Transcript An Overview of the Integration of the UCSF Dept. of

An Overview of the
Integration of the UCSF Dept.
of Radiology Teaching File
with MIRC
Wyatt M. Tellis
University of California San Francisco
Departments of Radiology and
Biological and Medical Informatics
Introduction
• MIRC is a “virtual” community of medical image
libraries that can be indexed and searched
• Libraries can be:
–
–
–
–
–
Teaching files
Research and reference image repositories
Clinical trial images and related data
Textbooks or papers
PowerPoint © and other types of presentations
• This presentation describes the process of
integrating an existing database driven teaching
file with MIRC
2
UCSF Teaching File Overview
• Web accessible
• Database driven
– 4th Dimension© RDBMS
• Integrated with PACS
– Images entered from the PACS display station
– Images automatically converted from DICOM to JPEG
– Reports automatically retrieved from RIS
• In use since 1998 with over 14,300 cases
– Used by all sections (neuroradiology, chest, abdominal
imaging, skeletal, nuclear medicine, etc.)
3
Input Screen on PACS Display
Select key images using PACS “summary series” tool
Demographics
& history filled
in automatically
ACR codes used
for indexing
Additional
content can be
entered
Search Page for Internal Use
5
UCSF Teaching File Display
UCSF TF Within MIRC
User
Query
Service
RSNA
UCSF uses its internal
database to index its
dynamically generated
documents.
Internet
Index
UCSF
Index
Index
Server
Server
Server
MedPix
Query/Storage Service Interactions
 User Submits Search
 Receives Matching Results
Query
Service
User

Broadcast
Query of
All
Storage
Services

User Views
Content Hosted by
Storage Services
Independent MIRC Storage Services
MIRC Storage Service Integration
• UCSF TF acts as MIRC “storage service”
– Integration achieved through MIRC “query interface”
• MIRC “query interface” responsibilities:
–
–
–
–
Receive “MIRCquery” document via HTTP POST
Extract query parameters from “MIRCquery” document
Perform actual query
Generate “MIRCqueryresult” document containing
matching cases
– Send “MIRCqueryresult” document back to query
service
9
MIRC XML Document Schemas
MIRCquery Document
MIRCqueryresult Document
Storage Service Model*
User
Storage Service
XML
Parser
Control
Logic
Query
Service
Web
Server
DB
Request
Handlers
*Courtesy
of Steve Moore
11
Storage Service Components
• Web Server
– 4D contains built-in web server
– Handles network traffic between browser (and
query service) and teaching file server
• Parses incoming HTTP requests
• Sends HTTP response data
• Database
– 4D contains built-in relational database
– Used to store all text data
• Images stored externally with pointers in database
12
Storage Service Components
• Control Logic
– Written in 4D’s scripting language
• Pascal-like syntax
• Uses third party “plug-ins” to extend scripting
language’s capabilities
– Responsible for authenticating users
– Delegates individual HTTP requests to
appropriate “handlers”
13
Storage Service Components
• XML Parser
– “Expat4D” plug-in provides XML support
• 4D does not inherently support XML parsing
• Freely available:
http://www.mitchenall.com/expat4d/
• Based on the open source, event driven, XML parser
“Expat”
– Used to parse MIRCquery document
14
Storage Service Components
• Request Handlers
– Individual “functions” written in 4D’s scripting
language
– Approximately 70 handlers, each dedicated to
processing a specific request:
• Perform query
• Send list of matching cases
• Send images
– Handlers can make direct calls to underlying
database to process a request
15
Storage Service Components
• Request Handlers (cont’d)
– Apply access restrictions
– Each handler generates the output that is sent
as part of the HTTP response:
• HTML
• Binary data
– JPEG
– PDF
• XML (for MIRC)
– New handler (“query interface”) added to
process requests from MIRC query service
16
Query Interface Overview
• Single request handler
– Receives MIRCquery document as input
• Uses Expat4D to parse XML
– Performs query of database using parameters
specified in MIRCquery document
• Translate MIRC’s Boolean query syntax into database
engine’s internal query language
• Map MIRCquery document elements to columns within
the database’s schema
– Generates MIRCqueryresult document which is
sent back to query service
17
Key Integration Issues
• Translating MIRC’s query syntax into
database engine’s internal query language
– Supports Boolean operators: “and” and “or”
– Uses quotes to define phrases
– Uses parenthesis to create complex Boolean
searches
– Sample query:
<MIRCquery>
(alpha | “bravo charlie”) delta
</MIRCquery>
– Reused code from internal search page to
process MIRC queries
• Query syntax is similar to the one already being used
to search teaching file
Key Integration Issues
• Mapping of MIRC XML document elements
to database schema
– 28 elements in MIRCquery document
– MIRC spec allows you to ignore those that can’t
be mapped
• Anonymization of cases
– Patient identifiers extensively used by
radiologists
– Flags added to specify whether case has been
fully anonymized and can be shared with MIRC
19
Conclusion & Future Directions
• Use of well established standards makes
integration with MIRC straightforward
– HTTP
– XML
• MIRC schema extensive enough to
accommodate our teaching file schema
– Schema continues to evolve to meet the
community’s demands
20
Conclusion & Future Directions
• Initially required only three days of
development time with subsequent
refinements over the next couple of years
– Schema changes
– Addition of anonymization support
• Use of the “MIRCdocument” to integrate
teaching file with authoring software
developed at UCSF for creating eCME
courses
21