Progressive and interactive modes of image transmission: optimized - - PowerPoint PPT Presentation
Progressive and interactive modes of image transmission: optimized - - PowerPoint PPT Presentation
Progressive and interactive modes of image transmission: optimized wavelet-based image representation Artur Przelaskowski Pawe Haasa Dominik Rives Institute of Radioelectronics, Warsaw University of Technology, Poland
11/2/2005 Telemedicine and Multimedia Communication 2
Outline
Context: Teleinformation medical system
(PACS-RIS-telediagnosis-reference database)
Props1: Interactive progression (user
interface)
Props2: Image indexing (reference database) Props3: JPEG2000 as medical (teleradiology)
standard
Props4: Faster encoder (archiving and
transmission)
Props5: Effective encoder (optimized) Conclusions (use it!)
11/2/2005 Telemedicine and Multimedia Communication 3
Context: Teleinformation Radiology System
Laboratory Registration Analyses Room Results Reports System/database Medical Order Medical Issue
Reports Patients Exam. Results Analyses Regist. Stuff
Central Unit (medical center)
Client Client (specialist)
network connections Wi-Fi Indexing Retrieval engine
Reference database
Interactive codec and data transmission protocols
Mobile Client
11/2/2005 Telemedicine and Multimedia Communication 4
Progressive Interactive Internet Codec
Teleconsultings and picture sharing
with other system clients
Multi-platform, system independent
architecture
Progression modes Region of interest (ROI) progression Interactive protocols Inteligent User Interface
11/2/2005 Telemedicine and Multimedia Communication 5
Progressive image data stream decoding
After 0.5% After 2% After 100%
- !"#$!!%
&'(
)*$!!"!$
11/2/2005 Telemedicine and Multimedia Communication 6
Interactive image exchange (JPEG2000 data stream)
selected ROI
flexible XML-based control connection high performance image stream via UDP Simple image transmission protocol (SITP)
$
- +
,+-'.
)*$!!"!$
11/2/2005 Telemedicine and Multimedia Communication 7
Image database indexing and retrieval
Reference diagnostic database Index in wavelet domain User Boxes (JPEG2000 data stream) Fast and precise retrieval Web access
Indexing Retrieval engine
Reference database
11/2/2005 Telemedicine and Multimedia Communication 8
Reference database
76% MMG 80% MR 84% CT Tanimoto, Var, LH, max n Correlaction, ADev, HL, max n Manhattan, Var, HH, max n Images
Profiles
Precision:
11/2/2005 Telemedicine and Multimedia Communication 9
JPEG2000 is medical standard? YES
DICOM:
it allows transmission of images with
improving resolution and quality, which will be extremely useful in teleradiology and in some PACS network environments
11/2/2005 Telemedicine and Multimedia Communication 10
JPEG2000 is medical standard? NO: irreversible compression
scientific community has not come to a
consensus
fear: valuable information might be lost
through compression
DICOM does not, and will never,
“approve” compression schemes for any particular use
11/2/2005 Telemedicine and Multimedia Communication 11
JPEG2000 is medical standard? YES
„avoiding all risks is the risk ” DICOM: professionals make decision
and take responsibility
Litigation risks: regulators rely on the
professions to lead the way
11/2/2005 Telemedicine and Multimedia Communication 12
Faster JPEG2000 encoder
Optimized PCRD: improvement of Jasper C
implementation (JPER)
Solution: Sorting instead of iterative
bisection
Effect: multi-layers coding without additional
time costs
Faster archiving and transmition (reduced
time costs, reduced memory costs, simplicity)
11/2/2005 Telemedicine and Multimedia Communication 13
Basic idea of PCRD in Jasper
11/2/2005 Telemedicine and Multimedia Communication 14
Basic idea of PCRD in JPER
11/2/2005 Telemedicine and Multimedia Communication 15
Experimental Effects
1 2 3 4 5 6 7 1 10 20 30 40 50
number of layers
compression time
J a s p e r JPER
progression
11/2/2005 Telemedicine and Multimedia Communication 16
JPER2000
11/2/2005 Telemedicine and Multimedia Communication 17
Increased compression efficiency
- Reduced bit rates
- Selective compression: diagnostic
quality progression
- Profiles (e.g. for mammograms: 17/11
filter bank, 8 levels of diadic, nz=0.6, layer progression, code-blocks 64x64)
11/2/2005 Telemedicine and Multimedia Communication 18
The results: diagnosis based on JPEG2000 compressed images
0.1bpp 0.05bpp
0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 0.01 0.05 0.10 0.20 0.30 0.40 0.50 0.80 0.90 1.0
TP/(FP+FN+1) bit rate
Quality subjectively: 1bpp
12bpp 0.3bpp
ROC-based detection: 0.1bpp CAD: 0.9bpp mammograms
11/2/2005 Telemedicine and Multimedia Communication 19