Dott. Ing. Letizia Squarcina, Ph.D.

Post Doctoral Fellow

Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico

University of Milan
Scientific Institute IRCCS “E. Medea”, Bosisio Parini, Italy.

NEUROSCIENCE AND PSYCHIATRY

Tecniche di analisi di MRI cerebrale
 Analysis with FreeSurfer

Cortical Surface Reconstruction

FreeSurfer creates computerized models of the brain from MRI data.

Input:
• T1-weighted (MPRAGE,SPGR)
• 1mm3 resolution
 Analysis with FreeSurfer

Cortical Surface Reconstruction

• Finds white/gray boundary – wm surface

• Finds pial/CSF boundary – pial surface
• Uses:

• Intensity information, spatial location, geometric structure

• Subcortical Segmentation
 Analysis with FreeSurfer

Surface Model

• Mesh (“Finite Element”)

• Vertex = point of 6 triangles
• XYZ at each vertex
• Triangles/Faces ~ 150,000
• Area, Distance
• Curvature, Thickness
 Analysis with FreeSurfer

Cortical Reconstruction Goals

• Geometrically Accurate surfaces

• Accurately follow the boundaries seen on the scan for each
of your individual subjects

• Topologically Correct surfaces

• Each surface is a 2-D continuous, non self-intersecting sheet
and can be inflated into a perfect sphere

• Surfaces are only as good as your scan.

 Analysis with FreeSurfer

MR Anatomy Caveats

• Dependent on data quality

• Contrast to noise
• Signal to noise
• Voxel resolution FreeSurfer Output
• Volumes
• MR Artifacts
• Surfaces
• MR susceptibility
• MR distortions • Surface Overlays
• ROI Summaries
 Analysis with FreeSurfer

Surface Mesh Inflation Surface ROI Group

D E Template
J

Curvature Sphere
C F I
Individual T1 Spatial
A Normalization

A Surface
Extraction B Thickness
Deformation
G H Field

2mm 4mm Apply

Volume ROI Deformation
Analysis with FreeSurfer
Analysis with FreeSurfer

Cortical Thickness

pial surface Distance between white and

pial surfaces along normal
vector.

1-5mm
 Analysis with FreeSurfer

Volumetric Segmentation

Cortex
Lateral Ventricle
White Matter

Thalamus

Caudate Putamen

Pallidum
Amygdala
Hippocampus
Not Shown:
Nucleus Accumbens
Cerebellum
 Analysis with FreeSurfer

Surface Segmentation
Precentral Gyrus
Postcentral Gyrus

Superior Temporal Gyrus

Based on individual’s folding pattern
 Analysis with FreeSurfer
Combined Segmentation

aparc

aparc+aseg

aseg
Analysis with FreeSurfer
Analysis with FreeSurfer
Analysis with FreeSurfer
 T1 analysis in Substance Abuse

Volume-based Morphometry
Purpose of the study
To investigate gray matter volumes and brain metabolism in psychotic
BD patients with and without substance abuse and in patients with
substance-induced psychosis coupling Magnetic Resonance Imaging
and Positron Emission Tomography.
Magnetic Resonance Imaging
High spatial resolution 
high definition of brain
structures

Low temporal resolution [s]

Positron Emission Tomography

Able to detect disease before
changes in the anatomy by
studying brain function through
biochemical functions
Low spatial resolution
PET scan risks caused by the
radioactive component used
during this procedure
 T1 analysis in Substance Abuse

Hypothesis

• Chronic use of drugs may alter the way the brain functions and therefore
we hypothesized that substance abuse may have direct effects on brain
regions underpinning the pathophysiology of BD patients
 T1 analysis in Substance Abuse

Participants and Methods

All Patients employed in this study have been recruited at the
Psychiatric Clinic, Fondazione IRCCS Ca’ Granda, Ospedale Maggiore
Policlinico, University of Milan, Milano, University of Milan

- 13 BD patients with substance abuse

- 20 BD patients without substance abuse
- 16 Substance-Induced Psychotic patients
- 27 Healthy Controls for MRI analysis
- 27 Healthy Controls for PET analysis

3T MRI and a 18-F-FDG-PET scanning in

resting condition.
Original T1 image Segment images (GM, WM & CSF) Normalise GM seg. to template
& remove extra-cerebral voxels to obtain norm. parameters

Pre-processing of Voxel-Based Morphemetry Approach

Smooth final images Apply norm. parameters to

original images
 Pre-processing of PET Approach

Original PET image Coregistered GM seg. to T1 images Smooth final images

to obtain norm. parameters
 VBM
Using the GLM for VBM
e.g, compare the GM/ WM differences
between 2 groups
Y = Xβ + ε

H0: there is no difference between

these groups

β: other covariates, not just the mean

 VBM

VBM: group comparison

GLM: Y = Xβ + ε
• Intensity for each voxel (V) is a function that models the different
things that account for differences between scans:

• V = β1(patient) + β2(control) + β3(covariates) + β4(global volume) + μ + ε

• V = β1(patient) + β2(control) + β3(age) + β4(gender) + β5(global

volume) + μ + ε

• which covariate (β) best explains the values in GM/ WM

• In practice, the contrast of interest is usually t-test between β1

and β2, ***
*** Eg, “is there significantly more GM (higher v) in the controls than in patients scans and does this explains the value in
v much better than any other covariate?”
 T1 analysis in Substance Abuse

Statistical analyses

- The same statistical analyses (ANCOVA) were performed for both the
MRI and PET approaches with age, sex and intracranial volumes as
covariates.

- Post-hoc analyses were performed to detect gray matter volumes and

brain metabolism abnormalities between the four groups.
 Brain region BA Laterality
BD patients without substance abuse <
Healthy Controls
Superior Frontal 9 Right
Superior Frontal 9 Left
Superior Frontal 10 Left
Superior Frontal 8 Left
Superior Frontal 8 Right
Superior Temporal 21 Left

Brain region BA Laterality

Substance-Induced Psychosis < Healthy
Controls
Superior Frontal 47 Right
Middle Frontal 9 Right
Precentral 6 Right
Precentral 4 Left
Superior Temporal 20 Left
Putamen - Right

Brain region BA Laterality

BD patients with substance abuse < Healthy
Controls
Inferior Frontal 47 Right

( all P <0.05 Family-Wise Error corrected

 Brain region BA Laterality

Substance-Induced Psychosis < BD patients

with substance abuse
Superior Frontal 10 Right
Superior Frontal 10 Left
Superior Temporal 38 Right
Middle Temporal 21 Right

Brain region BA Laterality

Substance-Induced Psychosis < BD patients
without substance abuse
Superior Frontal 10 Right
Superior Frontal 6 Left
Superior Frontal 11 Left
Superior Temporal 38 Right
Superior Temporal 38 Left
Inferior Temporal 20 Left

all P <0.05 Family-Wise Error corrected

T1 analysis in Substance Abuse

PET Results
 Brain region BA Laterality
BD patients without substance abuse < Healthy Controls
Precentral 9 Left
Inferior Frontal 46 Left
Inferior Frontal 47 Left
Precentral 44 Left
Middle Temporal 21 Left
Superior Temporal 41 Left
Middle Temporal 37 Right

Substance-Induced Psychosis < Healthy Controls

Inferior Frontal 9 Left
Middle Frontal 10 Left
Superior Temporal 42 Left
Superior Temporal 13 Right
Inferior Temporal 20 Right
Middle Temporal 21 Right

Substance-Induced Psychosis > Healthy Controls

Putamen - Right
Putamen - Left
Caudate - Right

BD with substance abuse < Healthy Controls

Medial Frontal 10 Left
Inferior Frontal 9 Left
Superior Temporal 42 Left
Inferior Temporal 20 Right
Superior Temporal 13 Right
Middle Temporal 21 Right

all P<0.005 uncorrected

Brain region BA Laterality
Substance-Induced Psychosis > BD
patients with substance abuse
Superior Frontal 10 Left
Precentral 6 Right

Brain region BA Laterality

Substance-Induced Psychosis < BD
patients without substance abuse
Superior Frontal 10 Right
Superior Temporal 22 Right
Superior Temporal 41 Left

all P<0.005 uncorrected

 T1 analysis in Substance Abuse

• Regardless of the presence of substance abuse, BD patients showed extensive

structural and metabolic fronto-temporal alterations compared to healthy
controls in brain regions well-known to be involved in this disorder, including
dorso- and ventro-lateral prefrontal cortices as well as superior , middle and
inferior temporal gyri (Brambilla et al., 2005; Selvaraj et al., 2012).

• Therefore, these findings further support the pivotal role of these structures
in the pathogenesis of BD which might, in turn, play a role in the genesis of
mood regulation and neurocognitive deficits observed in these patients.
 T1 analysis in Substance Abuse

• Substance-induced psychotic patients showed similar but more

extensive structural and metabolic alterations in fronto-temporal
regions, including dorso- and ventro-lateral prefrontal cortex as well as
superior and middle temporal regions compared to healthy controls
and BD patients with or without substance abuse.

• These results are in keeping with previous evidence reporting

significant prefrontal reductions in GM volumes (Bartzokis et al., 2000;
Makris et al., 2008a) and density (Dalwani et al., 2011) in drug addicts.

• These fronto-temporal alterations might be in line with what have

been reported in recent years by MRI studies in schizophrenia
(Buchanan et al.,2004; Zhang et al., 2015). Indeed it has been
hypothesised that substance-induced psychosis might lie within a
"grey" overlapping zone between schizophrenia or BD within a
psychosis continuum (Keshavan et al., 2011; Altamura and Goikolea,
2008).
 T1 and fractals
WHAT IS A FRACTAL?

Fractals are objects characterized by:

self-similarity: they are patterns repeating at different scales:

they are made of smaller copies of themselves

Romanescu

http://todaysorganicgardening.com/
 T1 and fractals
WHAT IS A FRACTAL?

Fractals are objects characterized by:

self-similarity: they are patterns repeating at different scales:

they are made of smaller copies of themselves

 T1 and fractals
Images were skull-stripped using FSL BET

Gray matter (GM) was obtained from brain volumes (FSL FAST)

GM maps were transformed (FNIRT + FLIRT) to MNI standard space to have all subjects aligned to a common reference
space, allowing slice-by-slice comparisons

BET

FAST

FNIRT
FNIRT FLIRT
FLIRT

MNI STANDARD SPACE

 T1 and fractals

Fractal dimension was computed on the whole brain and on some major structures:

- Frontal lobe

- Temporal lobe

- Occipital lobe

- Parietal lobe

Obtained from the MNI structural atlas in standard space.

We performed the analysis computing a FD value for each slice of the considered volume. To ensure
presence of contiguous GM regions the analysis was restricted to structure-specific ranges of slices.
 T1 and fractals

Thus, we created a template data from 10 HC, to obtain a reference value

for FD across the brain, and compared FD values with those of patients:

We obtained slice-by-slice group mean FD values

 T1 and fractals
frontal lobe temporal lobe

FRACTALS ON BD PATIENTS

parietal lobe occipital lobe

SCZ BD HC
 T1 and fractals
FRACTALS ON BD PATIENTS

This method allows to consider patients singularly instead as of part of a group

For example, for a randomly selected BD patient from our group in frontal lobe:

FD

Slice
 T1 and fractals

With fractal dimension we considered the complexity of gray matter, and of the whole, non-segmented brain, to see if its
possible modifications could be related to illness.

Quantitative indexes that discriminate brain images of subjects affected by psychiatric diseases from those of healthy
could potentially help to improve the diagnostic process.

Bipolar disorder is related with loss of GM volume (Bora et al., 2010; Yu et al., 2010). It is thus reasonable to hypothesize
that alterations in GM should be detectable when considering its geometry.