tcia-dataset-proposal-skill

name: TCIA Dataset Proposal Skill description: Guide users through proposing a new dataset or analysis results for TCIA, collecting all necessary metadata, and generating a standardized proposal package.

TCIA Dataset Proposal Skill

Overview

This Skill assists users in preparing a formal proposal for publishing a new dataset or analysis results on The Cancer Imaging Archive (TCIA). It mirrors the logic and requirements of the tcia-dataset-proposal.py application.

Workflow

1. Proposal Type Selection

Determine the type of submission:

• New Collection Proposal: For contributing original imaging data not previously on TCIA.
• Analysis Results Proposal: For contributing derived data (segmentations, annotations, radiomics) based on existing TCIA collections.

2. Contact Information

Collect Name and Email for three mandatory Points of Contact (POC):

• Scientific POC: Primary contact for proposal and data collection.
• Technical POC: Person responsible for data transfer.
• Legal POC: Authorized signatory for the Data Submission Agreement (should not be the PI).

3. Dataset Publication Details

• Title: Descriptive title for the dataset.
• Nickname: Short identifier (< 30 characters, alphanumeric and dashes only).
• Authors: List of authors. Action: Encourage providing ORCIDs. Use orcid_helper.py logic to validate and fetch profile details.
• Abstract: Brief overview of the dataset (Max 1,000 characters). Steering: Refer to tcia-cicadas-skill for high-quality abstract generation.

4. Data Collection Details

• Published Elsewhere: Has the data been published? If so, why TCIA?
• Adaptive Fields (New Collection):
  • Primary disease site/location (from permissible_values.json).
  • Histologic diagnosis (from permissible_values.json).
  • Image types (MR, CT, PET, etc.).
  • Supporting data (Clinical, Genomics, Radiation Therapy Plans/Structures, etc.).
  • Software/Related Resources:Standalone question about source code, Jupyter notebooks, web sites or other software.
  • File formats.
  • Modifications prior to submission.
  • Presence of patient faces.
• Adaptive Fields (Analysis Results):
  • TCIA collection(s) analyzed.
  • Types of derived data.
  • Specificity of image records.
  • File formats.

5. Additional Metadata

• Disk Space: Approximate size.
• Time Constraints: Any deadlines for sharing.
• Publications: Related dataset descriptors or derived publications.
• Acknowledgements: Funding or support statements.
• Why TCIA: Motivation for using TCIA.

Output Generation

The goal is to generate a ZIP package containing:

1. Proposal Summary TSV: {nickname}_proposal_summary_{date}.tsv
   • A single-row TSV containing all responses mapped to the labels in tcia-dataset-proposal.py.
2. Investigators TSV: {nickname}_investigators_{date}.tsv
   • A TSV containing investigator metadata (first_name, last_name, person_orcid, organization_name, email).
3. Proposal Summary DOCX: {nickname}_proposal_summary_{date}.docx
   • A formatted document using full-text question labels as headers.
4. Updated PDF Agreement: {nickname}_agreement_updated_{date}.pdf
   • A copy of the TCIA Data Submission Agreement with "Exhibit A" (Page 6) replaced by the dataset abstract.

Logic Reference

For specific implementation details on file generation and validation, refer to:

• tcia-dataset-proposal.py: Main application logic and file generation.
• tcia-remapping-skill/orcid_helper.py: ORCID validation and lookup.
• tcia-remapping-skill/resources/agreement_template.pdf: PDF template.
• tcia-remapping-skill/resources/permissible_values.json: Controlled vocabularies for sites and diagnoses.