Full Dicom de-identification

Description

This pipeline provides the highest level of anonymization by completely removing all text from both the image and metadata. It is ideal for preparing DICOM files for public sharing, research, or regulatory compliance, ensuring that no traceable information remains.

Comprehensive removal: Eliminates all visible text within images and removes or anonymizes most metadata fields, including patient identifiers, physician details, and hospital information.

Predicted Entities

Live Demo Open in Colab

How to use

dicom_df = spark.read.format("binaryFile").load(dicom_path)

pipeline = PretrainedPipeline("dicom_deid_full_anonymization", "en", "clinical/ocr")

result = pipeline.transform(dicom_df).cache()

val dicom_df = spark.read.format("binaryFile").load(dicom_path)

val pipeline = new PretrainedPipeline("dicom_deid_full_anonymization", "en", "clinical/ocr")

val result = pipeline.transform(dicom_df).cache()

Example

Input:

Output:

Model Information