Image Transfer and Storage

1. Image Transfer and Storage PACS DICOM Image Compression

2. PACS Picture Archival and Communication System means of storing and transferring digital images within the radiology / nuclear medicine department / hospital as opposed to Teleradiology which is the transfer of images outside of the hospital

3. PACS Components image acquisition devices image communication and storage protocols - DICOM computer hardware and software patient / hospital data interface - HIS / RIS image storage and archival devices image reporting and interpretation stations communication network output devices

4. Image Acquisition Devices Digital imaging systems Computed Radiography / Digital Fluoro / DSA CT / MRI Nuclear Medicine Ultrasound Analogue imaging systems require a film scanner to convert to digital data

5. Computer Hardware and Software Interface between image acquisition device and storage device can be an integral part of the acquisition device eg. an image manipulation workstation

6. Image Storage and Archival Devices Archival media: hard drives CD ROMs (WORM) magnetic optical disks (MOD) - write many times RAIDs (Redundant Array of Inexpensive Disks) digital tape

7. Communication Network LAN - Local Area Network Ethernet network Category 3 - �10BaseT � coaxial cable star topology 10Mbps for maximum segment length of 100 meters Category 4 16Mbps Token Ring Category 5 speeds up to 100Mbps eg. through optical fibre

11. WAN - Wide Area Network generally considered to be transfer of data to outside of the hospital can be specific link between sites through fibre optics or microwave (basically an extension of a LAN) more commonly used - connection through phone lines: - either specific data cables - ISDN - Integrated Services Digital Network or "I STILL DON'T know its NAME" or standard phone line this is more commonly called teleradiology

12. Network layers A network can be described at seven layers, often referred to in the business as the OSI network layer model: the seven layers are 1. Physical (eg Ethernet, Token Ring) 2. Media Access (a driver for a network card) 3. Network (e.g. I/P Internet Protocol, SLIP, PPP or X.25) 4. Transport (TCP - Transmission Control Protocol) 5. Session (e.g. Telnet, FTP, SNMP, or SMTP) 6. Presentation (e.g. Remote File Service) 7. Application (e.g. NFS - Network File System)

13. DICOM - What is it? Digital Imaging and COmmunication in Medicine

14. DICOM DICOM is a protocol that all manufacturers follow so as to be able to interchange information and images DICOM defines standards for: image files, specific for each modality transferring of images from one place to another

15. DICOM - What is it? DICOM was created by ACR - NEMA (American College of Radiologists & National Electronic Manufacturers Association) so as to survive the evolving nature of Radiology and to cope with technological advances It is not dependent on the network it works within.

16. Everyone that has more than one modality Everyone that wants their modalities to communicate with each other Everyone that wants to be free of single manufacturer restrictions. DICOM - Who needs it?

23. Normal Radiology DepartmentSame Manufacturers

24. Normal Radiology DepartmentDifferent Manufacturers

25. A Networked Department

26. So Why Do We Need DICOM don�t want to be committed to a single supplier of a network. to be able to easily connect any additional modality without another large expense. want to be confident that each modality will interact with each other.

27. How is DICOM structured DICOM consists of 13 different parts. Each of these parts are a set of rules that defines how each image or piece of information is treated. The core of DICOM are the service classes that describe what we will do with an image.

30. Service Classes Storage To Archive or Transfer Images Query / Retrieve Find Images and Send Them Print management Print formatted images to a laser printer Patient, Study and Results management Interact with HIS and RIS systems

32. We want to STORE an Image from one manufacturers CT to another�s Workstation

33. We will use the Storage Service Class to send the image from one modality to another

34. Which modality is the user and which is the provider?

35. The CT USES the Workstations ability to Store an Image

36. And the Workstation Provides the CT with the ability to Store an Image

37. The CT is the Storage Class User and the Workstation is the Storage Class Provider

38. So what next?

39. The CT sends CT images, so the Workstation must be able to accept CT images. We must define the Image Object

40. The Object is a CT Image and therefore we will need to tell the workstation some facts about the image

41. DICOM Image Format Header - that part of the image that contains information about the image DICOM Image Header patient data patient ID - name, URN, age etc clinical indications examination report modality data eg MRI - TR, TE, scan plan, location, scan time etc, etc, etc.

43. DICOM There is no central body that checks conformance statements. It is each manufacturers responsibility to correctly comply with the standard. DICOM is NOT Plug and Play.

44. DICOM DICOM does answer our interconnection problems DICOM is designed to grow with changes DICOM is supported by all major manufacturers The golden rule in DICOM, is to ask for a conformance statement for the equipment in question. If there is no Conformance Statement, then the equipment is not DICOM

45. Image Compression

46. What is image compression Image compression is reducing the size of the image file, with or without loss of image quality example: image has 1000 x 1000 spatial resolution and 10 bit depth (1024 gray values) gives an image size of 4,000,000 bytes (4,000 MB) � note: 2 bytes are require for 10 bits of data

47. if the image file is reduced to 2,000 MB it has been compressed comprehension ratio = size of original image size of new image in this case 2:1

48. Why do we need compression Large image files size Chest images � 4K x 4K x 10bit Multiple images M.R.I., C.T., digital fluoroscopy, digital cardiac Legal requirements Store images for given number of years

49. Why do we need compression reduce speed of transfer across the PACS network In medical imaging must: conform to DICOM standards compatible with DICOM transfer protocols

50. Types of Compression Lossless and Lossy Lossless no visible loss of image quality generally removes redundant data typical compression ratios of 2:1 to 10:1 Lossy small to large quality loss generally removes irrelevant data typical compression ratios of up to 100:1

51. Lossless Compression Colour images typically are 3 images of red, green & blue can uses a colour palette instead only use the values of the colour actually in the image

52. Run-length coding looks for sequential pixel values eg. 1 row of an image with the new code below have reduced the size from 15 bytes to 6 higher compression ratios when predominantly low frequency information typical compression ratios of 4:1 to 10:1

53. Huffman coding based on a lossless statistical method of the 1950s creates a frequency tree to obtain code

54. most common pixel value (with the highest frequency) has the shortest code table of pixel value versus the code must be sent with each image time of coding and decoding can be long typical compression ratios 2:1 � 3:1 LWZ coding (Lemple, Ziv and Welch) similar to run-length but with some statistical methods similar to Huffman

55. Lossy Compression does not necessarily mean loss of visible image quality Simple methods reduce image spatial resolution 50% reduction in both X & Y directions ? new image size 25% of original obvious loss of quality depending on purpose � can be acceptable

56. reduce image contrast resolution reduce from 10 or 12 bit depth to 8 bit ? new image size � of original if image histogram only spread over 8 bits � no loss of quality (depending on method) if image histogram spread over greater than 8 bit � some loss of quality but if image saved using an appropriate lookup table (brightness and contrast), may not be visible. these manipulation were done in Image Concepts practicals

57. JPEG / JPG Images Joint Photographic Experts Group (JPEG) have established an image file format that has both lossless and lossy compression lossless uses Huffman coding compression ratio upto 25 : 1 lossy uses discrete cosine transforms (DCT) compression ratio of 40 : 1 with little visible loss

59. Wavelet Compression latest method for image compression wavelet transforms were designed to over come time domain problems of Fourier uses a pre-defined �mother wave� as the basis of the transform Wavelet=small waves

60. signal analysis weighted sum of basis functions. Infinitely many possible sets of wavelets coefficients contain information about the signal basis functions Fourier representation reveals information about signal's frequency domain behaviour

61. Wavelet Compression compression levels greatly increased without apparent loss of image quality compression levels upto 200 : 1 new standard for JPEG compression JPEG compression is a standard in DICOM images