PDI DVD Preliminary-Evaluation

From IHE Wiki
Jump to: navigation, search
  • Editor DAC. 2007/10/11.

SUMMARY

For the purpose of evaluating the interoperability of DVD for PDI applications, an experiment will be conducted in which a defined set of uncompressed image files will be provided in the form of an ISO 9660 disk image that exceeds the capacity of a CD but otherwise conforms to the existing PDI profile; “master” DVDs will be burned using this image. “Replicator” and “author” participants may burn additional DVDs with the same file system but different media and/or burners, and/or author additional DVDs with their own file system but the same DICOM images, prior to the Connectathon. “Reader” participants will then be challenged at the Connectathon to read the “master” DVDs, as well as any additional DVDs created by other participants.

OBJECTIVE

The primary objective of this experiment is to confirm the hypothesis that with a reasonable choice of DVD burners, readers and media, interoperability can be achieved.

RATIONALE

The limited capacity of uncompressed PDI DICOM CDs has been identified as a problem for larger image sets, leading to the need for spanning, compression, larger media or a combination of these tools. DICOM has already provided media application profiles based on a DVD-ROM readability requirement, without a media technology preference, but there remain concerns about the interoperability of DVD media. Experience would suggest that interoperability is influenced to some extent by the choice of technology (DVD+/-R+/-W) and burner manufacturer but perhaps most significantly by the “quality” of the media itself. In the absence of a market for or supplier of “medical grade” media (DVD or CD), the question arises as to whether or not there is sufficient interoperability using existing DVD burners, DVD readers and consumer grade DVD media from “recognized” vendors (as opposed to dirt-cheap no-name brands).

The DICOM DVD profiles require the support of compression by the reader, but since that requires application support, for the purpose of this experiment, compression will not be tested. The DICOM profiles further require support of the UDF file system but since ISO 9660 support is also required, for the purpose of this experiment an ISO 9660 image will be used for the “master” test material.

METHODS

Participants

Reader” participants will be required demonstrate their ability to read all the images on this media successfully.

Replicator” participants may enhance the scope of the experiment by burning the ISO 9660 disk image on different media and/or using different burner hardware for use to further challenging the “reader” participants to read those.

Authoring” participants may enhance the scope of the experiment by taking the same set of DICOM images (as opposed the file system image) and authoring and burning their own DVDs and further challenging the “reader” participants to read those.

EVALUATION CRITERIA

Replicator and author participants will not be evaluated per se, only reader participants.

Reader – DICOM Content

For each piece of media, reader participants will be evaluated with respect to DICOM images as either:

  • PASS – able to mount the media and “read” all image files on the test media
  • FAIL PARTIALLY – able to mount the media but only able to “read” some of the image files, or reports read errors
  • FAIL COMPLETELY– unable to mount the media, or not able to “read” any of the image files

The definition of “read” in this case is somewhat dependent on the characteristics of each reading application but is intended to mean make all the DICOM images available to the user in the patient browser and either be able to display the image content to the user or import them into the system, depending on the functionality of the system.

Reader – Web Content

For each piece of media, reader participants that support the Web Content option will be evaluated as either:

  • PASS – able to mount the media and display in a web browser all iconic JPEG image files on the test page
  • FAIL PARTIALLY – able to mount the media but only able to display in a web browser some of the iconic JPEG image files on the test page, or reports read errors
  • FAIL COMPLETELY– unable to mount the media, or not able to display any of the iconic JPEG image files on the test page

Reader – Manifest

For each piece of media, reader participants capable of verifying the MD5 hash based manifest file (using either the Windows executable supplied on the media, or a locally installed Unix or GNU md5sum application) will be evaluated as either:

  • PASS – able to mount the media and verify the MD5 hash of all files in the manifest
  • FAIL PARTIALLY – able to mount the media but only able to verify the MD5 hash some of the files in the manifest, or reports read errors
  • FAIL COMPLETELY– unable to mount the media, or not able to verify the MD5 hash of any of the files in the manifest

TEST MATERIAL

Replicator and Author participants should download the ISO 9660 disk image from here: http://idisk.mac.com/dclunie-Public/IHEPDIDVD/ihepdidvd_20071011.iso.bz2


Since the purpose of the experiment is to assess DVD media interoperability rather than application functionality, a large number of uncompressed transverse images for the widely supported CT modality have been chosen, without any additional complications (such as localizers or presentation states or reports or an “excessive” number of images in a single series). Web content is supplied to provide an additional means of evaluating the readability of a large number of files. A manifest containing MD5 hash values is also provided to quickly evaluate the readability of every byte of every file on the entire media.

The master DVDs will consist of:

  • an ISO 9660 Level 1 File system with file version numbers and no Joliet or Rockridge extensions,
  • a root DICOMDIR file (for DICOM PS 3.10 compliance),
  • a root level INDEX.HTM file (for IHE PDI with Web Content option compliance),
  • a root level README.TXT file (for IHE PDI with Web Content option compliance),
  • a root level IHE_PDI directory, containing a single page with a tabular layout of icon images corresponding to all the DICOM images (for IHE PDI with Web Content option compliance),
  • a root level DICOM directory containing only DICOM PS 3.10 images,
  • a root level MANIFEST.MD5 file containing an MD5 hash value for all (other) files in the media
  • a root level MD5.EXE executable for verifying the MANIFEST.MD5 file
  • no other files (and in particular, no viewer and no AUTORUN.INF file).

The DICOM images will consist of a total of 1.28GB of data (uncompressed):

  • a single (de-identified) patient with a single study,
  • multiple (10) series constituting a coronary CT angiogram,
  • CT images only (2,618 images),
  • no private or standard extended attributes.

The DICOMDIR will:

  • reference all of the PS 3.10 images in the DICOM directory,
  • not contain Icon images,
  • contain the additional directory record keys required by the STD-GEN-DVD-JPEG Media Application Profile.

Formally, the media creator will comply as:

  • a DICOM FSC of the PS 3.11 2007 STD-GEN-DVD-JPEG Media Application Profile, using only the CT Image Storage SOP Class in Explicit VR Little Endian Transfer Syntax,
  • an IHE Portable Media Creator of the PDI Profile supporting the Web Content option, with the exception that DVD rather than CD media will be used.

RESULTS

Publication

Results will be tabulated and published disclosing the following:

  • media brand vendor version
  • drive burner vendor and version
  • drive reader vendor and version

The following information will not be published:

  • medical device hardware and software vendor and version