Wisdom Based Computing, Volume 1, Issue 1, April 2011

INTERNATIONAL JOURNAL OF WISDOM BASED COMPUTING, VOL. 1(1), 2011
1
On Experimenting with Functional
Magnetic Resonance Imaging of Both
Hands Movement
J. Satheeshkumar, Student Member, IEEE, R. Rajesh, Member, IEEE,
S. Arumugaperumal, Member, IEEE, C.
Kesavdas, Fellow, Functional Neuroimaging
Abstract
Functional magnetic resonance imaging is an improved method for identifying
neural correlates of psychological processes, regionally specific changes and mo-
tor/sensory activation region for specific function in human brain. These activations
and characteristics can be estimated based on Blood Oxygen Level-Dependent (BOLD)
phenomena, which are modeled in a statistical framework such as general linear modal
to obtain statistical parametric maps of the various neuronal activations for a particular
task. This article mainly focused on step by step procedure of identifying motor/sensory
activation region for specific stimuli, which helps the radiologists and researchers
for better interpretation. This article also presents the results and inferences from
neuroimaging data of hand movement experiment using statistical parametric mapping
(SPM). The results match with activation atlas by Penfield and Rasmussen (1950).
Index Terms
BOLD, fMRI, SPM, Hand movement.
I. INTRODUCTION
The recent developments in medical imaging from the past few decades allow
researchers to study more about cognitive tasks. Understanding the function of
human brain is still a complex task for researchers as well as radiologists.
Functional magnetic resonance imaging (fMRI) is a noninvasive technique with
high spatial and temporal resolution has provided a method of visualizing a
correlate of neural activity and helps researchers to address unanswered questions
in psychology, cognitive science, psychopathology, etc. [1], [2], [3], [4], [5],
[6], [7], [8], [9], [10], [11], [13]. Blood Oxygen level dependent (BOLD) is
an improved technique in fmri for measuring various motor/sensory activation
regions of human brain. This BOLD phenomena is based on difference of
concentration between oxygenated and deoxygenated hemoglobin. The changes
in the hemoglobin composition are proportionate to the neuronal activity of brain.
Recent advances in functional neuroimaging techniques are revolutionizing the
approach to surgical planning in tumor resection and in patients with intractable
epilepsy [15]. SPM2 is a relatively new tool used for mapping Functional Magnetic
J. Satheeshkumar and Dr. R. Rajesh are with Dept. of Computer Science and Engg., Bharathiar University,
Coimbatore - 46, India
Prof. Dr. S. Arumugaperumal is with Dept. of Computer Science, St. Hindu College, Nagercoil, India
Dr. C. Kesavdas, Additional Professor, Sree Chitra Tirunal Institute for Medical Sciences and Technology,
Thiruavanathapuram, India

---

INTERNATIONAL JOURNAL OF WISDOM BASED COMPUTING, VOL. 1(1), 2011
2
Fig. 1.
Paradigm design using boxcar.
Resonance Imaging (fMRI) to a statistical parametric map to identify the sensor,
motor and cognitive tasks in the specific regions of the brain.
This article presents a detailed experimental study of a tumor patient. The
study reported in this article forms only an initial analysis of finding the region
of influence of hand movement and to understand whether tumor is having any
influence in the hand movement region. This study is not for identifying and
diagnosing tumor. This article also explains sequence of steps involved for getting
statistical parametric map and the results match with the sensory/motor activation
atlas by Penfield and Rasmussen (1950) [12], [14].
This article is organized as follows. Section II explains the experimental setup,
materials and methods for the study reported in this article. Section III presents
the simulation and the results of the experimental study. Section IV concludes the
article.
II. MATERIALS AND METHODS
The fMRI data set of a subject used in this study was imaged from a 1.5T MRI
system (Siemens Avanto), Sree Chitra Tirunal Institute for Medical Sciences and
Technology, Thiruvananthapuram, India. A structural image of the brain with 176
slices of 1 mm thickness was obtained for overlaying the final results. Functional
images of 100 volumes were obtained during the experiment, where each volume
consists of 36 slices of 3mm thickness. Each volume scan was acquired over 3.58
seconds with voxel size of 64x64x36. The total experiment took around 6 minutes
to complete the scan.
The subject was asked to move both hands in specified timings. The experiment
started with 10 scans (35.8 sec) of rest followed by 10 scans (35.8 sec) of hand
movement. Five cycles of rest-hand movement completes the experiment obtaining
100 volumes of data. Fig. 1 shows the boxcar design for the paradigm.
The data were analyzed using statistical parametric mapping (SPM2, Wellcome
Department of Cognitive Neurology, London, UK) implemented in Matlab [3],
[4]. The scanning protocol used sequential axial-slice acquisition in ascending
order, activations in different slices were measured at different time points. Hence
the data was interpolated and resampled during realignment phase. A mean image
was created during the realignment phase. The structural MRI, acquired using a
standard three-dimensional
weighted sequence was co-registered to this mean
I
1/2
( ) image. The data were smoothed using a 8mm full width at half maximum
I
3/4

---

INTERNATIONAL JOURNAL OF WISDOM BASED COMPUTING, VOL. 1(1), 2011
3
0.08
x translation
y translation
z translation
0.06
0.04
0.02
mm
0
-0.02
-0.040
10
20
30
40
50
60
70
80
90
100
image
Fig. 2.
Translation of images with respect to the first functional image
isotropic Gaussian kernel. Data analysis was performed by modeling the different
conditions (hand movement and rest) as reference waveform in the context of the
general linear model [3]. t-statistics for all the voxels were obtained by testing
the specific stimuli with appropriate linear contrasts. The z-statistics, namely the
statistical parametric map (SPM) was obtained by transforming the t-statistics.
These statistical parametric maps are then interpreted. To identify the regions of
responses, the `hand movement' and `rest' stimulus conditions were compared.
The detailed simulations and the results of each steps of simulation along with
the parameters are discussed in the following sections.
III. SIMULATION AND RESULTS
All functional images were realigned to the first functional image using a six-
parameter rigid-body transformation [4]. A mean image was created during the
realignment phase. Fig. 2 shows the x, y, & z translations and Fig. 3 shows the
pitch, roll, & yaw rotations. The structural MRI, acquired using a standard three-
dimensional
weighted sequence was co-registered to the mean ( ) image.
I
I
1/2
3/4
For analysis using SPM, the images were spatially smoothed using Gaussian with
FWHM 8 mm.
In the design stage of fMRI, interscan interval (TR) was assigned as 3.58 secs,
where TR is the time taken to gather one whole-brain volume. The vector onsets
given for the experiment should match with the paradigm design and was assigned
as [
] (given in scans) with duration of 10 scans. Fig. 4 shows
1/21/2
1/2
1/2
1/2
1/2
the Time domain regressors for hand and Fig. 5 shows the Frequency domain
using 128 second High-pass filter.
After specifying the spm.mat file and all smoothed images in the data spec-
ification stage of fMRI, the design matrix is created and is showed in the Fig.
6.
Following the design and data specification, estimation is done which estimates
parameters, smoothness parameters and hyperparameters.
Statistical parametric mapping is done after estimation with t-contrast as [
],
1/2
1/4
where 1 represents the active condition [Statistical parametric mapping refers

---

INTERNATIONAL JOURNAL OF WISDOM BASED COMPUTING, VOL. 1(1), 2011
4
-3
x 10
1
pitch
roll
yaw
0.5
0
mm
-0.5
-1
-1.50
10
20
30
40
50
60
70
80
90
100
image
Fig. 3.
Rotation off images with respect to the first functional image
1
0.8
0.6
regressor[s] 0.4
0.2
0
10
20
30
40
50
60
70
80
90
100
scan
Fig. 4.
Time domain regressors for hands
0.2
0.18
0.16
0.14
0.12
0.1
relative spectral density 0.08
0.06
0.04
0.02
0.02
0.04
0.06
0.08
0.1
0.12
Frequency (Hz)
Fig. 5.
Frequency domain using 128 second High-pass filter

---

INTERNATIONAL JOURNAL OF WISDOM BASED COMPUTING, VOL. 1(1), 2011
5
Both Hands
Constant
Images
Parameters
Fig. 6. Design Matrix. The first column contains timing for the both hand movement. White strips correspond
to the hand movement. The second column corresponds to the constant.
to the construction of spatially extended statistical processes to test hypotheses
about regionally specific effects (Friston et al 1991). Statistical parametric maps
(SPMs) are image processes with voxel values that are, under the null hypothesis,
distributed according to a known probability density function, usually the Student's
T or F distributions. These are known colloquially as T- or F-maps]. This process
under goes T test and produces the T-map. Fig. 7 shows the T-map which gives the
maximum intensity projection on the so called "glass brain". Fields significantly
activated by hand-motion obtained after t-test (p-values are adjusted for search
volume) are shown in Table 1. Orthogonal sections of the threshold statistic image
overlaid on high resolution
weighted structural image is shown in Fig. 8, where
I
1/2
the white regions show the activations due to hand movement. The results match
with the sensory/motor activation atlas by Penfield and Rasmussen (1950). Fig. 9
shows the 3-D view of activation regions(white regions identified by arrow line)
for hand-movement.
Hemodynamic responses of a voxel at [-34 -44 17] (corresponds to global
maximum) have also been carried out and the results are as follows. Hemodynamic
responses of a voxel at maximum intensity projection ([-34 -44 17]) with radius
3mm is shown in the Fig. 10. Stimulus efficacy with 90% confidence intervals,
State variables for the first order kernel, Hemodynamic parameters, 1st order and
2nd order kernals for BOLD responses for the voxel ([-34 -44 17]) are shown
respectively in Figs. 11, 12, 13, 14, 15.
It is found from the experiment that the tumor has not making significant change
in hand activation region since the tumor is not exactly located in hand activation
region of the subject.
IV. CONCLUSION
Functional magnetic resonance imaging is an efficient method of medical
imaging, which provides much new information about the function of human
brain. This information helps medical researchers and radiologists for better

---

INTERNATIONAL JOURNAL OF WISDOM BASED COMPUTING, VOL. 1(1), 2011
6
Fig. 7.
Glass Brain. The motor activation region of the hand movement is visible as dark gray clusters in
the glass brain
Fig. 8.
Overlay of filtered SPM on a structural image. The white regions show the activations due to hand
movement
interpretation of results and for understanding about unknown activation regions,
which may responsible for various stimuli given to the subject.
This article presents statistical parametric maps of a tumor subject obtained
from Sree Chitra Tirunal Institute for Medical Sciences and Technology and the
results match with the sensory/motor activation atlas by Penfield and Rasmussen
(1950). The inferences shows that the tumor has not making significant change
in hand activation region since the tumor is not exactly located in hand activation
region of the subject.
ACKNOWLEDGEMENT
The first two authors are thankful to Dr. A.K.Gupta, Dr. C. Sujesh, Dr. B.
Thomas, Dr. Kapilamoorthy and colleagues in the Department of Imaging Sciences
and Interventional Radiology, Sree Chitra Tirunal Institute for Medical Sciences
and Technology for supporting them to do their research training in the institution.

---

INTERNATIONAL JOURNAL OF WISDOM BASED COMPUTING, VOL. 1(1), 2011
7
TABLE I
FIELDS SIGNIFICANTLY ACTIVATED BY HAND-MOTION OBTAINED AFTER T-TEST (P-VALUES ARE
ADJUSTED FOR SEARCH VOLUME)
Cluster-level
voxel-level
x, y, z mm
T
O
O
O
O
O
OO
O
UO
OO
O
I

OO
O
E
OO
O
UO
OO
O
0.000
464
0.000
0.000
0.000
12.75
Inf
0.000
-34
-44
17
0.000
0.000
7.54
6.65
0.000
-19
-40
28
0.000
0.000
6.30
5.74
0.000
-50
-40
2
0.000
348
0.000
0.000
0.000
9.31
7.80
0.000
28
-35
24
0.000
0.000
9.16
7.72
0.000
44
-39
13
0.000
0.000
8.35
7.20
0.000
40
-43
5
0.001
70
0.000
0.000
0.000
7.81
6.84
0.000
38
27
-55
0.000
0.000
7.03
6.29
0.000
31
15
-66
0.000
228
0.000
0.000
0.000
7.06
6.31
0.000
20
-63
-62
0.000
0.000
6.24
5.69
0.000
-12
-63
-62
0.004
0.000
5.55
5.15
0.000
20
-71
-85
0.001
75
0.000
0.000
0.000
6.96
6.24
0.000
-24
27
-51
0.000
115
0.000
0.000
0.000
6.94
6.22
0.000
-20
-67
-88
0.177
19
0.019
0.191
0.000
4.57
4.33
0.000
55
-16
-10
0.258
16
0.030
0.193
0.000
4.56
4.33
0.000
-8
-16
2
0.424
12
0.055
0.282
0.001
4.43
4.21
0.000
-47
-17
-2
0.228
17
0.026
0.415
0.001
4.28
4.08
0.000
-24
-9
-77
0.997
1
0.574
0.976
0.007
3.59
3.47
0.000
47
-27
-70
0.918
4
0.247
0.985
0.008
3.55
3.43
0.000
51
-12
-36
0.997
1
0.574
0.989
0.009
3.52
3.41
0.000
23
11
-51
0.985
2
0.414
0.998
0.013
3.39
3.28
0.001
15
-47
-92
0.997
1
0.574
1.000
0.018
3.25
3.16
0.001
-39
-75
-62
Fig. 9.
Three Dimensional view of activation regions(identified arrow line) for hand-movement

---

INTERNATIONAL JOURNAL OF WISDOM BASED COMPUTING, VOL. 1(1), 2011
8
Time(Seconds)
Fig. 10.
Hemodynamic motion of a voxel at [55 -16 26] with radius 3mm
Fig. 11.
Stimulus efficacy with 90% confidence intervals for hand movement
They are also thankful to all staff of the Department of Computer Science and
Engineering, Bharathiar University for their support.
REFERENCES
[1] Bandettini PA, Jesmanowicz A, Wong EC, Hyde JS: Processing strategies for time course data sets
in functional MRI of the human brain. Mag. Res. Med. 30: 161-173, 1993.
[2] Bullmore ET, Brammer MJ, Williams SCR et Al: Statistical methods of estimation and inference for
functional MR images. Mag. Res. Med. 35: 261-277, 1996.
[3] Friston KJ, Homles AP, Worsley KJ et Al: Statistical parametric maps in functional imaging: a general
linear approach. Human Brain Map 2: 189-210, 1995.
[4] Friston KJ, Ashburner J, Frith CD et Al: Spatial registration and normalization of images. Human
Brain Mapp 2:165-189, 1995.
[5] Friston KJ, Jezzard PJ, Turner R: Analysis of functional MRI time-series. Hum. Brain Mapp. 1:153-
171, 1994.
[6] Friston KJ, Frith CD, Turner R, and Frackowiak RSJ: Characterizing evoked hemodynamics with
fMRI. NeuroImage 2:157-165, 1995
[7] Friston KJ, Ungerleider LG, Jezzard P, and Turner R: Characterizing modulatory interactions between
V1 and V2 in human cortex with fMRI. Hum. Brain Mapp. 2: 211-224, 1995.
[8] Friston KJ, Williams S, Howard R, Frackowiak RSJ, and Turner R: Movement related effects in fMRI
time series. Mag. Res. Med. 35:346-355, 1996.
[9] Gerardin E, Lehericy S, Pochon JB et Al: Foot, Hand, Face and Eye Representation in the Human
Stratum. Cerebral cortex 13:162-169, 2003.
[10] Stefan Koelsch, Thomas Fritz, Katrin Schulze, David Alsop and Gottfried Schlauga: Adults and
children processing music: An fMRI study. NeuroImage 25, 1068-1076, 2005.
[11] Mrocz I, Karni A, Haut S, Lantos G, and Liu G: fMRI of triggerable aurae in musicogenic epilepsy.
NEUROLOGY 60, 705-709, 2003.
[12] Penfield W, Rasmussen T: The cerebral cortex of man. New York: Macmillan, 1950.

---

INTERNATIONAL JOURNAL OF WISDOM BASED COMPUTING, VOL. 1(1), 2011
9
Normalized Values
Time in Seconds for Both hands
Fig. 12.
State variables for the first order kernel - Hand activation
Fig. 13.
Hemodynamic parameters for Hand activation
[13] Sadato N, Okada T, Honda M et Al: Cross-model intergartion and plastic changes revealed by lip
movement, random-dot motion and sign languages in the hearing and deaf. Cerebral Cortex 15: 1113-
1122, 2005.
[14] Talairach J. Tournoux P: Co-planar stereotaxic atlas of the human brain. New york, Thieme 1998.
[15] fMRI Brain Mapping Paradigms: Task Requirements and Recommendations - White Paper. Neurog-
nostics, Inc., 1-9, 2006.

---

INTERNATIONAL JOURNAL OF WISDOM BASED COMPUTING, VOL. 1(1), 2011
10
Normalized flow signal
Time in Seconds for Both hands
Fig. 14.
BOLD response for hand movement - 1st order kernal
Time in Seconds for Both hands
Fig. 15.
BOLD response for hand movement - 2nd order kernal

---