Dicom Anonymizer

DICOM Anonymizer

A clinical-grade tool for batch anonymization of DICOM medical images, removing patient identifiable information while preserving essential imaging data for research and analysis.

Overview

This skill anonymizes DICOM (Digital Imaging and Communications in Medicine) files by removing or replacing Protected Health Information (PHI) while maintaining the integrity of the medical image data. It supports batch processing of entire directories and generates audit logs for compliance documentation.

Features

Usage

Command Line

# Anonymize a single file
python scripts/main.py --input patient_scan.dcm --output anonymized.dcm
Batch process a directory
python scripts/main.py --input /path/to/dicom/folder/ --output /path/to/output/ --batch
Preserve study relationships with pseudonyms
python scripts/main.py --input scans/ --output clean/ --batch --preserve-studies
Custom anonymization (keep age, remove birth date)
python scripts/main.py --input scan.dcm --output clean.dcm --keep-tags PatientAge

Python API

from scripts.main import DICOMAnonymizer
anonymizer = DICOMAnonymizer(preserve_studies=True)
result = anonymizer.anonymize_file("input.dcm", "output.dcm")
print(f"Tags anonymized: {len(result.anonymized_tags)}")
Batch processing
results = anonymizer.anonymize_directory("input_folder/", "output_folder/")

Parameters

| Parameter | Type | Default | Required | Description | |-----------|------|---------|----------|-------------| | --input, -i | string | - | Yes | Input DICOM file or directory path | | --output, -o | string | - | Yes | Output DICOM file or directory path | | --batch, -b | flag | false | No | Enable batch/directory processing | | --preserve-studies | flag | false | No | Maintain study relationships with pseudonyms | | --keep-tags | string | - | No | Comma-separated list of tags to preserve | | --remove-private | flag | true | No | Remove private/unknown tags | | --audit-log | string | - | No | Path for JSON audit log | | --overwrite | flag | false | No | Overwrite existing output files |

Anonymized DICOM Tags

The following PHI tags are anonymized by default:

Patient Information

Institution & Provider

Study & Series

Device & Acquisition

Output Format

Anonymized DICOM File

Audit Log JSON

{
  "timestamp": "2024-01-15T10:30:00Z",
  "input_file": "/path/to/original.dcm",
  "output_file": "/path/to/anonymized.dcm",
  "original_patient_id_hash": "sha256:abc123...",
  "pseudonym": "ANON_0001",
  "tags_anonymized": [
    {"tag": "(0010,0010)", "attribute": "PatientName", "action": "cleared"},
    {"tag": "(0010,0020)", "attribute": "PatientID", "action": "hashed"},
    {"tag": "(0010,0030)", "attribute": "PatientBirthDate", "action": "cleared"}
  ],
  "statistics": {
    "total_tags_processed": 150,
    "phi_tags_removed": 12,
    "private_tags_removed": 5,
    "image_data_preserved": true
  }
}

Technical Architecture

1. DICOM Loading: Use pydicom to read DICOM files with validation 2. Tag Analysis: Identify and categorize PHI-containing tags 3. Anonymization Engine: Apply configured anonymization rules per tag 4. UID Handling: Regenerate or hash UIDs to maintain/break linkage 5. Private Tag Removal: Strip manufacturer-specific private tags 6. Validation: Verify output is valid DICOM and image data intact 7. Audit Logging: Record all transformations for compliance

Dependencies

See references/requirements.txt for full dependency list.

Limitations & Warnings

⚠️ CRITICAL: This tool is designed as a helper, not a replacement for institutional review.

References

Technical Difficulty: High

Complex DICOM data structures, UID management, regulatory compliance requirements, potential pixel-data PHI.

Risk Assessment

| Risk Indicator | Assessment | Level | |----------------|------------|-------| | Code Execution | Python/R scripts executed locally | Medium | | Network Access | No external API calls | Low | | File System Access | Read input files, write output files | Medium | | Instruction Tampering | Standard prompt guidelines | Low | | Data Exposure | Output files saved to workspace | Low |

Security Checklist

Prerequisites

# Python dependencies
pip install -r requirements.txt

Evaluation Criteria

Success Metrics

Test Cases

Lifecycle Status

Command Line

# Anonymize a single file
python scripts/main.py --input patient_scan.dcm --output anonymized.dcm
Batch process a directory
python scripts/main.py --input /path/to/dicom/folder/ --output /path/to/output/ --batch
Preserve study relationships with pseudonyms
python scripts/main.py --input scans/ --output clean/ --batch --preserve-studies
Custom anonymization (keep age, remove birth date)
python scripts/main.py --input scan.dcm --output clean.dcm --keep-tags PatientAge

Python API

from scripts.main import DICOMAnonymizer
anonymizer = DICOMAnonymizer(preserve_studies=True)
result = anonymizer.anonymize_file("input.dcm", "output.dcm")
print(f"Tags anonymized: {len(result.anonymized_tags)}")
Batch processing
results = anonymizer.anonymize_directory("input_folder/", "output_folder/")

# Python dependencies
pip install -r requirements.txt