NA-MIC Project Weeks

Back to Projects List

Standardizing DICOM De-ID with Actionable Rules and Tests

Key Investigators

Dave Dinh (Consultant, Brigham and Women's Hospital, USA)
Deepa Krishnaswamy (Brigham and Women's Hospital, USA)
Matt McCormick (Fideus Labs, USA)
Andras Lasso (Queen's, Canada)
Steve Pieper (Isomics, USA)
Tina Kapur (Brigham and Women's Hospital, USA)

Project Description

DICOM de-identification (De-ID) efforts often overlap in the rules used to process DICOM metadata. This project aims to translate an existing DICOM De-ID standard into a set of actionable rules that can serve both as an implementation guide and as a verifiable audit trail for tools and AI systems. For example, a standard may permit multiple actions for a given metadata field. A baseline reference can recommend a default action, while still allowing users to specify alternative behaviors as needed. The proposed generator will take user input and produce:

A reference list of itemized, actionable rules
A unit and end-to-end test specification that users can apply to their specific use cases

The initial user interface for the generator will be a command-line (CLI) tool. It will operate based on a predefined decision tree and output rules and test specifications in formats suitable for humans, AI systems, and supported programming languages.

In practice, DICOM De-ID outputs can vary depending on modality, imaging protocols, contractual requirements, and application domains. However, there remains a consistent need to validate that outputs conform to a predefined set of rules. This library represents a step forward in programmatically generating verifiable rules for reliable and consistent implementation.

Objective

No response

Approach and Plan

We want to create a well tested CLI based tool that takes in user input and generates the following:

A reference list of itemized, actionable rules
A unit and end-to-end test specification that users can apply to their specific use cases

Progress and Next Steps

Illustrations