A Brief Introduction to DICOM
A Brief Overview of DICOM
When you look at the literature or on the web trying to get information about DICOM, what you find tends to fall into one of two categories. The first category assumes that you are an IT expert and are trying to learn how to implement DICOM in a process that you control. It is extremely technical in nature. This is especially true in the literature. The second category is more common on the web and these websites assume that you either don't need to know much about DICOM or that you already have a technical background in IT and they give you a lot of technical information. I don't find either of these approaches very useful for this class.
So, I'm going to give a brief discussion about DICOM and some of its issues and then I am going to intersperse a somewhat edited version of a website (From: https://www.neologica.it/eng/Tutorial) that does a pretty good job with explained DICOM without too much technical information.
DICOM stands for Digital Imaging and COmmunication in Medicine. It is both a file format standard and a transmission standard for any type of medical imaging, so it is not just limited to Radiologic Sciences. As I mentioned in the introduction to this module, there were three major issues that needed to be overcome in order to make digital imaging of possible. The first two were file size and transmission time. The third was common standards and DICOM is one of those two standards that we use. DICOM also assist with file size and with transmission time issues.
The following is adapted from https://www.neologica.it/eng/Tutorial:
DICOM was born back in year 1993 by initiative of the American College of Radiology (ACR) and of the National Electrical Manufacturers Association (NEMA). It is often referred to as “DICOM 3.0”, as it is an evolution of the previous ACR-NEMA 2.0 standard.
The main purpose of the DICOM standard is to allow cross-vendor interoperability among devices and information systems dealing with digital medical images, as long as all the involved actors comply with the DICOM standard: thanks to DICOM, a CT equipment of vendor “A” shall be able to send an exam to a digital archive of vendor “B”, or a diagnostic workstation of vendor “C” shall be able to query and retrieve information from a server of vendor “D”, etc.
A medical device supporting and implementing the DICOM standard is defined as a DICOM-compliant device. A “DICOM-compliant” device may be an acquisition device (e.g., CR equipment, CT equipment, MR equipment, etc.), or a workstation, or a server, or any other kind of device able to connect to the DICOM network exchange data with other nodes using the DICOM protocol. For this reason, almost all DICOM devices need to have a network interface.
DICOM devices attached to a DICOM network are often referred to also as DICOM nodes or DICOM peers.
Each vendor marketing a DICOM-compliant device is required to provide a DICOM Conformance Statement for that device: the DICOM Conformance Statement is a document having a well-defined format (which is also specified by the DICOM standard itself) describing which DICOM services the device implements, which optional portions of those services are supported by the device, plus some details about the specific implementation of those services by the vendor.
The DICOM network is the data network connecting DICOM-compliant devices within a medical institution or department. Normally, the DICOM network is a standard Local Area Network (LAN). Hence, the typical network interface of DICOM-compliant devices is a common Ethernet interface. This makes it particularly convenient to connect common PCs with dedicated DICOM-compliant software to a DICOM network.
The DICOM standard specifies several images-related services which are useful in the medical imaging workflow. Here is a short list of the most frequently-used DICOM services:
- Verification service: verifies DICOM connectivity between two DICOM nodes.
- Storage service: transfers DICOM images and other related digital data from a DICOM node to another DICOM node.
- Storage Commitment service: allows a DICOM client application to request a commitment to a server (such as CT unit to PACS server) for safekeeping of certain DICOM images and related data.
- Query/Retrieve service: used to query a DICOM archive (e.g., a PACS server) about its content, and to eventually retrieve some portions of that content to another DICOM node.
- Modality Worklist service: matches the work list created from the RIS and stored images.
Also, as of 2002 there is now a web service that allows images to be transmitted and stored on Internet services. You will sometimes see this referred to as DICOMWeb.
Lossless and Lossy Compressions
In the introduction to this module, I had mentioned about the file size for a typical uncompressed CR image. Below, label Table 1, is a current article that gives information about file sizes for a number of modalities. FYI: Bit depth is the number of shades of gray stored as a power of 2 so 16 bit = 216 = 65,536 shades of gray.
Table 1. Typical image dimensions and uncompressed file sizes for common medical imaging modalities.
|
Modality |
Anatomy |
Image Dimensions (x, y, z, t) in Pixels |
Bit Depth |
Uncompressed File Size |
|
Radiography |
Chest |
(2000, 2500, -, -) |
10–16 bits |
10 MB |
|
|
Abdomen |
(512, 512, 500, -) |
12–16 bits |
250 MB |
|
CT |
Brain |
(512, 512, 300, -) |
12-16 bits |
150 MB |
|
|
Heart |
(512, 512, 100, 20) |
12-16 bits |
1 GB |
|
Breast Tomosynthesis |
Breast |
(2457, 1890, 50, -) |
10–16 bits |
0.4 GB |
|
|
Abdomen |
(512, 512, 100, -) |
12–16 bits |
50 MB |
|
MRI |
Brain |
(512, 512, 200, -) |
12–16 bits |
100 MB |
|
|
Heart |
(256, 256, 20, 25) |
12–16 bits |
250 MB |
|
Ultrasound |
Heart |
(512, 512, -, 50)/s |
24 bits (color) |
38 MB/s |
|
PET Brain |
(256, 256, 50, -) |
16 bits |
6 MB |
|
|
|
Heart |
(128, 128, 40, 16) |
16 bits |
1 MB |
|
Digital Pathology |
Cells |
(30,000, 30,000, -, -) |
24 bits (color) |
2.5 GB |
http://www.mdpi.com/2078-2489/8/4/131/pdf
As you can see by the uncompressed file size column, file sizes in medical imaging today are huge! So big, in fact, that they complicate the transmission requirements to make a digital imaging system viable. One way to reduce transmission time of images is to reduce the size of the image. This is commonly done on the Internet. The problem with reducing image size is that you also reduce the amount of information in a given image. Most of you have probably seen images on the Internet where there was a lot of pixelation (large blocks of a single color on an image) that made it hard to see details. This occurs when an image is compressed too much. So, good question to ask is how much information is enough information? If a I is able to discern only objects that are larger than 0.1 mm in size, does it make any sense to have objects as small as 0.01 mm stored in the image? Actually, it does not. So, we could reduce the image size by reducing the amount of detail in the image enough so that we unaware that it has been compressed.
When we talk about DICOM images, there are two types of compression that are used. The first is lossless and the second is lossy. When a lossless compression is used with the DICOM image, the amount of information that is transmitted to the viewing station is exactly the same amount of information that was stored in the PACS. Without getting into too much technical information, a lossless compression is a way of writing a 'shorthand' of the ones and zeros in the image so that they can be reconstructed at the other end without loss of data.
With a lossy compression, we lose data. But, again, it's all an issue of how much data can we lose before it impacts the diagnostic value of an image as seen on a diagnostic workstation. With lossy compression, the uncompressed image in the PACs is reduced in size and transmitted to a workstation word is reconstructed. Below is another table that gives the current recommendations for the maximum amount of compression for various types of DICOM images as published by the Royal College of Radiologist in the UK. Interestingly enough, the American College of Radiologists has never taken a numerical approach to this problem but just states that compression should not interfere with diagnosis.
Table 2. Lossy compression ratio recommendations by the Royal College of Radiologists.
|
Modality |
Compression Ratio |
|
Mammography |
20:1 |
|
Chest Radiography |
10:1 |
|
Skeletal Radiography |
10:1 |
|
Ultrasound |
10:1 |
|
Digital Angiography |
10:1 |
|
CT (all areas) |
5:1 |
|
Magnetic Resonance |
5:1 |
|
Radiotherapy CT |
No lossy compression |
http://www.mdpi.com/2078-2489/8/4/131/pdf
So, there you have it. Pretty much everything you need to know about DICOM to get you through the day as you deal with all the technology around you.