Evolution Prediction of the Aortic Diameter Based on the Thrombus Signal from MR Images on Small Abdominal Aortic Aneurysms

Document Sample
Evolution Prediction of the Aortic Diameter Based on the Thrombus Signal from MR Images on Small Abdominal Aortic Aneurysms Powered By Docstoc
					                                                              (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                Vol. 9, No. 3, March 2011

Evolution Prediction of the Aortic Diameter Based on
  the Thrombus Signal from MR Images on Small
           Abdominal Aortic Aneurysms
                                     A. Suhendra1, C.M. Karyati2, A.Muslim3, A.B. Mutiara4
                            Faculty of Computer Science and Information Technology, Gunadarma University
                                             Jl. Margonda Raya No.100, Depok 16424, Indonesia

Abstract—The paper is about studying the T1 and T2 from                   parts of the human body. The human blood pressure will refer
Magnetic Resonance (MR) Images examination for the existence              to how much pressure in the arteries that brings blood to all
of thrombus in patient with Small Abdominal Aortic Aneurysms              cells of the human body through the delicate vessels
(SAAA) in order to know whether thrombus signal has                       (capillaries) which then will return to the heart through blood
correlation with the evolution of aortic diameter enlargement,            vessels and takes oxygen through the lungs. There are a little
which then can be used to predict the risk of rupture of aortic           description of the aorta which will be discussed further in this
wall. Data were derived from 16 patients with SAAA, whereas               study. It could be imagined if there are any damage to the
MR images obtained from 3T imager (Trio TIM, Siemens                      human aorta would result in abnormalities in blood flow in the
Medical Solution, Germany), which came from: the study of
                                                                          human body. In the following image, we can see the anatomy
anatomy, cine-MR images, pictures T1/T2, blood flow images,
and images after injection of contrast agents. The surface area of
                                                                          of the aorta and the arteries (figure 1) :
the aorta and luminal are determined by tracing manually, which
can be used to determine the surface area of thrombus. The
maximum diameter of the aorta are automatically obtained from
manual tracing on T1 images. The parameters to study the
thrombus signal are the mean, median, standard deviation,
skewness and kurtosis. Each parameter is calculated on the area
of thrombus, while for normalization we implement the signal in
the muscles. All parameters are compared to evolution of aortic
diameter. We found 13 out of 16 patients with SAAA have
thrombus. But there is no correlation between thrombus signals
and maximum diameter (mean (r = 0.318), median (r = 0.318),
skewness (r = 0.304)), or even with maksimum evolution diameter
(mean (r=0.512)). As the conclusion is the comparation between
mathematical and visual calculation of thrombus categories
reached 81% similar, but thrombus signal itself cannot be used to
                                                                                           Figure 1. Anatomy of the aorta [1]
predict the evolution of aortic diameter.
                                                                              The Studies of human aorta have been conducted and
   Keywords-component; Thrombus signal; evolution of aortic
diameter; T1 and T2 weighted images; Small Abdominal Aortic               successfully detected abnormalities in the aortic wall, both at
Aneurysms.                                                                the thoracic or abdominal aortas [1,2]. In general, the swelling
                                                                          of the aortic wall is very elastic, therefore if the swelling is
                                                                          occur then aortic wall will not be able to shrink back and it will
                       I.      INTRODUCTION                               be broken without being able to predict when the rupture risk
    Aorta is the larger artery that delivers blood from the heart         of the aortic wall. It could be in the risk of patient death.
of human beings throughout the body. In this way, the human
                                                                              An Abdominal Aortic Aneurysm, also called AAA, is a
blood flow will go through some branch, for example, that led
                                                                          bulging area in the wall of the aorta which is causing of an
to the arm (subclavian arteries), heading toward of the head
                                                                          abnormal widening or ballooning until greater than 50 percent
(carotid arteries), and headed toward of the chest (thoracic
                                                                          of the normal diameter. The the swelling of the aortic wall
aorta), then toward of the diaphragm to the stomach
                                                                          could be caused by age (more than 60), male (four to five times
(abdominal aorta). In the region around the stomach will be
                                                                          greater than females), family history (first degree relatives such
much more branching, including to the liver, intestines and
                                                                          as father or brother), genetic factors, hyperlipidemia (elevated
kidneys. And last, the branching will be forwarded to the
                                                                          fats in the blood), hypertension (high blood pressure), smoking
direction of human legs (iliac arteries).
                                                                          and diabetes.
   Human blood will be pumped by the heart into the aorta,
which then flows through the artery and its ramifications to all

                                                                                                      ISSN 1947-5500
                                                                (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                  Vol. 9, No. 3, March 2011
    Asymptomatic aneurysms may not require surgical                           According to the result of clinical data, there are difference
intervention until they reach a certain size or are noted to be            characteristics based on status of each patient (smoking/ex
increasing in size over a certain period of time. The parameters           smoking, fat in blood (dyslipidemied), and hypertency) as
for surgical decisions, but are not limited to, are as follows             shown in Table I.
   •    aneurysm size greater than 5 centimeters (about two
   •    aneurysm growth rate is arround 0.5 centimeters
        (slightly less than one-fourth inch) over a period of six
        months to one year
   •    patient’s ability to tolerate the procedure

                       II.   TROMBUS SIGNAL
    Thrombosis term will refer to the formation of a blood clot
(thrombus) in the blood vessels or human heart cavities.
Abdominal Aortic Aneurysms are often associated with the                                                                               (a)
thrombus (clots). This field have been studied and
demonstrated by the pathological, surgical, and clinical
examination based on the results of computed tomography
(CT), ultrasound imaging, angiography, traditional spin-echo
(SE) or cine-MRI. There are many methods have been created
or modified to prove the existence of intact thrombus signal in
the aorta. But until now, with a disorder that occurs in the
aorta, it is difficult to detect or properly evaluate the existence
of thrombus signal [2, 3].


                                                                           Figure 3. (a) T1- image and (b) T2- image at the level of Abdominal Aortic

                                                                                         TABLE I.        PATIENT’S CHARACTERISTICS

         Figure 2. Aneurysms with a formation of Thrombus [4]                                                   Age
                                                                                  Name of Patient     Sex                   Characteristics
    The selection of images for thrombus formation analyzing                      Patient 1          Male      65       Smooking
is very important. Images are selected from the result of
examination during relaxation took place (as shown in Figure 3                    Patient 2          Female 68          Dyslipidémie
of T1 and T2 images)[5].                                                                                                Smooking, Hypertensi,
                                                                                  Patient 3          Male      62
    This work analysed the T1 and T2 of thrombus of SAAA
patient examination to determine whether the thrombus signal                      Patient 4          Male      82       Ex Smooking
has correlation with the aortic diameter enlargement, and to                      Patient 5          Male      83       -
predict the rupture risk of the aorta wall.
                                                                                  Patient 6          Male      59       Ex Smooking

                III.    MATERIALS AND METHODS                                     Patient 7          Male      53       -
                                                                                                                        Ex Smooking,
A. Data                                                                           Patient 8          Male      79       Hypertensi,
    Data were obtained from 16 patients with Small Abdominal                                                            Dyslipidémie
Aortic Aneurysms (SAAA) who have been examined since                                                                    Ex Smooking,
July 2006 until January 2010. Each patient has been examined                      Patient 9          Male      77       Hypertensi,
at least 1 to 4 times with examination period between 6 to 12                                                           Dyslipidémie
months (depend on the patient). MR Images were acquired on a                      Patient 10         Male      71       Smooking,
3T Imager (Trio TIM, Siemens Medical Solution, Germany).

                                                                                                            ISSN 1947-5500
                                                                           (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                             Vol. 9, No. 3, March 2011
        Name of Patient     Sex                  Characteristics
        Patient 11         Female 74         Ex Smooking
        Patient 12         Male     69       -
                                             Ex Smooking,
        Patient 13         Male     55       Hypertensi,                                                                   (a)                             (b)
                                             Ex Smooking,
        Patient 14         Male     51
                                             Ex Smooking,
        Patient 15         Male     73       Hypertensi,
        Patient 16         Male     59       Smooking

                                                                                      Figure 5. (a) Anterior-Posterior Diameter, (b) Transversal Diameter, (c)
B. Protocol Small Abdominal Aortic Aneurysms                                                                   Maximum Diameter
   In this study protocol, images originating from: the study of
anatomy, cine-MR images for 3D/4D modeling, images T1/T2,
blood flow images, and images after injection of contrast
agents have been used to study the aspects of inflammation.
For each patient, the images are located in the same position
between one to another examination.

C. Processing
   We used MatLab software to precess the data. Preliminary
examination has been conducted for predictive aspect, and final                          (a)                                                     (b)
examination has been conducted as well for data which has
more important thrombus, more areas, and more signals. The
borders have been manually traced to define the Aorta Surface
and Luminal Surface, therefore Thrombus Surface = Aorta
Surface – Luminal Surface, (see figure 4).
    In aortic wall surface calculation, thrombus is found if the
thrombus surface area is greater than 30% of aortic surface                                                          (c)
area. Diameter of aorta is achieved by tracing manually the
aorta surface. There are three kinds of diameter positions:                           Figure 6. (a) T1-W image and (b) T2-W image after manual tracing, (c)
                                                                                                        Normalization area in the muscle
Anterior-Posterior Diameter, Transversal Diameter and
Maximum diameter, as shown in the figure 5.
                                                                                    D. Paramaters
  The muscle signal are slightly differences among each
examinations, therefore we normalized the data of muscle area.                          Maximum aortic diameter was automatically obtained from
                                                                                    manual tracing on T1 image in all examinations. Then we
                                                                                    calculated the evolution of the aortic diameter (mm/year) = ∆
                                                                                    maximum diameter (mm) / ∆ examination date (day) * 365
                                                                                    days. Several parameters were used to study the thrombus
                                                                                    signal, such as mean, median, standard deviation, skewness that
                                                                                    describes the degree of asymmetry of the signal histogram by
                                                                                    using the equation ∑ni(xi-x)3/Ns3, and the kurtosis that
                                                                                    describes how sharp the peak of the signal histogram which is
                                                                                    defined by using the equation ∑ni(xi-x)4/Ns4-3, where ni is
                                                                                    number of pixel at aorta xi , x is mean value of the aorta, s is the
                                  (a)                                (b)            SD, and N is the total number of pixels. [5]
                                                                                        Each parameter is calculated for the thrombus area, and the
Figure 4. (a) Manual tracing in Aorta Surface, (b) Manual tracing in Luminal        signal in the muscle is used to normalize the mean of signal in
                           Surface (in green line)
                                                                                    thrombus, the median of signal in thrombus and the standard
                                                                                    deviation of signal in thrombus. These parameters are
                                                                                    compared to the evolution of the aortic diameter. By using

                                                                                                                    ISSN 1947-5500
                                                                       (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                         Vol. 9, No. 3, March 2011
mean/median/SD signal of the aorta and normalized                                       Patient     Thrombus Categories      Thrombus Categories
mean/median/SD signal of muscle, the thrombus is classified as                                      (based on parameters)   (based on visualization)
follow: Homogeneous Thrombus (if T1 = T2 = Low signal);                                Patient 5    Heterogeneous           Heterogeneous
                                                                                       Patient 6    Indefinite              Indefinite
Heterogeneous Thrombus (if T1 = T2 = High signal); and                                 Patient 7    Heterogeneous           Heterogeneous
Indefinite Thrombus (if T1 ≠ T2 (low and high signal, or high                          Patient 8    Homogeneous             Heterogeneous
and low signal)).                                                                      Patient 9    Heterogeneous           Heterogeneous
                                                                                       Patient 10   Heterogeneous           Heterogeneous
                  IV.   RESULTS AND DISCUSSION                                         Patient 11   Homogeneous             Heterogeneous
                                                                                       Patient 12   Homogeneous             Heterogeneous
    We found 13 out of the 16 patients with SAAA have a                                Patient 13   Homogeneous             Homogeneous
thrombus. Figure 7 and 8 are a sample of T1 image which can                            Patient 14   Heterogeneous           Heterogeneous
describe about presence of thrombus in SAAA.                                           Patient 15   Heterogeneous           Heterogeneous
                                                                                       Patient 16   Homogeneous             Homogeneous

   Figure 7. Surface thrombus : 243mm² (11,6%)      without thrombus                                                                      (a)


                                                                                  Figure 9. P13, Male, 55, ex smooking, hypertensi, dyslipidémie, ∆ Max
                                                                                Diameter = 2.80 mm/year, 40% surface thrombus, Homogeneous T1 = T2 =
    Figure 8. Surface thrombus : 1026mm² (48,4%)      with thrombus
                                                                                              Low, T1= 0.391 < 0.815, (b) T2= 0.327 < 0.788

    Based on height’s distribution of thrombus signal, there
were 3 patient without thrombus, 5 patiens with homogenous
thrombus, 7 with heterogeneous thrombus and 1 with indefinite
thrombus. Figure 9, 10, and 11 shows the categories of
thrombus presence. If we compare the used of parameters to
the visual, there were 3 differences of the result of thrombus
categories as shown in Table II. There are three categories are
different (patient number 8, 11, 12). It indicates that 81,25% of
thrombus categories determination using parameters are the
same with the result of based on visualization.
       Patient     Thrombus Categories      Thrombus Categories
                   (based on parameters)   (based on visualization)
      Patient 1    Without thrombus        Without thrombus
      Patient 2    Without thrombus        Without thrombus
      Patient 3    Homogeneous             Homogeneous
      Patient 4    Without thrombus        Without thrombus

                                                                                                               ISSN 1947-5500
                                                                     (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                       Vol. 9, No. 3, March 2011
                                                                                 and Table IV). From those tables, there are many values of R <
                                                                                 0.3 (not good correlation), a few values of R > 0.3, which
                                                                                 indicates a correlation between thrombus signals and the
                                                                                 evolution of the aortic diameter.

                                                                                            TABLE III.       PARAMETERS VS MAXIMUM DIAMETER

                                                                                   Name of                    T1                              T2
                                                        (b)                                       R²     R         Equation       R²      R        Equation
                                                                                 Mean/Mean   0.099 0.314 y = 0.030x – 0.010 0.098 y = 0.005x +
 Figure 10. P5, Male, 83, ∆ Max Diameter = 2.27 mm/year, 90.85% surface          Muscle                  0.325                    0.485
thrombus, Heterogeneous T1 = T2 = High, (a) T1= 2.675 > 0.815, (b) T2 =          Mean/Median 0.099 0.314 y = 0.030x – 0.045 0.212 y = 0.012x +
                              0.881> 0.788                                       Muscle                  0.329                    0.253
                                                                                 Mean/SD     0.006 0.078 y = 0.041x + 0.071 0.266 y = 0.145x +
                                                                                 Muscle                  4.002                    6.542

                                                                                 Median/Mean 0.101 0.318 y = 0.031x – 0.009 0.097 y = 0.005x +
                                                                                 Muscle                    0.364                    0.471
                                                                                 Median/Median 0.101 0.318 y = 0.031x – 0.045 0.212 y = 0.011x +
                                                                                 Muscle                    0.368                    0.249
                                                                                 Median/SD     0.008 0.089 y = 0.045x + 0.063 0.252 y = 0.136x +
                                                                                 Muscle                    3.774                    6.483

                                                                                 SD/Mean      0.055 0,234 y = 0.005x + 0.004 0.063 y = -0.002x +
                                                        (a)                      Muscle                   0.038                     0.340
                                                                                 SD/Median    0.055 0,234 y = 0.005x + 0.005 0.069 y = 0.005x +
                                                                                 Muscle                   0.0383                    0.176
                                                                                 SD/SD Muscle 0.000 0.001 y = -0.000x + 0.000 0.005 y = 0.001x +
                                                                                                          1.550                     4.873

                                                                                 Skewness       0.093 0.304 y = -0.011x + 0.015 0.121 y = -0.007x +
                                                                                                            0.549                     0.669
                                                                                 Kurtosis       0.000 0.010 y = -0.000x + 0.040 0.200 y = -0.025x +
                                                                                                            2.321                     3.649

                                                                                            TABLE IV.    PARAMETERS VS ∆ MAXIMUM DIAMETER

                                                                                   Name of                    T1                              T2
Figure 11. P6, Male, 59, ex smoking, ∆ Max Diameter =1.33 mm/year, 6.93%                          R²     R         Equation       R²      R        Equation
surface thrombus, Indefinite T1 = Low ≠ T2 = High, (a) T1= 0.691 < 0.815,
                            (b) T2 = 0.853 > 0.788                               Mean/Mean   0.262 0.512 y = 0.024x +           0.029 0.171 y = -0.004x +
                                                                                 Muscle                  0.931                              0.717
Then all parameters for generate thrombus categories that have                   Mean/Median 0.001 0.028 y = -0.010x +          0.019 0.134 y = -0.024x +
                                                                                 Muscle                  1.125                              0.843
correlation with the evolution of aortic diameter compared
with parameters that don’t have correlation with the evolution                   Mean/SD        0.031 0.176 y = -0.101x +       0.014 0.112 y = -0.160x +
                                                                                 Muscle                     5.441                           12.953
of aortic diameter which is indicated by many occurrence of
value r <0.3 (r is the coefficient of determination on the
graph). But there are also some parameters indicate a linear                     Median/Mean 0.000 0.02       y = -0.007x +     0.019 0.137 y = -0.018x +
                                                                                 Muscle                       1.109                         0.705
correlation between thrombus signal with a maximum
                                                                                 Median/Median 0.001 0.022    y = -0.008x +     0.019 0.137 y = -0.022x +
diameter, where the mean value (r = 0.314), median (r =                          Muscle                       1.123                         0.798
0.318), skewness (r = 0.304), or thrombus signal with the                        Median/SD      0.024 0.154 y = -0.091x +       0.013 0.114 y = -0.156x +
evolution of maximum diameter (mean (r = 0.512).                                 Muscle                     5.391                           12.437

    But there are some parameters that showed a linear
correlation between thrombus signal with a maximum diameter                      SD/Mean        0.002 0.044 y = 0.003x +        0.022 0.148 y = -0.011x +
                                                                                 Muscle                     0.250                           0.327
(mean (r = 0.314), median (r = 0.318), skewness (r = 0.304)) or
                                                                                 SD/Median      0.002 0.04 y = 0.002x +         0.027 0.163 Y = -0.013x +
correlation between thrombus signal and the evolution of                         Muscle                     0.252                           0.361
maximum diameter (mean (r = 0.512) (as shown in Table III

                                                                                                                   ISSN 1947-5500
                                                                  (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                    Vol. 9, No. 3, March 2011
SD/SD Muscle 0.005 0.07    y = 0.010x +   0.010 0.102 y = -0.060x +                                      REFERENCES
                           1.266                      5.277
                                                                           [1]   Health Library, Aneurysm Overview, New York-Presbyterian Hospital,
                                                                                 30 Novembre 2008
Skewness     0.046 0.215 y = -0.021x +    0.000 0.017 y = 0.003x +         [2]   Shin Ito, Koichi Akutsu, Yuiichi Tamori, Shingo Sakamoto, Tsuyoshi
                         0.178                        0.479                      Yoshimuta, Hideki Hashimoto, Satoshi Takeshita, Differences in
                                                                                 Atherosclerotic Profiles Between Patients With Thoracic and Abdominal
Kurtosis     0.021 0.146 y = -0.016x +    2.000 0.005 y = 0.002x +               Aortic Aneurysms, The American Journal of Cardiology, Volume 101,
                         2.367                        2.855                      Issue 5, Pages 696-699, 1 March 2008
                                                                           [3]   Toshio Honda, Mareomi Hamada, Yuji Matsumoto, Hiroshi Matsuoka,
                                                                                 Kunio Hiwada, Diagnosis Thrombus and Blood Flow in Aortic
                                                                                 Aneurysms by use of Tagging Cine MRI, International Journal of
                          V.   CONCLUSION                                        Angiology, 6:203-206, 1997
                                                                           [4]   M. Xavier, A. Lalande, P.M.Walker, C. Boichot, A. Cochet, O. Bouchot,
    The similarity level between thrombus catagory and                           E. Steinmetz, L. Legrand, F. Brunotte, Dynamic 4D Blood Flow
visualization is 81.25%. There are no good correlation between                   Representation in the Aorta and Analysis From Cine-MRI in Patients,
thrombus signal and evolusi of aortic diameter on SAAA                           Computers in Cardiology, 34:375−378, 2007
(many values of r < 0.3), therefore thrombus signal itself                 [5]   Marco Castrucci et al, Mural Thrombi in Abdominal Aortic Aneurysms:
cannot be used as parameter to predict the evolution of aortic                   MR Imaging Characterization-Useful before Endovascular Treatment ?,
diameter. The correlation between blood flow, thrombus signal                    RSNA, 197, Italy, October 1995
and bilogy is still studied. For the next research, we will                [6]   Christopher M. Kramer, Magnetic Resonance Imaging Identifies the
                                                                                 Fibrous Cap in Atherosclerotic       Abdominal Aortic Aneurysm ,
implement other comparation parameters to aortic diameter,                       Circulation ; 109 ; 1016-1021, 2004
such as: blood flow speed with 3D/4D modeling (The aspect of               [7]   Eric M. Isselbacher, Thoracic and Abdominal Aortic Aneurysms ,
laminar and turbulance, maximum spped, radial speed, and                         Circulation, 2005
shear stress). and othehr parameters are comparison our                    [8]   Michèle Coutard, Thrombus versus Wall Biological Activities in
thrombus categories with the visualization categories is 81%.                    Experimental Aortic Aneurysms, Journal of Vascular Research,2009
For the evolution of the aortic diameter, we found no                      [9]   Shin Matsuoka, Quantification of Thin-Section CT Lung Attenuation in
correlation between thrombus signals with the evolution of the                   Acute Pulmonary Embolism : Correlations with Arterial Blood Gas
aortic diameter in Small AAA (R < 0.3), but the parameters                       Levels and CT Angiography, American Roentgen Ray Society,
                                                                                 186:1272-1279, May 2006
were used. The methodologies to measure and the
normalization area with muscle signal will be discussed. We
cannot use thrombus signal alone as a parameter to predict the                                        AUTHORS PROFILE
evolution of the aortic diameter. Relationship between flow
data, thrombus signal and biology findings will be studied.                A. Suhendra is a Lecturer of Informatics Engineering, Industrial Engineering
                                                                                Fakulty of Industrial engineering, Gunadarma university.
    Currently, comparison of the blood flow velocity with                  C. M. Karyati, Graduate from Master Program in Information System,
3D/4D modeling (aspect laminar flow and turbulence,                             Gunadarma Unviversity, 1998. She is now a Ph.D-Student at Groupe
maximum speed, radial speed, and shear stress) with evolution                   Imagerie Médicale, Le2i, UMR CNRS 5158, Faculté de Médecine,
                                                                                Université de Bourgogne, Dijon, France
of the maximum diameter was performed.
                                                                           A Muslim, Graduate from Master Program in Information System,
                                                                                Gunadarma Unviversity, 1997. He is a Ph.D-Student at Groupe Database
                    ACKNOWLEDGEMENTS                                            Sistem Information et Image, Le2i, UMR CNRS 5158, Faculté de
                                                                                Science et L’Enginer, Université de Bourgogne, Dijon, France
This research was conducted because of aid from the MRI and
                                                                           A. B. Mutiara is a Professor of Computer Science at Faculty of Computer
Nuclear Medicine Department at the Centre Hospital                              Science and Information Technology, Gunadarma University
Universitaire (CHU) de Bocage in Dijon, France. More
specifically the authors want to thank Nicolas Abello who has
been so helpful in terms of procurement data. A.B.M. also
gratefully acknowledges financial support of the Gunadarma
Education Foundation during the research.

                                                                                                           ISSN 1947-5500