DICOM Basics using Java - Transfer Syntax and Compression

Nov 7, 2025

Introduction

This is part of my series of articles on the DICOM standard. In this tutorial, we'll explore DICOM transfer syntaxes, which define how DICOM data is encoded (value representation, byte ordering, and compression). Understanding transfer syntaxes is essential for interoperability and storage optimization.

Choosing the right transfer syntax affects storage size, transmission speed, and viewer compatibility.

Prerequisites

Before you begin, ensure you have the following:

Java JDK installed (Java 8 or later)
PixelMed Java DICOM Toolkit
Understanding of basic DICOM file structure
You can find all the code demonstrated in this tutorial on GitHub here

“Time heals what reason cannot.” ~ Seneca

Common Transfer Syntaxes

Uncompressed:

UID	Name	Notes
1.2.840.10008.1.2	Implicit VR Little Endian	Default, most compatible
1.2.840.10008.1.2.1	Explicit VR Little Endian	Recommended
1.2.840.10008.1.2.2	Explicit VR Big Endian	Retired

Lossless Compression:

UID	Name	Ratio
1.2.840.10008.1.2.5	RLE Lossless	2:1 - 3:1
1.2.840.10008.1.2.4.57	JPEG Lossless (Process 14)	2:1 - 3:1
1.2.840.10008.1.2.4.70	JPEG Lossless SV1 (Recommended)	2:1 - 3:1
1.2.840.10008.1.2.4.90	JPEG 2000 Lossless Only	2:1 - 3:1
1.2.840.10008.1.2.4.80	JPEG-LS Lossless	2:1 - 3:1

Lossy Compression:

UID	Name	Ratio
1.2.840.10008.1.2.4.50	JPEG Baseline (8-bit)	10:1 - 20:1
1.2.840.10008.1.2.4.51	JPEG Extended (12-bit)	10:1 - 20:1
1.2.840.10008.1.2.4.91	JPEG 2000 (lossy)	10:1 - 50:1

Transfer Syntax Conversion

package com.saravanansubramanian.dicom.pixelmedtutorial;

import com.pixelmed.dicom.AttributeList;
import com.pixelmed.dicom.FileMetaInformation;
import com.pixelmed.dicom.TagFromName;
import com.pixelmed.dicom.TransferSyntax;

public class TransferSyntaxConversionDemo {

    public static void main(String[] args) {

        try {

            System.out.println("=== DICOM Transfer Syntax Conversion ===\n");

            String inputFile = "C:\\path\\to\\input.dcm";
            String outputFile = "C:\\path\\to\\output.dcm";

            // Read the DICOM file
            AttributeList list = new AttributeList();
            list.read(inputFile);

            // Get current transfer syntax
            String currentTS = list.get(TagFromName.TransferSyntaxUID)
                .getSingleStringValueOrNull();
            System.out.println("Current: " + currentTS);
            System.out.println("Description: " + getDescription(currentTS));

            // Convert to different transfer syntax
            String targetTS = TransferSyntax.ExplicitVRLittleEndian;
            System.out.println("\nConverting to: " + targetTS);

            // Remove old file meta information
            list.removeMetaInformationHeaderAttributes();

            // Add new file meta information with target transfer syntax
            FileMetaInformation.addFileMetaInformation(
                list, targetTS, "OurSourceAET");

            // Write the file
            list.write(outputFile);

            System.out.println("Conversion complete: " + outputFile);

        } catch (Exception e) {
            e.printStackTrace(System.err);
        }
    }

    public static String getDescription(String tsUID) {
        if (tsUID == null) return "Unknown";

        switch (tsUID) {
            case "1.2.840.10008.1.2":
                return "Implicit VR Little Endian";
            case "1.2.840.10008.1.2.1":
                return "Explicit VR Little Endian";
            case "1.2.840.10008.1.2.5":
                return "RLE Lossless";
            case "1.2.840.10008.1.2.4.50":
                return "JPEG Baseline (Lossy, 8-bit)";
            case "1.2.840.10008.1.2.4.70":
                return "JPEG Lossless SV1 (Recommended)";
            case "1.2.840.10008.1.2.4.90":
                return "JPEG 2000 Lossless Only";
            case "1.2.840.10008.1.2.4.91":
                return "JPEG 2000 (Lossy)";
            default:
                return "Other";
        }
    }
}

Compression and Decompression

// Read DICOM file (PixelMed automatically decompresses)
AttributeList list = new AttributeList();
list.read(inputFile);

// Save with RLE compression (lossless)
saveWithTransferSyntax(list, TransferSyntax.RLE, "compressed_rle.dcm");

// Save with JPEG Lossless
saveWithTransferSyntax(list, TransferSyntax.JPEGLosslessSV1,
    "compressed_jpeg_lossless.dcm");

// Save with JPEG 2000 Lossless
saveWithTransferSyntax(list, TransferSyntax.JPEG2000Lossless,
    "compressed_j2k_lossless.dcm");

// Helper method
private static void saveWithTransferSyntax(AttributeList list,
        String transferSyntax, String outputPath) throws Exception {

    // Clone to avoid modifying original
    AttributeList outputList = (AttributeList) list.clone();

    // Remove existing file meta information
    outputList.removeMetaInformationHeaderAttributes();

    // Add new file meta with target transfer syntax
    FileMetaInformation.addFileMetaInformation(
        outputList, transferSyntax, "OurAET");

    // IMPORTANT: In a real implementation, you must also transcode
    // (compress/decompress) the pixel data to match the target
    // Transfer Syntax. This example only updates the metadata header.

    // Write the file
    outputList.write(outputPath);
}

Note: This simplified example only updates the Transfer Syntax UID metadata. In a production implementation, you must also transcode the pixel data to match the target Transfer Syntax, otherwise the resulting DICOM file will be invalid.

The Theory Behind Transfer Syntaxes

Transfer syntaxes define the encoding rules for DICOM data. Understanding the theoretical foundations helps in making informed decisions about compression, interoperability, and system design.

The Three Components of Transfer Syntax

Every transfer syntax specifies three independent parameters:

Value Representation (VR) Encoding: Implicit (VR must be looked up in dictionary) vs. Explicit (VR embedded in data stream)
Byte Ordering: Little Endian (LSB first, x86 native) vs. Big Endian (MSB first)
Pixel Data Compression: Uncompressed, RLE, JPEG variants, JPEG 2000, etc.

These are independent axes, though not all combinations exist as registered transfer syntaxes. For example, Implicit VR is always Little Endian in DICOM.

Information Theory and Medical Image Compression

Medical images have unique statistical properties that affect compression efficiency:

High Bit Depth: 12-16 bits vs. consumer images' 8 bits, meaning more unique values and lower entropy reduction
Spatial Correlation: Adjacent pixels tend to have similar values (smooth gradients in tissue), exploitable by predictive coding
Modality-Specific Noise: CT quantum noise, MR thermal noise, US speckle all reduce compressibility
Dynamic Range: Medical images often use only a portion of the available bit range, creating opportunities for bit-plane optimization

Why Lossless Compression Matters for Medicine

In consumer imaging, lossy compression is universally accepted because humans can't perceive the differences. Medical imaging is different:

Diagnostic Threshold: A subtle finding might differ from normal by just a few pixel values
Quantitative Imaging: SUV calculations in PET, Hounsfield units in CT require exact values
Legal Evidence: Images may be used in litigation where authenticity matters
Algorithm Input: CAD/AI systems may be sensitive to compression artifacts

However, research has shown that for certain modalities and clinical tasks, carefully controlled lossy compression may be acceptable. The key is validation: any lossy compression scheme must be validated for its specific clinical use case.

The JPEG 2000 Advantage

JPEG 2000 offers significant advantages over older JPEG for medical imaging:

Native Support for 16-bit: No need for 12-bit extended JPEG workarounds
Progressive Transmission: Low-resolution preview available before full image loads
Region of Interest Coding: Encode diagnostically important regions at higher quality
Unified Lossless/Lossy: Same algorithm scales from lossless to high compression
No Blocking Artifacts: Wavelet transform avoids JPEG's characteristic 8x8 blocks

Lossless vs Lossy Compression

Lossless Compression:

Original pixel values preserved exactly
Required for diagnostic interpretation
Typical ratio: 2:1 to 3:1
Examples: RLE, JPEG Lossless, JPEG 2000 Lossless

Lossy Compression:

Some pixel information is lost
Acceptable for: teaching files, thumbnails, web display
Typical ratio: 10:1 to 50:1
Examples: JPEG Baseline, JPEG 2000 Lossy

Compression Algorithm Comparison

Algorithm	Type	Speed	Compatibility	Best For
RLE	Lossless	Fast	Universal	Uniform images
JPEG Lossless SV1	Lossless	Medium	Wide	General use
JPEG 2000 Lossless	Lossless	Slower	Growing	Modern systems
JPEG-LS	Lossless	Fast	Limited	Medical imaging
JPEG Baseline	Lossy	Fast	Universal	Web display
JPEG 2000	Lossy	Slower	Growing	High compression

Choosing the Right Transfer Syntax

Use Case	Recommended Transfer Syntax
Maximum compatibility	Explicit VR Little Endian
Archival storage	JPEG Lossless SV1 or JPEG 2000 Lossless
Newer systems	JPEG 2000 Lossless
Web/mobile display	JPEG Baseline or JPEG 2000 (lossy)
Network transfer	Implicit VR Little Endian

Best Practices

NEVER apply lossy compression to diagnostic images
Keep original uncompressed copies for legal compliance
Use JPEG Lossless SV1 for maximum compatibility
Use JPEG 2000 Lossless for newer systems
Test viewer compatibility before deployment
Consider storage vs bandwidth tradeoffs

Lossy Compression Attributes

When lossy compression is applied, these attributes should be updated:

Tag	Name	Values
(0028,2110)	Lossy Image Compression	”00” = No, “01” = Yes
(0028,2112)	Lossy Image Compression Ratio	e.g., “10” for 10:1
(0028,2114)	Lossy Image Compression Method	Algorithm identifier

Conclusion

DICOM transfer syntaxes provide flexibility in how medical images are encoded and compressed. Choosing the appropriate transfer syntax is critical for balancing storage efficiency, transmission speed, and image quality requirements.

Understanding the differences between lossless and lossy compression, and when each is appropriate, is essential for building compliant DICOM applications that maintain image quality for diagnostic purposes while optimizing storage and bandwidth. In the next tutorial in this series, I will cover DICOM character set handling for international text support. See you then!