Transcript Defense Presentation PowerPoint Slide, 2000

Development of a Web-based,
Multimedia Database for
Collection, Organization and
Analysis of Biomedical Signals
M.S.C.S. Problem Report Defense
Lan Guo
Advisor:
Dr. Stephanie Schuckers
Committee: Dr. Wils Cooley
Dr. Bojan Cukic
Dr. George Trapp
Computer Science and Electrical Engineering
West Virginia University
Morgantown, WV
Background: Biomedical Signal (ECG)
with an Event (VF)
VF
Normal
Normal
Biomedical Signal Analysis
Prediction of Sudden Cardiac Death
 Continuous-time
monitoring
 Analysis of large time-series datasets
 Data mining, Oracle
 Prediction and prevention
Collection of Biomedical Signals
Surface ECG
5
0
-5
Intracardiac ECG
10
0
-10
Surface ECG
5
0
Storage and Processing of Signals
-5
Intracardiac ECG
10
0
-10
CSEE
CEMR
0
2
4
6
8
10
12
time (second)
14
16
Example of Biomedical Signals
18
20
Objectives
 To
build a web-based database which provides
efficient management on data collected by
Biomedical Signal Analysis Laboratory
 To
provide these signals to students through webbased front-end to facilitate application-oriented
class projects
 To
provide these signals to the public for outreach
purposes.
 To
utilize database structure for analysis of signals
-- temporal data mining
Project Outline
 Redesign
and implement the back-end
database to incorporate other biomedical
signal besides ECG signal
 Recreate a user-friendly web-based interface
for data insert and data query
 Develop a secure FTP method which shields
the HP9000 data server from public access
 Develop an Oracle/Java interface to
dynamically display signal on-line
Project Overview
 Three
tiered web-based database application
HP9000
Data Server
Java
Applet Servlet
Database link
Oracle 8i
Database
Servlet
WebDB
Client
Client
Client
J
D
B
C
Web
Server
Windows
NT
Features of Back-end
Includes several Large Object (LOB) types,
BFILE, which point to biomedical signal file
and associated files
 Provides relevant information regarding the
biomedical signals and corresponding signal
processing data
 Includes mapping of events within a file and
relationship of events to similar events in other
files
 A small database managing a terabyte of data
on HP 9000 server

Back-end Implementation (1)
 Data
structure was redesigned to incorporate
other biomedical signals besides ECG signal
 Tables were recreated to enable users to
query both multiple data files and
multiplexed data files which contain multiple
channels
 Event tables were modified for querying
same events in multiple locations within a
data file
Back-end Implementation (2)
 Views
were created for different user groups
to query database based on given privileges
 Views were created for joining all tables for
displaying all data associated with a specific
biomedical signal
 Indexes were created for primary key, foreign
key, and unique constraint
 Synonyms were created for each table
 Sequences were created for each primary key
Table Layout of ECG Signal Database
by Chuanyin Dong 09/06/1999
Signal
Other_Signal
UID
Type_of_Signal
ID
Type_of_Signal
Collector
Subject
Collector_ID
Last_Name
First_Name
Status
Phone
e-mail
Subject_ID
Last_Name
First_Name
Type
Age
Sex
Description
Location
Location_ID
Address1
Address2
City
State
Zip_Code
Collector_Location
Location_ID
Collector_ID
ECG_Signal
ECG_ID
Collector_ID
Subject_ID
Property_ID
Start_Date
End_Date
Channel_ID
Instrumentation
Way_to_Collect
Format
Access_level
Memo
ECG_File
Event_location
Event_ID
ECG_ID
Location
Description
Event
Event_ID
Event_Name
Beat_File
Class
Access_level
Description
Surface_Channel
Surface_Channel_ID
Channel_ID
Surface_Channel_Name
Property
Subject_Location
Subject_ID
Location_ID
Property_ID
Length_Time
S/N_Ration
Sample_Rate
Input_Range
Resolution
Amplification
Channel
Channel_ID
Num_Channel
Description
IC_Channel
IC_Channel_ID
Channel_ID
IC_Channel_Name
Anode
Cathode
Description
Features of Front-end(1)
A
front-end web site was developed in Oracle
WebDB
 The web server is a Windows NT
 Users are categorized as public, student, and
lab employee with different access privileges
to database
 Provides query capability and user-friendly
data insert interface including BFILE insert
Features of Front-end(2)
 Client
can download signal data files from
query result by secure FTP method
 Client can view the ECG Signal on-line
based on query result
 Client can view all data stored in the
database associated with a specific
biomedical signal to which he/she has access
Front-end Implementation(1)
 Oracle
8i database biomed was created on the
web server
 Oracle WebDB was installed into the biomed
database
 Three Data Access Descriptors (DAD) were
created to manage three different user groups
 Different accounts were created for different
user groups with corresponding privileges
Front-end Implementation(2)
A
database link and views were created for
transferring data between the data server and web
server
 PL/SQL procedures were developed and WebDB
was explored for developing a web-based
interface
 Java Applets were upgraded to receive parameters
 Java Servlets with JDBC were created as an
Oracle/Java interface for secure FTP and dynamic
signal displaying
Interface to Manage User Groups

Restrict access to
interface for
student and lab
employee by
checking login
information
Interface to Insert Data
Provides a userfriendly interface
to insert data,
including BFILE
 Data referential
integrity is
reinforced by
displaying
foreign keys as
LOV, which
dynamically
select the current
entry being
referenced

Interface to Query Database
Can query database
fields by different
condition, such as =,
>, <, in, not in, etc.
 Can dynamically
query database on
various criteria
 Can select the
output format,
including file
format and report
format

Interface to View Query Result

Can view the query
result and choose
biomedical signal
for further
information and
operations
Interface to Perform Various Operations
Can choose to
download signal
files associated with
a specific
biomedical signal
 Can choose to view
biomedical signals
in all channels for a
specific signal
record
 Can choose to view
all information in
database regarding a
specific biomedical
signal

Download Signal File
Can choose channel
to download
corresponding
signal file
 Can only download
files from the web
server without
knowing the actual
data source

Interface to View Signal Data
Can select channel
to view signal at
different speed and
. different amplitude
 Can manually
measure signal at
both time and
amplitude direction
 Information is
dynamically passed
to Java Applet from
PL/SQL front-end
interface by Servlet

Conclusions (1)
This database project provides an efficient means to
organize the biomedical signal files and related
information
The web-based database system built in this project
provides a seamless mechanism to search and view
the information, and to view the signal on-line
The database provides information about particular
medical events which can occur in multiple locations
within a file and in different files
Conclusions (2)
The framework of organization will allow analysis
of biomedical signal for research conducted in the
Biomedical Signal Analysis Laboratory
The web-based database project supports on-line
class teaching in the area of biomedical signal
processing
The web-based database project provides secure
public access to a portion of the signal database
Future Work
 Link
events that occur during one signal file with
the graphical interface
 Upgrade Java Applet to display more than two
channels at the same time for DSI format
 Upgrade Java Applet to display data file of size
greater than 1 MB for ASCII format
 Develop automatic insert routines which both scan
the data for relevant information, interpret, and
insert into the database
 Apply temporal data mining techniques to further
research
Acknowledgements
Dr. Wils Cooley
 Dr. Bojan Cukic
 Dr. Stephanie Schuckers
(Chair)
 Dr. George Trapp

Dr. John Atkins
 Alan Butcher
 Dr. Juggy Jagannathan

Qunying Fan
 Wei Huang


Members in Dr. Schuckers’ lab