Skip to content

Latest commit

 

History

History
62 lines (43 loc) · 3.8 KB

README.md

File metadata and controls

62 lines (43 loc) · 3.8 KB

Back to Projects List

Updating Batch Anonymizer

SlicerBatchAnonymize is a Slicer Extension that strips off metadata from dicom files, and converts them to various file formats. The work during project week will involve investigating and creating prototypes for defacing in medical images, support and single file dicom export.

Key Investigators

  • Hina Shah (UNC Chapel Hill)
  • Juan Carolos Prieto (UNC Chapel Hill)
  • Jonas Boanchi (University of Michigan)
  • Lucia Cevidanes (University of Michigan)

Project Description

The very first step to make any medical data available to research community is it's anonymization. SlicerBatchAnonymize is a 3D Slicer extension to anonymize a batch of DICOM images by stripping most of metadata (image information stays intact). The tool currently provies a user-friendly UI, supports export to several popular research formats including DICOM series, and also generates a crosswalk files for future uses.

Objective

  1. Add support for exporting CBCT images to a single DICOM file.
  2. Add support for keeping certain metadata fields (example: age and gender) intact during anonymization process
  3. Improve current defacing algorithm

Approach and Plan

  1. CBCT export to dingle file DICOM images will need some exploration into DICOM standards to be careful that correct modailities are assigned correct SOP IDs. Ask experts what is the right way to convert a multi-file DICOm images to a single file.
  2. Will be using inspiration from existing metadata anonymization tools to implement "selective" metadata stripping, with initial options of keeping gender and age intact. This is per the request of our clinicians who will be the primary users of this tool.
  3. Current defacing approach creates noise in the back of the head, and is not robust to intensity changes. We'll work on implementing frontal region detection, and make the algorithm robust to intensities. Community is welcome to add their own/other standard defacing algorithms in the

Progress and Next Steps

  • Suupert added for keeping age and gender intact during anonymization.
  • Creating a summary report.
  • Plans for CBCT anonymization created. The defacing will be evaluated using visual inspection and a survey by clinicians.
    • Find the frontal face, and run anonymization on just that part.
    • Make defacing robust to intensity changes through normalization.
    • Retrain AMASSS. and consider adding more anatomical structures

Illustrations

image SlicerBatchAnonymizeScreenshot

image

CBCT Defacing pipeline

image

Examples of CBCT defacint screenshots for evaluation

Slicer Extension link:

SlicerBatchAnonymize Slicer extension

SlicerBatchAnonymize tutorial video

DICOM standard guidelines for multi-frame volume generation