Transcript National Alliance for Medical Image Computing: Namic

The Analysis of Brain Lesions in
Neuropsychiatric Systemic Lupus Erythematosis
H Jeremy Bockholt
Charles Gasparovic
The MIND Institute / UNM
Albuquerque, NM
NA-MIC All Hands Meeting
20070111
Salt Lake City, Utah
National Alliance for Medical Image Computing
http://na-mic.org
Background and Significance
• Systemic lupus erythematosus (SLE) is an autoimmune
disease affecting multiple tissues, including the brain
–
•
the facial rash of some people with lupus looked like the bite or scratch of a wolf
("lupus" is Latin for wolf and "erythematosus" is Latin for red). patients may feel
weak and fatigued, have muscle aches, loss of appetite, swollen glands, and
hair loss, sometimes have abdominal pain, nausea, diarrhea, and vomiting.
Estimates of SLE prevalence range from 14.6-372 per 105
– About 1.5 million americans, 90% diagnosed are female
• Neuropsychiatric SLE (NPSLE), a term that subsumes the
neurologic and psychiatric complications of SLE, occurs in
up to 95% of SLE patients
• While MRI often reveals distinct white matter
abnormalities in active NPSLE, the pathologic processes
underlying these lesions, whether purely autoimmune or
vascular (e.g., hemostasis), are unknown
National Alliance for Medical Image Computing
http://na-mic.org
Aims of the RO1 Study
• Test hypotheses concerning the possible
thrombotic or embolic origin of white matter brain
lesions in NPSLE
• Examine whether the incidence of lesions
correlates with either levels of thrombosis markers
or emboli in the blood or a potential source of
emboli in the heart
• Examine whether overall lesion load or the levels
of particular classes of lesion correlate with
cognitive function
National Alliance for Medical Image Computing
http://na-mic.org
Background and Objective
• Critical to understanding the etiology of brain lesions in
NPSLE will be the accurate measurement of their location,
size, and time course.
• Lupus brain lesions are known to vary in MRI intensity and
temporal evolution and include acute, chronic, and
resolving cases.
• Monitoring the time course of image intensity changes in
the vicinity of lesions, therefore, may serve to classify them
based on their temporal characteristics.
• Major objective of this DBP will be the evaluation of
existing tools and the development new tools using the NAMIC kit for the time series analysis of brain lesions in
lupus.
National Alliance for Medical Image Computing
http://na-mic.org
Summary of MRI Protocol
• The MRI data in this project are collected on a 1.5T
Siemens Sonata scanner using an 8-channel head coil.
– T1-weighted 3D fast low angle shot (FLASH) sequence
(TR/TE = 12/4.76ms, flip angle = 20deg,
FOV =
220x220mm, resolution = 192x192, 120 1.5-mm slices, total
time = 6m32s)
– T2-weighted fast spin echo sequence (TR/TE = 9040/64ms,
turbofactor=5, FOV = 220x220mm, resolution = 192x192,
120 1.5-mm slices, total time = 6m2s)
– Fluid Attenuated Inversion Recovery (FLAIR)/fast spin
echo sequence (TR/TE = 1000/105ms, TI = 2500ms, echo
train=9, field of view (FOV) = 220x220mm, resolution =
192x192, 88 1.5-mm slices, total time = 9m2s).
National Alliance for Medical Image Computing
http://na-mic.org
Additional Protocol
• transcranial doppler ultrasound is used to
detect microemboli in the brain
• transesophageal
echocardiography
is
performed to evaluate general cardiac status
and to detect the presence of heart valve
vegetations, as potential sources of emboli.
• markers of hemostasis,
– analyses for platelets, coagulation, fibrinolysis, and
anti-phospholipid antibodies.
• battery of neuropsychological tests is
administered to evaluate cognitive function.
National Alliance for Medical Image Computing
http://na-mic.org
Summary of Study Design
• Clinical Assessments and MRI will be
performed on approximately 60 SLE patients
and 30 normal control subjects over a period
of 48 months.
• Subset of this group will be approximately 15
SLE subjects with NPSLE and 15 SLE
subjects without NSLE
• Subset of initial SLE groups will be reexamined in the each of the subsequent years
(2-4) of the study.
National Alliance for Medical Image Computing
http://na-mic.org
Image Processing Needs
• Co-registration of T1, T2, and FLAIR.
• A robust and reliable method capable of segmenting
the brain into at least four classes: gray matter, white
matter, cerebrospinal fluid, and white matter lesions.
• Longitudinal follow-up registration
• Correlations such as appearance of and changes in
particular lesions correlate with the onset and
remission of the neuropsychiatric symptoms of
NPSLE.
• Others to be pointed out during discussion and perhaps
ones that we do not even know about yet
National Alliance for Medical Image Computing
http://na-mic.org
Goals of the NPSLE DBP
• Use and extend the NA-MIC kit to make a fully automated
lesion analysis tool.
– Input data
• image data from the T1-weighted, T2-weighted, and FLAIR
sequences
– Output data
• will be probability maps for each tissue class, the number of
lesions, the volume of each lesion, and the total lesion volume at
each time point
• Changes in lesion size and changes in pixel intensity within the
volume of each lesion will be displayed graphically
• Time course data will also be amenable to time series analysis
by statistical tools such as general linear modeling (GLM),
independent component analysis (ICA), or potentially Bayesian
analysis
National Alliance for Medical Image Computing
http://na-mic.org
Example NPSLE Lesion
Hypointense on T1
Hyperintense T2
National Alliance for Medical Image Computing
http://na-mic.org
Hyperintense on FLAIR
TMI/UNM Current Tool Use
• A number of different MRI data analysis tools are used by
researchers at TMI and UNM:
– Slicer
• Mutli-modal Scientific Visualization
– BRAINS2, FSL, SPM, Freesurfer, and Slicer
• for image segmentation and labeling
– FSL, SPM, and AFNI
• used fMRI data analysis
– GTRACT, DTIStudio
• Used for DTI/DWI analysis
National Alliance for Medical Image Computing
http://na-mic.org
Goals Aug 2007 - July 2008
•
•
•
Collect Baseline data points
Hire experienced C++ programmer, train and
mentor to become expert at using NA-MIC kit
Evaluate the algorithm/approach performance of at
least four methods for lesion segmentation of
NPSLE brain images:
–
–
–
–
EM-Segment Method, developed by Sandy Wells
Modification of EM-Segment, developed by Vincent
Magnotta
K-means+discriminant analysis, developed by Vincent
Magnotta
Manual classification by an expert rater.
We will use the STAPLE and/or a method using the
Williams Index to cross-validate these methods.
National Alliance for Medical Image Computing
http://na-mic.org
Goals Aug 2008 - July 2009
• Complete collection of follow-up data points
• Participate in the overall NA-MIC grant renewal process
• Use and extend the NA-MIC kit to develop a lesion
analysis module that provides a workflow for the
registration of T1,T2, and FLAIR within and across
different scanning sessions, automated segmentation into
gray, white csf, and lesion, and summary of lesion
location, size, and intensity. A fully-functional prototype
will be completed in time for grant resubmission.
• Extend lesion analysis module to provide time series
analysis tools as the follow-up data points become
available
National Alliance for Medical Image Computing
http://na-mic.org
Goals Aug 2009 - July 2010
• Complete a final, stable production time series
lesion analysis module to be made available in NAMIC kit
• Provide final publically available data-sets to
support a robust training tool for using the module
• Contingent on renewal funds, try out different
statistical methods of time series analysis, drawing
from methods that have been successful in the
analysis of fMRI data, including GLM and ICA
• Use GLM or multiple regression analysis to
examine correlations between changes in lesion
intensity or size and measures of thrombo-embolic
activity or the onset of clinical symptoms of NPSLE
National Alliance for Medical Image Computing
http://na-mic.org
NPSLE DBP Driving Force
• Methods developed in this DBP will have a broad
impact on the study of brain diseases involving
MRI-visible lesions
• Characterization of the time evolution of these
lesions will undoubtedly help to elucidate not only
the origins of the lesions but their relationships to
disease symptoms.
• No current image analysis package currently
permits this level of automated lesion time series
analysis--this will make the NA-MIC kit unique
and more desirable to be used by the broader
community
National Alliance for Medical Image Computing
http://na-mic.org
NPSLE DBP Summary
• Using the NA-MIC kit, we will augment, develop, and
validate tools for the quantification of brain lesions thought
to underlie the cognitive dysfunction of NPSLE.
• We will extend NA-MIC kit to analyze changes in these
lesions with time and to relate these changes to the
fluctuating symptoms of NPSLE
• We will gain greater insight of NPSLE etiology
• The automated lesion time series analyses should
generalize well to other vascular disorders such as vascular
dementia, myotonic dystrophy, and multiple sclerosis.
National Alliance for Medical Image Computing
http://na-mic.org
The MIND Institute / UNM
The Analysis of Brain Lesions in
Neuropsychiatric Systemic Lupus Erythematosis
INVESTIGATORS:
H. Jeremy Bockholt
Charles Gasparovic, Ph.D.
CONSULTANTS:
Vincent Magnotta, Ph.D.
Vince Calhoun, Ph.D.
PROGRAMMER:
Sumner Williams, M.S.
Subject Type
Baseline
SLE
30
Lupus
NPSLE
30
Healthy Normal Volunteers 30
year 1
15
15
0
year 2
15
15
0
BACKGROUND:
• NPSLE is an autoimmune disorder that
causes neurological and psychiatric
complications
• Afflicted patients have distinct white matter
lesions that vary over time
• To understand the etiology of brain lesions in
NPSLE, accurate measurement of lesion
location, size, and time course must be
achieved
AIMS:
•
Evaluate existing tools and develop new tools
using the NA-MIC kit for time series analysis
of brain lesions found in NPSLE
DATA:
•
Data collected under R01-NS35708-04
National Alliance for Medical Image Computing
http://na-mic.org
MRI, DTI, perfusion, transcranial ultrasound,
echocardiography, and neuropsychology
References
1. Aladjem, H (Editor): LFA study shows between 1,400,000 and
2,000,000 people diagnosed with lupus. Lupus News 14:12,
1994.
2. Sibbitt WL Jr, Brandt JR, Johnson CR, Maldonado ME, Patel
SR, Ford CC, Bankhurst, AD, BrooksWM: The incidence and
prevalence of neuropsychiatric syndromes in pediatric-onset
systemic lupus erythematosus. J Rheum 2002 29:1536-42.
3. Warfield SK, Zou KH, Wells WM. Simultaneous truth and
performance level estimation (STAPLE): an algorithm for the
validation of image segmentation. IEEE Trans Med Imaging.
2004 Jul;23(7):903-21.
4. Martin-Fernandez M, Bouix S, Ungar L, McCarley RW,
Shenton ME. Two methods for validating brain tissue
classifiers. Med Image Comput Comput Assist Interv Int Conf
Med Image Comput Comput Assist Interv. 2005;8(Pt 1):51522.
National Alliance for Medical Image Computing
http://na-mic.org
Acknowledgements
•
•
•
•
•
•
•
•
Charles Gasparovic
Wilber Sibbit
Carlos Roldan
Bruce Rosen
John Rasure
DOE Grant No. DE-FG02-99ER62764
Function BIRN
Lupus: R01-NS35708-04
National Alliance for Medical Image Computing
http://na-mic.org
Questions and Discussion
•
•
•
•
•
•
•
Preprocessing and filtering
Baseline Shape Analysis
Longitudinal Change Shape Analysis
Anatomical Localization of Lesions
Longitudinal Matchup of Lesions
Is the timeline reasonable?
Should we extend and generalize the EM
Segment Module or develop a standalone
module?
• Other items and discussion?
National Alliance for Medical Image Computing
http://na-mic.org