DICOM & Medical Imaging

What is DICOM?

DICOM stands for Digital Imaging and Communications in Medicine. It's the universal standard for medical imaging - the reason a CT scan taken in Tokyo can be viewed by a radiologist in London without compatibility issues.

Think of DICOM as both a file format and a network protocol. It defines how medical images are stored (like JPEG defines photos) and how they're transmitted between devices (like HTTP moves web pages). Every X-ray, MRI, CT scan, and ultrasound you've ever had was likely stored as a DICOM file.

Why DICOM Matters

Before DICOM, medical imaging was chaos. Each vendor used proprietary formats. A GE scanner couldn't talk to a Siemens workstation. Images were trapped in systems, requiring film printouts to share studies between facilities.

DICOM solved this. Now, regardless of manufacturer, medical images flow seamlessly. Radiologists view studies from any scanner. Images transfer between hospitals. AI algorithms process scans from different vendors. This interoperability isn't just convenient - it enables better patient care and medical research at scale.

Understanding DICOM Files

A DICOM file is more than just an image. It's a package containing:

• The image data: The actual pixels of the scan
• Patient information: Name, ID, date of birth
• Study details: What type of scan, which body part, when it was taken
• Technical parameters: Scanner settings, radiation dose, image dimensions
• Interpretation: Sometimes includes measurements or annotations

This bundling is crucial. The image and its context travel together. You never lose track of whose scan you're looking at or what body part it shows.

DICOM Tags

DICOM files contain hundreds of data elements called "tags." Each tag has a number and stores a specific piece of information:

• (0010,0010) = Patient's Name
• (0020,000D) = Study Instance UID (unique identifier)
• (0008,0060) = Modality (CT, MR, XR, etc.)
• (0028,0010) = Rows (image height in pixels)

Software reads these tags to understand what it's looking at. Want to find all chest X-rays for a patient? Search for their patient ID and "XR" modality. Need to route cardiac MRIs to a specialist? Check the body part tag and modality.

The Medical Imaging Ecosystem

Modalities

Modalities are the devices that create images:

• CT (Computed Tomography): Cross-sectional X-ray imaging
• MR (Magnetic Resonance): Detailed soft tissue imaging
• XR (X-Ray): Traditional radiography
• US (Ultrasound): Sound wave imaging
• PT (PET Scan): Metabolic activity imaging

PACS (Picture Archiving and Communication System)

A PACS is the central repository for medical images. It receives images from modalities, stores them long-term, and serves them to workstations for viewing. Think of it as the imaging department's database and file server combined.

When a radiologist opens a study, they're pulling it from the PACS. When studies need to be sent to another facility, they're exported from the PACS. Everything imaging-related flows through the PACS.

Viewers and Workstations

These are the applications radiologists use to examine images. They provide tools for:

• Adjusting brightness and contrast (windowing)
• Measuring distances and areas
• 3D reconstruction from CT/MRI slices
• Comparing current and prior studies
• Creating reports with key images

DICOM Network Communication

DICOM isn't just files - it's also a protocol for moving those files between systems.

DIMSE (DICOM Message Service Element)

The traditional DICOM network protocol. Systems establish peer-to-peer connections and exchange images using DIMSE operations:

• C-STORE: Send an image to another system
• C-FIND: Query for studies matching certain criteria
• C-MOVE: Request images be sent to a third system
• C-GET: Retrieve images directly

DIMSE works, but it's old. It requires complex configuration of Application Entities (AE Titles), IP addresses, and ports. Firewalls complicate things. Each connection must be explicitly configured on both ends.

DICOMweb

The modern approach. DICOMweb uses RESTful APIs over HTTP/HTTPS, making it work like any web service. Instead of specialized network configuration, you use standard HTTP requests:

• WADO (Web Access to DICOM Objects): Retrieve images via URL
• QIDO (Query based on ID for DICOM Objects): Search for studies
• STOW (Store Over the Web): Upload images

DICOMweb makes medical imaging accessible to web developers. You can build a viewer in JavaScript that pulls images from a PACS via HTTPS. No special DICOM networking knowledge required - if you can call a REST API, you can work with DICOMweb.

Real-World Workflow

Here's what happens when a patient gets a CT scan:

1. Order Created: Doctor orders CT scan in EHR, HL7 message sent to imaging system
2. Patient Arrives: Tech verifies patient identity, enters into modality
3. Scan Performed: CT scanner creates images, stores as DICOM files
4. Images Sent: Scanner sends DICOM files to PACS via C-STORE
5. Radiologist Reads: Opens study from PACS using viewing workstation
6. Report Created: Radiologist dictates findings, report goes to EHR
7. Results Delivered: Ordering doctor views images and report

If the patient needs a second opinion, the PACS can send studies to another facility via DIMSE or export them to portable media. The images remain accessible for future comparison - crucial for monitoring disease progression.

Modern Developments

Cloud PACS

Traditional PACS required on-premise servers and expensive storage arrays. Cloud PACS move this infrastructure to AWS, Azure, or GCP. Benefits include:

• No hardware to maintain
• Elastic storage that scales with volume
• Access images from anywhere
• Disaster recovery built in

Cloud PACS typically use DICOMweb for image retrieval, making them web-native. Radiologists access studies through browser-based viewers rather than thick client workstations.

AI and DICOM

AI algorithms for detecting disease in medical images consume DICOM files. The standard metadata helps AI systems:

• Identify what type of scan they're analyzing
• Extract clinical context (patient age, indication)
• Route findings back to appropriate systems
• Track which images were used for training

Zero-Footprint Viewers

New viewers run entirely in web browsers, requiring no software installation. Using DICOMweb and JavaScript, they provide full diagnostic viewing capabilities. This enables:

• Specialists consulting from home
• Emergency reads from mobile devices
• Patient portals with imaging access
• Simplified IT - no workstation management

Challenges and Considerations

File Sizes

Medical images are large. A single CT study might be 500MB. An entire cardiac MRI could be several gigabytes. This impacts:

• Network bandwidth - transferring studies takes time
• Storage costs - years of images accumulate
• Viewer performance - loading large studies requires optimization

Privacy

DICOM files contain patient information. When sharing for research or teaching, images must be de-identified - patient names, IDs, and dates removed. This is tricky because some tags might indirectly identify patients.

Interoperability Issues

Despite being a standard, DICOM implementations vary. Vendors interpret specifications differently. Some use private tags that other systems don't understand. Integration projects often require troubleshooting vendor-specific quirks.

Getting Started

For developers interested in medical imaging:

• Learn the basics with sample DICOM files (many are available online)
• Use a DICOM viewer to explore file structure and tags
• Try libraries like dcm4che (Java), pydicom (Python), or cornerstone.js (JavaScript)
• Experiment with a test PACS like Orthanc (open source)
• Build a simple viewer using DICOMweb APIs

For those evaluating systems:

• Verify DICOM conformance statements from vendors
• Ask about DICOMweb support for future-proofing
• Understand storage requirements and growth projections
• Consider cloud vs on-premise PACS
• Ensure viewers meet specialty needs (cardiology, mammography, etc.)

The Bottom Line

DICOM is the invisible foundation of modern medical imaging. Radiologists, technologists, and clinicians use it daily without thinking about it - which is the mark of a successful standard.

As healthcare moves to the cloud and AI transforms diagnostics, DICOM continues evolving. DICOMweb makes imaging web-accessible. New supplements address advanced modalities. The standard that started in the 1980s remains relevant because it solves a fundamental problem: ensuring medical images can be viewed and understood regardless of when, where, or how they were created.