part of the nucleus being in focus was annotated
    - value: During cell counting, cell division was tracked using anillin localization.
        Mitosis was considered completed when anillin stop localising  at the contractile
        ring and re-entered the cell nucleus, at which point the dividing cell was
        considered two different cells.
    from_schema: https://w3id.org/BioImage-Archive/bia-mifa-models
    rank: 1000
    alias: annotation_criteria
    owner: Annotations
    domain_of:
    - Annotations
    range: string
  annotation_coverage:
    name: annotation_coverage
    description: Which images from the dataset were annotated, and what percentage
      of the data has been annotated from what is available
    examples:
    - value: All images were annotated
    - value: 'Images: image_1, image_2 and image_3 were not annotated due to the abnormal
        cell death rates compared to other samples.'
    from_schema: https://w3id.org/BioImage-Archive/bia-mifa-models
    rank: 1000
    alias: annotation_coverage
    owner: Annotations
    domain_of:
    - Annotations
    range: string
  annotation_confidence_level:
    name: annotation_confidence_level
    description: Confidence on annotation accuracy
    examples:
    - value: Curators had 15 and 20 years of experience working in cancer pathology
    - value: To characterize the quality of gold standard segmentation annotations,  we
        calculated the average and standard deviation performance of the three independent
        annotators  relative to the gold truth that was established by merging the
        triplets of manual annotations.  For the whole dataset this metric was 0.904±0.081
    from_schema: https://w3id.org/BioImage-Archive/bia-mifa-models
    rank: 1000
    alias: annotation_confidence_level
    owner: Annotations
    domain_of:
    - Annotations
    range: string
  authors:
    name: authors
    description: a collection of the authors of a study
    from_schema: https://w3id.org/BioImage-Archive/bia-mifa-models
    rank: 1000
    multivalued: true
    alias: authors
    owner: Annotations
    domain_of:
    - AuthorCollection
    range: Author
    inlined: true
    inlined_as_list: true
  file_metadata:
    name: file_metadata
    description: a collection of the file level metadata for each annotation
    from_schema: https://w3id.org/BioImage-Archive/bia-mifa-models
    rank: 1000
    multivalued: true
    alias: file_metadata
    owner: Annotations
    domain_of:
    - FileLevelMetadataCollection
    range: FileLevelMetadata
    inlined: true
    inlined_as_list: true
tree_root: true