Single DICOM file

Processes one individual file. Best for quick checks or single-instance de-identification.

Study (folder)

Processes entire folders. Preserves study consistency (same input UIDs map to same anonymized UIDs) and generates a DICOMDIR.

The application relies on the robust DCMTK (DICOM Toolkit) library, compiled to WebAssembly (WASM). This brings native C++ performance and strict DICOM standard compliance directly to the web.

• Streaming Mode: Used when processing single files. Data is streamed in chunks, allowing the app to process massive DICOM files (e.g., multi-frame imaging or large videos) without crashing your browser tab by loading the entire file into RAM at once.
• Web Workers (Study Mode): Used for processing entire folders. The application spins up a pool of background worker threads. It automatically calculates the optimal number of workers based on your CPU cores and available system memory. Files are processed in parallel, drastically reducing overall processing time.
• Deterministic UID Generation: Consistent linkage across a study is crucial. The app maintains cross-file consistency for generated UIDs (Study Instance UID, Series Instance UID, SOP Instance UID) using a deterministic SHA-256 hashing algorithm. This ensures that multiple files belonging to the same series will still belong to the same series after anonymization, even when processed independently by different worker threads.

Single DICOM File

• Processing utilizes the memory-efficient streaming engine.
• Detailed metrics, including processing time down to the millisecond, original size, and output size, are displayed.

Study (Folder Upload)

• Parallel Processing: Files are automatically batched and sent to the Web Worker pool.
• DICOMDIR Generation: A standard DICOMDIR index file is automatically built at the end of the process to catalog the newly anonymized dataset, making it ready to be burned to a CD or imported into a PACS.
• Real-time Dashboard: A live status board displays processing progress, files completed per second, total throughput (MB/s), ETA, and a count of any failures.

General Options

• Delete private tags: Removes non-standard, vendor-specific tags (odd-numbered groups). These tags are notorious for hiding PHI such as patient names or device serial numbers. (Enabled by default and highly recommended)
• Delete unknown tags: Strips any tag not explicitly defined in the standard DICOM data dictionary.

PS3.15 Advanced Options

• Retain longitudinal temporal information: Keeps dates and times (e.g., Study Date, Content Time) intact. Useful for timeline analysis or tracking disease progression.
• Retain patient characteristics: Keeps generic biometric data like Age, Sex, Size, or Weight, which are often required for research cohorts, while still removing direct identifiers.
• Retain device identity: Keeps scanner manufacturer, model name, and software versions.
• Retain UIDs: Disables the generation of new UIDs, preserving the original identifiers. Warning: Not recommended unless you explicitly need to match anonymized data against a known external database.
• Retain safe private tags: Attempts to keep private tags that are strictly known to be safe, rather than blindly stripping all of them.
• Retain institution identity: Keeps the hospital, clinic name, and institutional addresses.
• Clean graphics / structured content / descriptors: Aggressively strips overlay graphics, embedded structured reports, and descriptive text fields which technicians occasionally use to type in patient names.

Custom Tag Replacements

• Placeholders: Use dynamic placeholders: {patient}, {study}, {series} for consistent pseudo-random IDs.
• Allow custom profile deviation: If a tag is mandated to be deleted (X) or kept (K) by the strict PS3.15 profile, the engine will normally ignore your custom replacement. Checking this box overrides the strict profile and forces the app to apply your custom value.

Error Logging and PHI Safety

When processing files, some might encounter parsing errors and fail to be anonymized. To protect patient privacy (PHI) that might be present in the original filenames or error messages:

• By default, error logs (_errors.txt) are NOT included in the final ZIP archive.
• You can view errors in the browser or manually download them using the "Save to file" buttons in the results summary or the Errors Modal.
• If you must include the error log in the ZIP, you can enable the "Include errors log in ZIP" checkbox in the Study Output settings, but ensure you review it before sharing.

Study Output Modes

1. Direct ZIP to selected path (Highly Recommended): Uses the modern File System Access API. As each file is processed, it is immediately compressed and streamed directly to your hard drive. This bypasses browser storage quotas and RAM limits entirely, making it the only reliable way to process massive datasets (e.g., 10GB+ studies).
2. Browser storage, then ZIP: Temporarily stages the anonymized files inside the browser's hidden internal storage sandbox (Origin Private File System / OPFS). Useful for older browsers but may fail on very large studies due to browser storage limits.

ZIP Codec & Compression

• Native (CompressionStream): Uses the browser's built-in C++ compression engine. Faster and uses an optimized compression level.
• Software (zip.js): Uses a JavaScript-based compression engine. Allows manual adjustment of the ZIP compression level (0-9).
• Use web workers for ZIP: Offloads the heavy ZIP compression workload to background threads.

When you select a folder, the app scans it for an existing DICOMDIR file. If found, a Metadata Review Modal will pop up.

• The app reads the original Study Descriptions and Series Descriptions from the index.
• You can choose to preserve these descriptions in the anonymized output.
• This ensures your anonymized studies still have meaningful, human-readable names when opened in a DICOM viewer, rather than just showing a list of generic IDs.

• Verbosity: Controls the internal log level of the WASM engine (Quiet to Trace). Set to Debug or Trace to see deeply detailed C++ logs in the browser's Developer Tools (F12) console.
• Study Worker Count: If your browser crashes or freezes, try manually reducing the worker count. Lowering it to 1 or 2 uses significantly less RAM.
• Storage Panel: In "Browser storage, then ZIP" mode, the Storage Panel lets you inspect the hidden files the app has staged. Use "Clear App Storage" to free up disk space if a previous run was interrupted.

This application relies on modern web standards. For the best, most stable experience, we strongly recommend using the latest version of:

• Google Chrome, Microsoft Edge, or other Chromium-based browsers. (Full support for WebAssembly, File System Access API, and Hardware Concurrency).
• Mozilla Firefox (Excellent support, though direct ZIP output may require manual permission grants).
• Safari and older browsers (like Internet Explorer) may fall back to slower processing modes and cannot stream ZIP archives directly to disk.