The Guidelines for Data Acquisition outline the types of data that are collected and annotated (prepared) for machine learning model development. While the general requirements for data acquisition can be broadly defined, the subset of data used for annotations needs to be more specific. Additional data collected may be utilized for future, yet unspecified developments.
Related Documentation:
- SOP Machine Learning Model Development
- Intended Use
- Guidelines for Data Annotation
- Algorithm Validation Report
Data Collection
Context of Use
| Context of use: | \<e.g., AI-driven decision support in radiology> |
| Pathology: | \<e.g., prostate cancer> |
| Target quantities: | For example, XXX tumor cases with biopsies, XXX benign cases, XXX healthy cases |
| Additional notes: | \<e.g., Prefer images from specific manufacturers> |
| (…) | (…) |
Medical Details
| Patient group: | \<e.g., European male patients> |
| Medical report contents: | \<e.g., Relevant clinical scores, biopsy data> |
Technical Details
| Manufacturer: | \ |
| Exclusions: | For example, non-MRI images or medical reports excluded |
| (…) | (…) |
Annotation Data Specifications
| Inclusion criteria: | For example, male patients in Europe with specific diagnoses, (…) |
| Exclusion criteria: | For example, tumors caused by metastases of other cancers, implants (e.g., for radiation therapy), (…) |