CTSC:ARRA:Mockup

From NAMIC Wiki
Revision as of 12:58, 16 March 2010 by Pieper (talk | contribs) (→‎PACS)
Jump to: navigation, search
Home < CTSC:ARRA:Mockup

Back to CTSC ARRA Supplement

Goals

The mockup installation allows the development team to conduct experiments outside of the clinical network environment.

WARNING: These services are not secured -- DO NOT upload any protected health information (PHI) or other confidential information to these servers.

Implementation

A set of virtual machine instances on the Amazon Web Services infrastructure allow our work to happen outside of the institutional firewalls. This minimizes the risk security issues on the clinical networks and allows developers to have control over server allocation, firewall rules, IP addresses, and similar issues that would otherwise require interaction with clinical IS staffs.

Servers

CentOS 64 bit linux is the standard OS used in this project (compatible with i2b2 standard usage).

Contact a mi2b2 researcher for username and password information for these servers.

PACS

The dcm4chee software allows implementation of basic PACS functionality.

The mi2b2 project provides instructions for installation on CentOS 64 bit linux.

The server can be accessed at: http://mi2b2.slicer.org:8080/dcm4chee-web/

This server is being populated by phantom datasets and anonymized generic data.

  • IP Address: mi2b2.slicer.org (174.129.38.17)
  • Port: 11112
  • AE Title: MI2B2

These parameters have been tested and are working with OsiriX on Mac OSX. Other DICOM clients should also work.

Access the PACS using dcmtk

Push a set of DICOM files:

storescu -aec MI2B2 mi2b2.slicer.org 11112 dicom/*

To query by patient name:

Create a file patqry.dump containing the following:

# query patient names and IDs
(0008,0052) CS [PATIENT]     # QueryRetrieveLevel
(0010,0010) PN []            # PatientsName
(0010,0020) LO []            # PatientID

Run the following command to create a dicom (binary) file from the ascii description:

dump2dcm patqry.dump patqry.dcm

Use the CFind Service Class User (SCU) with the query file to ask for all patient with names starting with the letter F

findscu -aec MI2B2 -P -k 0010,0010=F* mi2b2.slicer.org 11112 patqry.dcm

As of this writing the query will return four response records:

$ findscu -aec MI2B2 -P -k 0010,0010=F* mi2b2.slicer.org 11112 patqry.dcm | grep PatientsName
(0010,0010) PN [FROGG^KERMIE]                           #  12, 1 PatientsName
(0010,0010) PN [FROG^KERMIT TCH ]                       #  16, 1 PatientsName
(0010,0010) PN [FROG^KERMIT ]                           #  12, 1 PatientsName
(0010,0010) PN [FROG^KERMIT ]                           #  12, 1 PatientsName

i2b2

  • Webclient: An instance of i2b2 is installed at this address: http://i2b2.slicer.org/webclient
  • Workbench:
    • Download and install the workbench from www.i2b2.org.
    • Edit i2b2workbench.properties file and add the following line at the end of the file:
I2b2.2=i2b2demo,REST,http://i2b2.slicer.org:9090/i2b2/rest/PMService/

XNAT

Initially XNAT Central is being used for testing. Optionally, an amazon instance may be set up to run XNAT locally.

See http://xnat.org for more information about XNAT interface, clients, and usage.

Issues and Considerations

  • At each deployment site we will need a catalog of site-specific attributes that are not modeled accurately by this mockup. Possible issues include:
    • Extra security layers and protocols
    • Bandwidth caps to avoid disrupting clinical workflows
    • Accounting for off-line (tape) storage that may require throttling accesses
    • Bugs or other non-standard behavior of vendor PACS installations (or bugs in dcm4chee)
  • Since the data in the mockup is non-clinical it may not accurately reflect the real installations