---
title: "DICOM Anonymization & HIPAA De-identification"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{DICOM Anonymization & HIPAA De-identification}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

## Overview

`dicomtk` implements HIPAA Safe Harbor and the **DICOM PS3.15 Basic Application Level Confidentiality Profile** via `dicom_anonymize()`. Unlike legacy wrappers that invoke external `dcmodify` subprocesses, `dicomtk` performs de-identification natively in C++17 memory.

```{r anon-workflow}
library(dicomtk)
library(dplyr)

# Create a synthetic DICOM file containing PHI tags
mat <- matrix(seq(0, 255), nrow = 16, ncol = 16)
raw_dcm <- tempfile(fileext = ".dcm")
write_dicom(
  mat,
  raw_dcm,
  metadata = list(
    PatientName = "DOE^JOHN",
    PatientID = "MRN-987654",
    PatientBirthDate = "19800415",
    StudyDate = "20251016",
    InstitutionName = "GENERAL HOSPITAL",
    Modality = "CT"
  )
)

# Inspect original PHI tags
dicom_tag_value(
  raw_dcm,
  c("PatientName", "PatientID", "PatientBirthDate", "StudyDate", "InstitutionName")
)

# Anonymize with longitudinal date shifting (-100 days), UID regeneration, and private tag removal
anon_dcm <- dicom_anonymize(
  raw_dcm,
  patient_name = "TRIAL^SUBJECT_001",
  patient_id = "SUBJ_001",
  shift_days = -100L,
  erase_private = TRUE,
  regenerate_uids = TRUE
)

# Verify de-identification
dicom_tag_value(
  anon_dcm,
  c("PatientName", "PatientID", "StudyDate", "InstitutionName", "PatientIdentityRemoved")
)
```