Transcript Journal Club 11/5/2014: DWI in Head and Neck Tumors

Gloria J. Guzman, MD, MSc
Christopher Owen, MA
Dan Marcus, PhD
Russ Hornbeck, MCSC
Matthew Smyth, MD, FAANS, FACS, FAAP
Tammie Lee Benzinger, MD, PhD
Financial Disclosures
 Gloria Guzman, Christopher Owen, Dan Marcus, Russ
Hornbeck, Matthew Smyth : None
 Tammie Lee Benzinger:
 Over 10k: She participates in clinical trials sponsored by
Eli Lilly, Avid Radiopharmaceuticals, and Roche. and
research funding from Avid Radiopharmaceuticals
(current)
 Less 10k: Advisory Board membership for Eli Lilly (2011
She also reports providing expert testimony and
receiving compensation from Kujawaski & Associates
(2011)
PURPOSE
 To prepare an educational exhibit on the workflow for
volumetric MRI quantification in patients with
Alzheimer’s dementia, multiple sclerosis and in
medically intractable epilepsy
APPROACH/ METHODS
What is volumetric imaging?
 Technique used to assess volumetry of brain regions or
whole brain volume after acquisition of high resolution
MR images for the delineation of anatomical
boundaries
 Most frequently used sequences are T1, T2 and proton
density
 MR scanners are either 1.5 Tesla (T) or 3T systems. 3T
systems offer higher resolution of tissue contrast (i.e.
increased visualization of the borders between grey
matter and white matter and cerebrospinal fluid)
Why is volumetry important?
 Early in the course of dementia, it is difficult to distinguish between
dementia due to Alzheimer’s Disease (AD) versus other, potentially
treatable causes of dementia, such as normal pressure hydrocephalus
(NPH). Tools that more accurately diagnose early manifestations of this
disease, such as hippocampal volumetry, are critical in patient management.
It is also useful as a quantitative measure of disease progression
 Likewise, evaluation of general whole brain atrophy in Multiple Sclerosis
(MS) patients can identify those that present with subtle or no clinical
symptoms. These patients could benefit from early aggressive treatment if
they demonstrate quantifiable and progressive brain volume loss. It is also
useful as a quantitative measure of disease progression
 Finally, focal cortical dysplasia (FCD) and other cortical anomalies can be
difficult lesions to identify radiographically. These patients have medically
intractable epilepsy that significantly affects their quality of life. Surgical
excision of these lesions results in marked improvement to complete
resolution of the epileptic episodes. A color-coded volumetric map that can
easily demarcate abnormal cortical areas would have a significant impact on
patient management and outcomes
How is volumetry processed?
 The FreeSurfer software suite is an MRI-based
brain imaging software package used in functional brain
mapping that facilitates the visualization of the different
anatomical regions of the brain cortex
 Contains both volume based and surface based analysis
which can be used for the reconstruction of topologically
correct and geometrically accurate models of both the
gray/white matter and pial surfaces
 Surfaces used in conjunction with segmentation can be
used for measuring cortical thickness, surface area and
folding, and for computing inter-subject registration based
on the pattern of cortical folds
Processing of volumetric map for MS and AD
 MR scans are moved to the Radiology supercomputer for
parcelation and segmentation via the FreeSurfer software
suit
 Whole brain, gray matter, and white matter volume are
calculated and normalized via a comparison of subject
intracranial volume to intracranial volume of a
supernormal cohort (called ICV or eTIV in FreeSurfer)
 After normalization, the large ROI's are compared to a
loess regression weighted against age generated with the
supernormal cohort. Standard deviations and percentiles
of specific subjects are calculated from the aforementioned
loess regression. Subjects can be displayed longitudinally
one a graph to show changes of volume over time with
respect to the loess regression
Processing of MRI volumetric map for FCD
 MR scans are moved to the Radiology supercomputer
for parcelation and segmentation via the FreeSurfer
software suit
 Thickness values for each vertex on the pial layer are
extracted from FreeSurfer assessor files, and
normalized against the supernormal group
 A weighted loess regression is generated for every
vertex in the FreeSurfer surface using the supernorm
data set, and the z-score (number of standard
deviations away from the mean) is calculated for the
vertices of a subject's surface
Processing of MRI volumetric map for FCD
 The resulting z-scores are stored in a text file generated by
the R-script and then converted to mgh (a file format
created by Massachusetts General Hospital), which in this
case is an "overlay" for the FreeSurfer suite
 An overlay can then be displayed on the patient's gray
matter surface using freeview or tksurfer, and images
generated from the z-score map
 The images can then be compared to what can be seen
though tkmedit. In tkmedit, we look at the areas of the
brain that have abnormal values in our z-score map to
ensure that the z-scores are not due to errors in the surface
themselves, rather than being a reflection of physiology
Example of adapted FreeSurfer using “R”
coding program for AD
 For normalizing the supernormal group and generating a reference graph:
#Find linear model between ROI and ICV
hippLeftReg = lm(super$Left_Hippocampus_volume~super$IntraCranialVol)
#Get head size (ICV) corrected hippocampal volume
super$corrHippLeftVolume = (super$Left_Hippocampus_volume hippLeftReg$coefficients[2]*(super$IntraCranialVol-mean(super$IntraCranialVol)))
#Smooth population data with loess (weighted regression)
smoothHippLeftFit = loess(super$corrHippLeftVolume~super$Age,degree=1,span=0.7)
#Generate graph represeting Supernormal Data
smoothHippLeftPred = predict(smoothHippLeftFit,newdata=43:90,se=TRUE)
 For normalizing the patient's data:
#Calculate corrected hippocampal volume for patient data using the linear model
already created
normHippLeftVolume =
all$Left_Hippocampus[all$Session==eval(parse(text=paste("\"",Subject_number,"\"",sep
="")))]
normCorrHippLeftVolume = normHippLeftVolume hippLeftReg$coefficients[2]*(all$IntraCranialVol[all$Session==eval(parse(text=paste("\"
",Subject_number,"\"",sep="")))]-mean(super$IntraCranialVol))
Normalization of data
 For MS and AD: The Super-Norm cohort is composed
of scans from individuals who have tested cognitively
normal in the Cognitive Dementia Rating (CDR 0) and
have normal biomarker Mean Cortical Binding
Potential (MCBP<0.18) or other normal cerebrospinal
fluid (CSF) markers for at least three years after the
included data
 For FCD: The normal children’s cohort is composed of
patients from ages 4 to 15 who have had normal MRI
brain scans at Saint Louis Children’s Hospital
FreeSurfer map
FreeSurfer map quality control
 Example of how the pial surface can extend into dura, and
how you can fix it by editing brainmask.mgz
FINDINGS/DISCUSSION
CASES
Multiple sclerosis (MS)
CASE 1 - MS
52 year-old female with relapsing remitting MS currently on Aubagio. Visual evaluation on
anatomical MR of volume loss is abnormal, but not strikingly so
2014
Multiple sclerosis
2014
CASE 1 - MS
52 year-old female with relapsing remitting MS currently on Aubagio. Visual evaluation on anatomical
MR is not strikingly abnormal. However, patient has advanced symptoms. She is unable to walk without
assistance of a walker, and can only do so for 5 feet. Whole brain volume in the 2nd percentile for age,
very advanced compared to expected from visual qualitative evaluation
CASE 2 - MS
52 year old, right-handed, Caucasian female with a history of relapsing remitting multiple
sclerosis that was diagnosed in 1997. Significant visual atrophy between 2007 (top) and 2014
(bottom)
Multiple sclerosis
2014
2014
CASE 2 - MS
52 year-old female with relapsing remitting MS currently on Aubagio. Visual evaluation
demonstrates significant interval atrophy between 2007 and 2014. On whole brain volume
evaluation done in 2014, brain volume is in the 9th percentile for her age
CASE 3 - MS
51 year old female, initially diagnosed in 2006 at age 42. She presents with worsening progression in
number and size of T2/FLAIR hyperintense lesions. Patient was changed from Tysabri. She began Tecfidera
in January 2015. This was begun, after failing Aubagio based on active MRI lesions. Below is MRI from 2015
(top) and 2014 (bottom) showing increase in size of left periventricular lesion (blue arrows)
2015
2015
2014
CASE 3 - MS
51 year old female, initially diagnosed in 2006 at age 42 year old. She presents with worsening
progression in number and size of T2/FLAIR hyperintense lesions. Below is whole brain
volumetry curve showing how the patient’s total brain volume is 2 standard deviations below
the mean, at the 2nd percentile
Alzheimer’s Disease (AD)
CASE 1 - AD
74 year-old female. Presented at age 64, had nursing as a profession. Her mother died of AD at age 70 with
disease onset at age 64. Patient’s initial complaint is of “word-finding difficulties”, has a clinical dementia
rating score (CDR) of 0 and mini mental state examination (MMSE) score of 30. At age 70 has CDR of 0.5
and MMSE of 29. At age 74 has CDR of 1 and MMSE of 23. MRI Brain at age 74 shows moderate aging
changes, see below
2014
2014
CASE 1 - AD
74 year-old female. Presented at age 64, had nursing as a profession. Her mother died of AD at
age 70 with disease onset at age 64. Patient’s initial complaint is of “word-finding difficulties”,
and at presentation has a clinical dementia rating (CDR) of 0 and mini mental state
examination (MMSE) score of 30. At age 70 has CDR of 0.5 and MMSE of 29. At age 74 has CDR
of 1 and MMSE of 23. Below is her volumetry graph demonstrating increasing loss of
hippocampal volume, with more than 2 standard deviations below the mean: 1st percentile on
right and 0th percentile on left
CASE 2 - AD
69 year-old male with no family history of AD. Presented to our institution at age 62 with memory loss, was
an aerospace engineer forced to retire by age 60, with initial onset of symptoms at age 57. At age 62 had a
CDR of 0.5 and MMSE of 30. By age 63 had a CDR of 1 and MMSE of 27. Remained CDR 1 through age 68
with MMSE of 24 and by age 69 had CDR of 2 and MMSE of 19. By visual evaluation, loss of whole brain
volume and hippocampal volume between 2009 and 2014.
2009
2014
CASE 2 - AD
69 year-old male with no family history of AD. Presented to our institution at age 62 with
memory loss, was an aerospace engineer forced to retire by age 60, with initial onset of
symptoms at age 57. At age 62 had a CDR of 0.5 and MMSE of 30. By age 63 had a CDR of 1 and
MMSE of 27. Remained CDR 1 through age 68 with MMSE of 24 and by age 69 had CDR of 2
and MMSE of 19. Below is volumetry graph demonstrating increasing loss of hippocampal
volume, with more than 2 standard deviations below the mean: 2nd percentile on right and 1st
percentile on left
CASE 3 - AD
73 year-old female. Presented at age 70, had nursing as a profession in early youth and then was a homemaker after having children. Paternal history of AD at age 72 with death at age 79. Patient’s initial complaint
is of problems with memory and has a clinical dementia rating (CDR) of 0.5 and mini mental state
examination (MMSE) score of 28 at age 70. At age 73 has a CDR of 0.5 and MMSE of 28. Initially diagnosed
with early amnesic disorder versus depression. Visually, there is decrease in volume between 2012 and
2014.
2014
2014
2012
CASE 3 - AD
73 year-old female. Presented at age 70, had nursing as a profession in early youth and then
was a home-maker after having children. Paternal history of AD at age 72 with death at age 79.
Patient’s initial complaint is of problems with memory and has a clinical dementia rating (CDR)
of 0.5 and mini mental state examination (MMSE) score of 28 at age 70. At age 73 has a CDR of
0.5 and MMSE 28. Initially diagnosed with early amnesic disorder versus depression. Although
there is no change in CDR or MMSE, there is decline of hippocampal volume bilaterally. Also
note that even though the hippocampal volume is normal for age, IT IS DECREASING. During
2014 clinical visit, definite diagnosis of AD made, discarding diagnosis of depression
EXTRA RELATED CASE
68 year old male with right occipital infarct (blue arrow) presents with new onset memory loss
after the stroke. Evaluation to discriminate between vascular dementia or AD. Initial read of
chronic right occipital stroke, no other notable findings on original report
2014
2014
EXTRA RELATED CASE
68 year old male with right occipital infarct presents with new onset memory loss after the stroke. Evaluation to
discriminate between vascular dementia or AD. Initial read of chronic right occipital stroke, no other notable
findings. Images below show significant asymmetric atrophy of the right hippocampus (blue arrow) compared to
the left, not noted on the original report. The occipital infarct extended anteriorly to include right hippocampus.
An addendum added to the report once the volumetric data was available, which shows 2nd percentile right and
34th percentile left hippocampal volume
Focal Cortical Dysplasia (FCD)
CASE 1 - FCD
13-year-old boy who has been suffering from medically refractory seizures since the first grade.
He typically has clusters of seizures, better or partially controlled with a combination of Dilantin
or Topamax. During the seizures, he abruptly begins yelling and moves his right arm and leg in
a disorganized fashion. The seizures are associated with drowsiness but can occur in the
daytime. He has some delays in language and development, but is at regular school. Imaging
demonstrates focal cortical thickening in the right frontal lobe (superior frontal sulcus), with
correlated area of decreased radiotracer uptake on PET-Brain
CASE 1 - FCD
13-year-old boy who has been suffering from medically refractory seizures since the first grade.
Cortical thickening identified with the help of PET BRAIN. He has had a comprehensive
evaluation by our multidisciplinary epilepsy center culminating in a recommendation for
surgery for a probable right frontal lobe onset with or without invasive electrode monitoring.
Cortical volumetric mapping demonstrates lesion in red, as a region of increased thickening
compared to adjacent normal cortex
CASE 2 - FCD
Patient is an otherwise healthy, 11-year-old boy who had the onset of seizures at age 5. His seizures occur
approximately once per month, but they can cluster with several in a month and he also went one period of
two years with no seizures. He has been trialed on the number of anticonvulsant medications and has been
evaluated by our Multidisciplinary Epilepsy Center including MRI, PET scan, and neuropsychological testing
and interictal and ictal video EEG evaluation. Imaging shows cortical thickening at the right frontal lobe
(superior and middle frontal gyrus) with correlated area of decreased radiotracer uptake on PET-Brain)
CASE 2 - FCD
Patient is an otherwise healthy, 11-year-old boy who had the onset of seizures at age 5. He has
been trialed on the number of anticonvulsant medications and has been evaluated by our
Multidisciplinary Epilepsy Center including MRI, PET scan, and neuropsychological testing and
interictal and ictal video EEG evaluation. Imaging shows cortical thickening at the right frontal
lobe (superior and middle frontal gyrus) with correlated area of decreased radiotracer uptake
on PET-Brain). Cortical volumetric mapping demonstrates lesion in red, as a region of increased
thickening compared to adjacent normal cortex
SUMMARY
In summary
 Volumetric imaging can be a powerful quantitative
tool for diagnosis and follow-up assessment of
multiple neurological conditions, among them
Alzheimer’s Disease, Multiple Sclerosis, and Focal
Cortical Dysplasia
 Useful information is acquired from single point data,
but more importantly, from longitudinal data that
allows evaluation of disease progression that may be
too subtle to assess clinically or by anatomical imaging
alone
Where can I find this software?
 FreeSurfer can be downloaded for free at:
https://surfer.nmr.mgh.harvard.edu/fswiki/Download
AndInstall
 The “R” code used to produce these volumetric maps
will be made available shortly in the open source
website GitHub.com. As well, they can be acquired
from Christopher Owen from our NeuroImaging Lab
at: [email protected]
References
 Barkovich AJ. "Current concepts of polymicrogyria.
"Neuroradiology” 2010: 52(6), 479-487
 Free SL, Bergin, PS, Fish DR, et al. Methods for normalization of
hippocampal volumes measured with MR. “American Journal of
Neuroradiology” 1995: 16(4), 637-643
 Cleveland, WS. Robust locally weighted regression and
smoothing scatterplots. “Journal of the American statistical
association” 1979: 74(368), 829-836
 Buckner, RL, Head D, Parker J, et al. A unified approach for
morphometric and functional data analysis in young, old, and
demented adults using automated atlas-based head size
normalization: reliability and validation against manual
measurement of total intracranial volume. “Neuroimage”
2004:23(2), 724-738
Thank you for your attention!