„Personalized Medicine“ –
Challenges & Chances for
Imaging
Konstantin Nikolaou
Abteilung für Diagnostische und Interventionelle Radiologie
„Personalized Medicine“
Challenges & Chances for Imaging
Increase in Workload + Costs of Imaging
 „Value-based Healthcare“
Imaging Biomarker  Quantatitive Imaging
Radiomics  „Big Data“ + Artificial Intelligence
Potential & Innovations in Magnetic Resonance Imaging
Personalized Medicine at the University Hospital Tuebingen
1
Transforming Radiology:
Large (functional) information in relation to „Big Data“
WEB 2.0/CLOUD
PACS
Personal Communication
Information
Business
Imaging
Business
Decision
making
business
1895.........................................1997.................................2025
Courtesy of Maximilian Reiser, LMU Munich
Diagnostic MRT
Diagnostic CT
24000
9200
23000
9000
22000
8800
21000
8600
20000
8400
19000
8200
18000
17000
8000
16000
7800
15000
7600
2008 2009 2010 2011 2012 2013 2014 2015 2016
2008 2009 2010 2011 2012 2013 2014 2015 2016
McDonald, Acad Radiol 2015
Department of Radiology, UKT
2
„Value-based healthcare“
Kashani, Perspective for Managers Newsletter, 2010
Radiology, 2010
Surrogate Imaging, Imaging Biomarker and Radiomics
as an Integral Part in Precision Medicine
Radiology
Modified from: Jakka and Rossbach, The HUGO Journal, 2013
3
Imaging Biomarkers
•
•
can reliably be used to test medical hypotheses, cross the first gap
becoming useful ‘medical research tools’
•
Any new biomarker has to cross the gaps of technical validation,
biological/clinical validation, and cost effectiveness
 then it becomes a ‘clinical decision-making tool’
To generate QUANTITATIVE Imaging Biomarkers….
• …medical images have to be within describable limits of bias and variance.
• …imaging protocols have to be standardized across imaging centers.
• …acquisition and reconstruction standards have to be defined.
Courtesy of G. Krestin, Rotterdam
NATURE REVIEWS | CLINICAL ONCOLOGY, 2016
Inter- and intratumoral tumor heterogeneity
Do we have to biopsy every lesion?
Radiology, 2016
“Multiscale HCC”, eMed BMBF project, University Hospital Tuebingen
“Intratumor heterogeneity can lead to underestimation of the tumor genomics
landscape and may present major challenges to personalized-medicine and
biomarker development.”
New England Journal of Medicine 366;10 nejm.org march 8, 2012
4
Radiomics

Scan
Images  Features
(Advanced) Imaging

Biomarkers
Big data

Prediction
Artificial intelligence
Courtesy of F. Nensa, University of Essen
Artificial Intelligence
Adap+ng to ar+ficial intelligence
•
Radiologists and pathologists need to adapt
•
Artificial Intelligence will (partly) perform the information extraction from imaging data
•
Radiologists/pathologists will transform to…
 Information Specialists:
 information processing
 putting it into clinical context
 communication
5
OK!, but….
Why Tübingen?
Why MRI?
Tübingen: Center for Personalized Medicine
(ZPM)
6
CentraXX
Biobank / Study Mgt. / Research
data integration solution
Research data
Identified an validated User
(pseudonymized web‐view)
XML
HL7
Patient management
(SAP)
Web
Interface
Biobanking
Pathology
by MPI
by MPI
XML
Biobanking
University gynecological
hospital
Clinical Cancer
Registry
GTDS
XML
XML
Biobanking
Neuropathology
Clinical Study center
Courtesy of N. Malek, Chairman, ZPM & Med. Dpt., University Hospital Tuebingen
Innovations in MRI
Milestones of MRI development
1972
1
1975
1980
3
Lauterbur &
Mansfield
1983
1990
2000
2003
2005
First MR images in humans
2010
6
2015
PET-MR
Basics for modern
MRI
7
5
2
First experimental
MR Systems
Noble Price for
Lauterbur und
Mansfield
8
4
MR Fingerprinting
Hyperpolarized MRI
Clinical MR Systems
0,5 Tesla
1,5 Tesla
3 Tesla
7 Tesla
7
with motion
Motion Artifacts
Motivation
w/o motion
Motion Blur
• respiratory motion
• cardiac movement
• patient and organ
movement
Courtesy of Th. Küstner
Dpt. Radiology, UKT
Aliasing
 Need for adaptation and motion correction strategies
Personalized Medicine:
Role of (standardized) Imaging, Quantification & Reporting
Response
Therapy
Relevance
Quantitative /
Structured Data
Reading / Analysis
Image Acquisition
Patient
8
Role of imaging in personalized medicine
Acquire, identify, structure and curate data
• Radiologist as an INFORMATION SPECIALIST
Multidisciplinary research
• clinical partners, IT, Statisticians, data scientists
Apply classifications to new data
 Provide decision support
27.02.2017
Thank you!
Where do we see the MAGNETOM VIDA?
Test
Modified from: Jakka and Rossbach, 2013
9