liminfo

DICOM Reference

Free reference guide: DICOM Reference

26 results

About DICOM Reference

The DICOM Reference provides a comprehensive, searchable guide to the Digital Imaging and Communications in Medicine (DICOM) standard, covering essential patient information tags (Patient Name, Patient ID, Birth Date, Sex), study and series identification with the UID hierarchy (Study Instance UID, Series Instance UID, SOP Instance UID), and imaging modality codes (CT, MR, US, CR, PT).

This reference includes image pixel attributes (Rows, Columns, Bits Allocated, Photometric Interpretation, Pixel Data), CT Hounsfield Unit scales, Window Center/Width presets for lung/bone/brain viewing, DICOM network services (C-STORE, C-FIND, C-MOVE), DICOMweb RESTful APIs (WADO-RS, STOW-RS, QIDO-RS), and DICOM structural elements (VR types, Transfer Syntax, SOP Classes).

Designed for radiology IT professionals, PACS administrators, medical imaging developers, and healthcare informaticists, this tool covers PACS architecture (Orthanc, DCM4CHEE, OHIF), de-identification/anonymization techniques for PHI removal, Radiation Therapy objects (RT Structure Set, RT Plan, RT Dose), and programming tools including DCMTK command-line utilities and pydicom for Python.

Key Features

  • Patient information tags: Patient Name (PN VR format), Patient ID, Birth Date, Sex with value representation details
  • Complete UID hierarchy reference: Study Instance UID, Series Instance UID, SOP Instance UID for DICOM object identification
  • Image pixel attributes: Rows, Columns, BitsAllocated, Photometric Interpretation (MONOCHROME1/2, RGB), and Pixel Data encoding
  • CT Hounsfield Unit scale (air=-1000, water=0, bone=1000-3000) and Window Center/Width presets for clinical viewing
  • DICOM network services: C-STORE (image transfer), C-FIND (query), C-MOVE (retrieve), and association workflow
  • DICOMweb RESTful API reference: WADO-RS (retrieve), STOW-RS (store), QIDO-RS (query) with URL patterns
  • DICOM structure: VR data types (AE, CS, DA, DS, LO, PN, SQ, UI, US), Transfer Syntax UIDs, and SOP Class definitions
  • Open-source tools: DCMTK (dcmdump, storescu, findscu), pydicom Python library, PACS systems (Orthanc, DCM4CHEE, OHIF)

Frequently Asked Questions

What is the DICOM UID hierarchy and why does it matter?

DICOM uses a three-level UID hierarchy: Study Instance UID identifies a complete examination (e.g., a CT scan session), Series Instance UID identifies a specific series within a study (e.g., contrast-enhanced vs. non-contrast), and SOP Instance UID uniquely identifies each individual DICOM image file. This hierarchy is fundamental for PACS queries, image routing, and ensuring correct patient-study-series-image relationships.

What is the difference between MONOCHROME1 and MONOCHROME2 Photometric Interpretation?

MONOCHROME1 maps higher pixel values to darker display (used in some X-ray systems), while MONOCHROME2 maps higher pixel values to brighter display (standard for CT and most modern systems). Most DICOM viewers handle both, but incorrect interpretation causes inverted images. RGB is used for color images, and YBR_FULL is used with JPEG compression transfer syntaxes.

How do CT Hounsfield Units (HU) work and what are common Window presets?

CT pixel values are converted to Hounsfield Units using HU = pixel_value * RescaleSlope + RescaleIntercept. The scale ranges from air (-1000 HU) through water (0 HU) to cortical bone (1000-3000 HU). Standard Window Center/Width presets: Lung (C=-600, W=1500), Bone (C=400, W=1500), Brain (C=40, W=80). Window settings control brightness and contrast for optimal tissue visualization.

What are the main DICOM network services and how do they work?

C-STORE transfers images from SCU (Service Class User) to SCP (Service Class Provider) through A-ASSOCIATE, Presentation Context negotiation, C-STORE-RQ/RSP, and A-RELEASE. C-FIND queries PACS for studies matching criteria (Patient Name, Study Date) at PATIENT/STUDY/SERIES/IMAGE levels. C-MOVE requests the SCP to send images to a specified destination via C-STORE. Modern alternatives include DICOMweb RESTful APIs.

How does DICOMweb differ from traditional DICOM networking?

DICOMweb provides RESTful HTTP/HTTPS APIs that are easier to integrate with web applications compared to the traditional DIMSE protocol. WADO-RS (Web Access to DICOM Objects - RESTful Services) retrieves studies/series/instances via GET requests. STOW-RS stores DICOM objects via multipart POST. QIDO-RS queries for DICOM objects using URL query parameters. DICOMweb uses standard HTTP and JSON, making it firewall-friendly and web-developer accessible.

What DICOM tags must be removed for patient de-identification?

At minimum, remove or replace Patient Name (0010,0010), Patient ID (0010,0020), Patient Birth Date (0010,0030), and other PHI tags per DICOM PS3.15 Confidentiality Profile. Additional tags include Accession Number, Referring Physician Name, Institution Name, and any private tags that may contain identifying information. Tools like CTP (Clinical Trial Processor) and dcm4che dcmanon automate this process while preserving image data integrity.

What is a DICOM Transfer Syntax and why is it important?

Transfer Syntax defines how DICOM data is encoded: byte ordering (Little Endian/Big Endian), VR encoding (Implicit/Explicit), and pixel data compression. Implicit VR Little Endian (1.2.840.10008.1.2) is the default. Explicit VR Little Endian (1.2.840.10008.1.2.1) is most common. JPEG 2000 (1.2.840.10008.1.2.4.90) enables lossy/lossless compression. Transfer Syntax negotiation during DICOM association determines which encoding both parties support.

How do I read and manipulate DICOM files with pydicom?

Install pydicom (pip install pydicom), then use pydicom.dcmread("image.dcm") to load a DICOM file. Access tags by keyword (ds.PatientName, ds.StudyDate) or tag number (ds[0x0010,0x0010]). Access pixel data as a NumPy array with ds.pixel_array. For command-line operations, DCMTK provides dcmdump (view tags), storescu (send images), findscu (query PACS), movescu (retrieve images), and dcm2pnm (convert to PNG/JPEG).