2010 • Volume 2, Number 4
In this issue:
Why a Quantitative Imaging Curriculum Should be Included
in Residency Training Programs
M. BOONE, PhD, FAAPM, FSBI, FACR
ANALYSIS TOOLS AND
The Challenges of Making
By JAMES VOYVODIC,
RSNA 2010: Quantitative
Imaging/Imaging Biomarkers and QIBA Meetings and Activities
BIOMARKERS IN THE LITERATURE
Search on Quantitative Imaging in the Residency Curriculum
IN MY OPINION
Why a Quantitative Imaging
Curriculum Should be Included in Residency Training Programs
By JOHN M. BOONE, PhD, FAAPM, FSBI, FACR
today learned their craft in a largely qualitative educational landscape, and for
many clinical settings the differential diagnosis which is the standard
qualitative reporting procedure will remain the heart and soul of the radiology
report. However, the future of radiology reporting will gradually embrace
quantitative metrics, providing critical information in an increasing number of
radiology settings. Therefore, it is essential that residency programs begin to
teach both the necessity of quantitative reporting techniques, and develop the
infrastructure by which quantitative reporting can be achieved.
The Need for
Oncologic imaging is the
most obvious example where quantitative image metrics such as tumor diameter,
volume, standard uptake value, or vascular permeability are necessary in
treatment response assessment. With the proliferation of 3D volume imaging
techniques, image data sets are now rich with quantitative information, which
will eventually have important diagnostic value in general radiology
practiceâ€”well beyond oncology. In addition to anatomic
metrics, functional data are available on all 3D modalities when injected agents
are used. These data, determined by the interpreting radiologist in many cases
using automated software tools, will lead to more definitive and ultimately more
accurate diagnostic conclusions.
The explosion in biological
discovery in the last two decades has led medicine down a path from art to
science, and this will continue. Radiology must follow this trend to keep pace
with the sophistication of referring physicians, and quantitative imaging is an
important signpost on this journey. With the promotion of reimbursement slogans
such as "evidence-based medicine" and "pay-for-performance," a quantitatively
based radiological diagnosis is a necessary component in the radiology report of
the future. Now that this door has been opened for us, we in the radiology
communityâ€”including residentsâ€”need to step
through it and become part of this process.
How to Get
Although patient images
currently reside on PACS and the radiologist's text report resides on the
radiology information system (RIS) at most institutions, this dichotomy will
gradually erode as PACS and the RIS become more integrated. The radiology report
of the future will be an electronic document (e.g., web page) where the
radiologist's text report is supported by embedded key images, movies with
rotating maximum intensive projections of anatomy with overlaid functional
information, quantitative measurements detailed and highlighted, color bar charts
showing differential diagnostic probabilities based on quantitative and
qualitative findings, radiation dose estimates for X-ray and gamma ray
procedures, and a Twitter link to the radiologist in case a quick electronic
consultation is wanted. But how do we get there?
We need to work with our IT
departments to break down artificial barriers between computer
systemsâ€”we did it with PACS/DICOM, we can do it again. We have
the tools: voice dictation, the DICOM structure reporting object, emerging
software for image segmentation, spreadsheet software with linked drop-down
menus, word processing tools for putting it all together, and conversion software
to prevent alteration. We need radiology vendors to step up to the plate and
provide effective and efficient integrated software tools to advance the
information (quantitative and qualitative) content and clinical value of the
Building a reporting
infrastructure is necessary for quantitative imaging to happen, but developing
the science to support quantitatively pertinent radiological reporting is a
project that clinical academic radiologists can and should embrace. I suspect
that much of the quantitative reporting data is already in the heads of
experienced radiologists. Peer-reviewed literature in radiology generally does a
good job at reporting statistically justified quantitative data such as
sensitivity, specificity, etc. Radiologists read these papers and assimilate them
into their subjective minds, but the data are right there in the literature to
convert into a drop-down menu which would facilitate a more quantitative
By combining numerous
sources of statistically meaningful peer-reviewed clinical data with decision
support tools such as multiple regression analysis, fuzzy logic or neural
networks, quantitative data in the literature can be used to create a
quantitative reporting tool (QRT). Given the breadth
and depth of imaging in medicine, numerous quantitative reporting tools will be
necessary for each clinical subspecialty. Not only do we need to teach radiology
residents quantitative reporting, we should capitalize on their computer-savvy
upbringing to help create the quantitative report systems that will be the
essential tools of their future careers.
 Neural Networks in Radiological Diagnosis II. Interpretation of Neonatal
Chest Radiographs. Invest Radiology 1990; 25:1017-1023. Gross G.W., et
M. Boone, Ph.D. is professor of radiology and biomedical engineering at the
University of California, Davis, and is vice-chair of radiology (for research).
His research interests focus on the development of breast imaging systems,
primarily breast CT, on computer modeling of imaging systems and dose
distribution, and on quantitative imaging. He is a member of the QIBA Steering
Committee and was a primary investigator on the original Imaging Response
Assessment Teams (IRAT) program.
ANALYSIS TOOLS & TECHNIQUES
The Challenges of
Making fMRI Reproducible
By JAMES VOYVODIC, Phd
Functional MRI (fMRI) has
become a commonplace tool for basic research studies of brain function, and it
has great potential for becoming an important clinical imaging procedure.
Currently, however, the only routine clinical application of fMRI is for
localization of critical brain regions (e.g. speech and motor areas) in treatment
planning for brain surgery.
A major obstacle to broader
clinical application is the fact that standard fMRI methodologies tend to produce
results that are difficult to quantify and are not highly reproducible. Multiple
scans of a single individual performing the same behavioral task typically
produce similar brain activation maps,but with significant variability in the
details of active regions identified in different scans.This lack of reproducibility has made it difficult
to assess confidence in the accuracy of individual maps, to standardize
quantification of fMRI results, or to perform rigorous validation testing of
clinical fMRI procedures.
There are three fundamental
reasons why reproducibility is a problem in fMRI. The first is that fMRI is an
inherently indirect method for mapping brain function. It is based on mapping
regional changes in the blood oxygen level-dependent (BOLD) MR signal, which is
highly correlated with changes in brain activity.The BOLD signal is also sensitive, however, to
other factors that contribute to variability in blood flow or blood oxygenation.
For example, changes in anxiety or arousal levels, recent consumption of
cigarettes or alcohol, and vascular disease or brain tissue pathology can all
affect the coupling between neuronal functional activity and the observable BOLD
The second major obstacle to
reproducibility is the fact that fMRI analysis methods tend to identify active
brain regions based on the statistical significance of the task-dependent BOLD
signal compared to task-independent signal fluctuations. Because task-dependent
signals are typically comparable in magnitude to physiological noise levels,
signal averaging is usually essential.
The problem with
reproducibility arises because traditional fMRI mapping defines "active" brain
regions based solely on the statistical significance of the averaged
signal-to-noise ratio rather than on the BOLD signal itself. Again, this means
that factors such as attention, anxiety, or scan duration that affect the noise
level will produce variability in fMRI map results, even if the task-evoked
pattern of brain activity is constant.
The third major obstacle to
reproducibility is the fact that brain function is inherently complicated and
changing. Even the simplest reading task involves many brain regions including
vision, eye movement, and language comprehension areas. Moreover, the spatial
pattern of brain activity levels change if the person changes how he performs the
task or simply as the same task becomes easier with practice.
For fMRI to become a
reliable biomarker of brain activity, these reproducibility problems must be
addressed. Empirical studies are needed to better understand the relationship
between specific clinical task behaviors and brain activity and between brain
activity and other components of BOLD signals.
Most importantly, we need
improved statistical analysis methods that use statistical significance to assess
confidence while providing relatively noise-independent quantitative maps of
activity-dependent BOLD signal levels.
 Reproducibility of fMRI at 1.5T in a Strictly Controlled Motor Task.
Magn. Reson. Med., 2004; 52:751-760. Liu J.Z, et al.
 Neurophysiological Investigation of the Basis of the fMRI Signal.
Nature, 2001; 412:150-157. Logothetis N.K., et al.
Voyvodic, PhD, is an associate professor of radiology and neurobiology at Duke
University Medical Center in Durham, N.C He leads the clinical fMRI research
effort and is actively involved in developing real-time image analysis and data
quality assessment algorithms.
Quantitative Imaging/Imaging Biomarkers and QIBA Meetings and
RSNA Awarded $2.4 million
NIBIB Grant for Quantitative Imaging
RSNA has been awarded a two-year, $2.4 million contract from the
National Institute of Biomedical Imaging and Bioengineering (NIBIB) to support
RSNA's quantitative imaging and biomarkers
programsâ€”specifically the Quantitative Imaging Biomarkers
Alliance (QIBA), formed in 2008 to advance quantitative imaging and the use of
imaging biomarkers in clinical trials and practices.
The contract provides $1.2 million
each year to support a coordinated effort to establish an infrastructure for the
collection and analysis of imaging biomarker data. The long-term objective is to
establish processes and profiles leading to acceptance by the imaging community,
clinical trial industry and regulatory agencies of quantitative imaging
biomarkers as proof of biology, changes in pathophysiology and surrogate
endpoints for changes in the health status of patients.
RSNA 2010: QIBA Meetings and
QIBA held a working meeting at RSNA
2010 that provided attendees with a recap of significant accomplishments for the
year. These include:
â€¢ the award of a
two-year contract to RSNA by the National Institute of Biomedical Imaging and
Bioengineering (NIBIB) to support the ongoing work of QIBA
visibility achieved, in part, by publication of the MITA (Medical Imaging
Technology Assessment) White Paper, Why QIBA is a good thing for Radiology in
General, and the Imaging Manufacturers in Particular, and reflected by an
overwhelming interest in and attendance at the RSNA Special Interest Session,
Imaging Biomarkers for Clinical Care and Research
â€¢ convening a
workshop on standards for imaging endpoints, jointly sponsored by SNM, RSNA and
â€¢ continued work
on QIBA CT, MR and PET profiles which include standardized protocols
â€¢ acceptance for
publication by Radiology of two QIBA-related articles, "A Collaborative
Enterprise for Multi-Stakeholder Participation in the Advancement of Quantitative
Imaging," and "Quantitative Imaging Test Approval and Biomarker Qualification:
Inter-related but Distinct Activities."
The Quantitative Imaging
RSNA 2010 featured The
Quantitative Imaging Reading Room. This educational showcase featured 23
educational exhibits that provided visual and experiential exposure to
quantitative imaging and biomarkers through exhibitor products that integrate
quantitative analysis into the image interpretation process. Participants learned
through hands-on exhibits featuring informational posters, computer-based
demonstrations and Meet the Expert presentations scheduled throughout the
QI/IMAGING BIOMARKERS IN THE
PubMed Search on
Quantitative Imaging in the Residency Curriculum
Each issue of QIBA Quarterly will
feature a link to a dynamic search in PubMed, the National Library of Medicine's
interface to its MEDLINE database. Click here
to view a PubMed search on "why quantitative imaging curriculum should be
included in residency training programs."
For additional information
concerning quantitative imaging and structured reporting related to this topic,
please click here.
Take advantage of the My NCBI
feature of PubMed that allows you to save searches and results and includes an
option to automatically update and e-mail search results from your saved
My NCBI includes additional features for highlighting search terms, storing
an e-mail address, filtering search results and setting LinkOut, document
delivery service and outside tool preferences.