NIBIB Health IT Initiatives and Medical Image Sharing

1
NIBIB Health IT Initiativesand Medical Image
Sharing
The 4th US-China Roundtable Conference on
Scientific Data Cooperation March 29-30, 2010

• James Luo Ph.D.
• Program Director, Biomedical Informatics
  Programs
• National Institute of Biomedical Imaging and
  Bioengineering
• National Institutes of Health

2
US Healthcare Expenditures

• US total healthcare expenditures reached 2.3
  trillion in 2008
• 7,681 per person
• 16.2 of total GDP
• Projection it will reach 4.48 trillions (19.3
  of GDP) in 2019
• Source DHHS

3
Health IT ARRA

• A 2005 RAND study projects that the adoption of
  Health IT in healthcare sectors
• A mean annual savings of almost 42 billion in US
• The American Recovery and Reinvestment Act of
  2009 (ARRA) provides over 19 billion stimulus
  funds for the development and adoption of Health
  IT
• 2 billion for ONC to set up the standards and
  meaningful use
• Healthcare Information Technology Standards Panel
  (HITSP)
• IHE, HL7, DICOM, etc.
• Nationwide Health Information Network (NHIN)
• Establishment of Certification Programs for
  Health IT
• Certification Commission for Healthcare IT
  (CCHIT)
• 17 billion incentives for adoption of EHR

4
Benefits of Health IT

• Interoperable Health IT can improve individual
  patient care in numerous ways
• Provide complete and accurate health information
  at the point of care.
• Allow secure exchange between patients and
  providers.
• Allow more informed decision making to enhance
  the quality and reliability, while reducing
  errors
• Provide increased efficiencies in care and
  administration
• Reduce unnecessary or repetitive tests.
• Improve population health.
• Integrated EHR systems with image, genomics,
  pharmacogenomics (PGx) data and AHRQ clinical
  guidelines support evidence-based clinical
  decision and personalized medicine

5
Kaiser Case Study

• A Kaiser study published in Health Affairs showed
  that using EHR in 2004 to 2007
• office visit rate in Kaisers Hawaii region
  dropped 26.2 percent.
• phone visits increased more than 8 fold,
• online messaging rose nearly 600 percent.
• In 2007
• office visits 66 (compare to
  100 in 2004)
• phone visits 30
• online consultations 4
• The use of EHR and better connectivity with
  patients (phone, online) has made Kaiser more
  efficient

6
Kaiser Case Study 2

• With complete patient information available to
  them in the EHR, physicians can respond to
  patients questions about minor problems without
  seeing them.
• These modes of patient contact dont lower either
  patient satisfaction or the quality of care.
• it reduces errors.
• Integrated EHR systems reap the benefits due to
  increased efficiency, reducing office visits,
  avoiding redundant tests and prescriptions

7
Image in Health IT

• Medical images play a critical role in
• diagnosis and prognosis of diseases
• therapeutic planning
• medical decision-making, safety assessment, and
  risk management
• clinical research to discover effective
  technologies, therapeutics, diagnostics, and
  prevention strategies for different populations
• tracking specific diseases and response to drug
• analyses the effectiveness of therapeutic
• Important part of electronic health records (EHR)

8
NIBIBs Initiatives in Health IT and Clinical
Image Sharing

• NIBIB launched its Health IT and clinical image
  sharing program in 2009
• NIBIBRSNA research project develop a network
  for patient-controlled medical image sharing
  built upon the IHE (Integrating the Healthcare
  Enterprise) HITSP and ONC accepted standards.
• Allow image sharing across RHIOs UCSF, U.
  Maryland, Mayo, U. Chicago, Mount Sinai.
• NIBIB awarded Grand Opportunity (GO) grants in
  clinical image sharing.
• Address image sharing in RHIOs
• U. Alabama, Birmingham
• Wake Forest U.

9
Objectives of Clinical Image Sharing

• To enable the sharing of radiology images across
  health care institutions and vendor systems.
• To aim toward increasing the speed and accuracy
  of data on which medical decisions are based,
• To reduce imaging redundancy and overutilization.
• To improve the quality of patient care by making
  images immediately available.
• A key feature patients control the access to and
  sharing of the images, e.g.
• Consumer based control and ownership of their
  imaging exams through Personal Health Records
  (PHRs)
• Rural, underserved populations or academic
  patient care environment image sharing is
  encouraged.

10
T

• T
• A
• T
• A
• T
• A

11
T

• T
• A
• T
• A
• T
• A

12
Integrating Data, Models, and Reasoning in
Critical Care

• Challenges
• data overload, false alarms, poor data
  organization in the ICU
• early warning signs often difficult to recognize

• Opportunity
• Richness of ICU data makes possible advanced
  monitoring systems to track and predict
  pathophysiologic state of patients.

Roger Mark, Massachusetts Institute Technology,
R01 EB001659 (BRP)
13
Roger Mark, Massachusetts Institute Technology,
RO1-EB001659 (BRP)
Nursing notes and discharge summaries are
de-identified automatically for the research
database.
Techniques to assess signal quality

• Predictive alerts for impending hemo-dynamic
  instability

14
MIMIC II Database Multi-parameter Intelligent
Monitoring for Intensive Care

• a massive research-enabling database
• supports development and evaluation of advanced
  patient monitoring systems
• contents 30,000 patient records 4,000 include
  waveforms
• data includes physiologic trends discharge
  summaries nurses notes IV meds physician
  orders lab reports ventilator settings etc.

De-identified database is made freely available
to research community via PhysioNet
(www.physionet.org)
Roger Mark, Massachusetts Institute Technology,
R01 EB001659 (BRP)
15
PhysioNet the research resource for complex
physiologic signals
16
Design of the PhysioNet Website
17
What is PhysioBank?
PhysioBank currently includes gt40 collections
of cardiopulmonary, neural, and other biomedical
signals from healthy subjects and patients with a
variety of conditions with major public health
implications, including sudden cardiac death,
congestive heart failure, epilepsy, gait
disorders, sleep apnea, and aging.
18
Who Uses PhysioNet / Where?

• gt30,000 researchers, students, manufacturers,
  educators, each month
• From all 50 US states and DC
• Users from gt100 other countries

19
Image Data Sharing in Research

• Alzheimer's disease neuroimaging initative (ADNI)
  gt60 million, PPP
• Goal collection of data and samples (800 cases)
• to establish a brain imaging biomarker,
• to identify the best markers for following
  disease progression and monitoring treatment
  response
• Determine the optimum methods for acquiring,
  processing, and distributing images and
  biomarkers in conjunction with clinical and
  neuropsychological data.
• Validate imaging and biomarker data by
  correlating with neuropsychological and clinical
  data.
• Provide public access to all data and
  bio-specimens http//www.loni.ucla.edu/ADNI/

20
Hippocampal Atrophy as a Quantitative Trait in a
Genome-Wide Association Study Identifying Novel
Susceptibility Genes for Alzheimers Disease
UC Irvine S. Potkin, G Guffanti, A Lakatos, JA
Turner, F Kruggel, JH Fallon, Other
Contributors AJ Saykin, A Orro, S Lupoli, E
Salvi, M Weiner, F Macciardi, ADNI

• The case-control analysis identified APOE and a
  recent risk gene, TOMM40, at a genome-wide
  significance level of p-value 10-6
• The quantitative trait analysis identified 21
  genes or chromosomal areas with at least one SNP
  with a p-value 10-6.
• Apoptosis, cell cycle impairment and the
  alteration of protein folding and degradation
  through ubiquination are among the candidate
  pathophysiological mechanisms

Adapted from Potkin, Guffanti, et al. (2009)
PLoS ONE 4(8) e6501
21
Shen et al 2010 Overview
QCed genotyping data
FreeSurfer 56 volume or cortical thickness
measures
530,992 SNPs
Baseline MRI Scans
142 QTs
GWAS of Imaging Phenotypes
Strong associations represented by heat maps
VBM 86 GM density measures
Refined modeling of candidate association
GWAS of candidate QT
VBM of candidate SNP
22
Shen et al 2010 Findings

• Whole genome, whole brain ROI analysis
• As expected, SNPs in the APOE and TOMM40 genes
  were confirmed as markers strongly associated
  with multiple brain regions.
• Other top SNPs were proximal to the EPHA4, TP63
  and NXPH1 genes.
• Refined analysis for a candidate SNP
• rs6463843 (flanking NXPH1) was associated with
  reduced global and regional GM density across
  diagnostic groups in TT relative to GG
  homozygotes.
• Interaction analysis indicated that AD patients
  homozygous for the T allele showed differential
  vulnerability to right hippocampal GM density
  loss.
• NXPH1 codes for a protein implicated in promotion
  of adhesion between dendrites and axons, a key
  factor in synaptic integrity, the loss of which
  is a hallmark of AD.
• A genome wide, whole brain search strategy has
  the potential to reveal novel candidate genes and
  loci warranting further investigation and
  replication.

23
ADNI Genetics UCLA, Thompson Lab
Voxelwise GWAS Ran genome-wide association for a
quarter of a million points across 700 subjects -
new gene discovery method many new SNPs power
calculations for replication (Jason Stein et al,
NeuroImage, in press) GRIN2b, a common glutamate
receptor genetic variant, is associated with
greater temporal lobe atrophy and with AD
NMDA-receptor is a target for memantine therapy
(Jason Stein et al, NeuroImage, in press) FTO,
an obesity risk gene carried by 46 of Europeans,
is associated with 10-15 frontal and occipital
atrophy, and with a 1.7kg weight gain, on
average (April Ho et al, PNAS, in revision)
Stein et al, NeuroImage, in press