Download Automated Screening of Fetal Heart Chambers from 2

24 International Journal of Biomedical and Clinical Engineering, 1(2), 24-33, July-December 2012
Automated Screening of
Fetal Heart Chambers
from 2-D Ultrasound CineLoop Sequences
N. Sriraam, Department of Medical Electronics, M.S Ramaiah Institute of Technology,
Bangalore, Karnataka, India
S.Vijayalakshmi, Department of Biomedical Engineering, Rajalakshmi Engineering College,
Chennai, India & Trivitron Medical Systems, Chennai, Tamil Nadu, India
S.Suresh, Mediscans Pvt Ltd, Chennai, Tamil Nadu, India
ABSTRACT
Fetal cardiac ultrasonic imaging technique has become increasingly popular in the recent years for the detection of fetal congenital abnormalities at an early stage. Due to the low signal-to-noise ratio of the ultrasound
imaging, the automatic detection methods should incorporate suitable preprocessing filtering techniques to
enhance the segmentation techniques efficiently. This paper suggests the application of median and morphological filtering operation for removing speckle noise. Then the four heart chambers are segmented independently
based on the shape priors. The amorphous snake’s helps in identifying the contours of the chamber edges
individually based on shape pattern. Experimental study involves the ultrasound cine-loop sequences of the
apical four chamber view of fetal heart with a constant frame rate of 25 frames per second (fps) with varying
duration of 10-30s encompassing a range of 20-40 complete cardiac cycles. The simulation result confirms
the suitability of proposed scheme for automated screening of fetal heart chambers.
Keywords:
Amorphous Snakes, Apical Four Chamber View, Cine-Loop Sequence, Congenital Abnormalities,
Contours, Fetal Heart Biometry
INTRODUCTION
Congenital heart defects happen because of incomplete or abnormal development of the fetus’
heart during the very early weeks of pregnancy.
Some are known to be associated with genetic
disorders, such as Down syndrome, but the cause
of most congenital heart defects is unknown.
The major congenital heart diseases have
characteristic abnormalities seen in the fetal
ultrasound images taken at different trimesters
of pregnancy. The sweep technique involves
DOI: 10.4018/ijbce.2012070103
Copyright © 2012, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
International Journal of Biomedical and Clinical Engineering, 1(2), 24-33, July-December 2012 25
sweeping the transducer beam in a transverse
plane from the level of the four-chamber view
towards the fetal neck. By doing so, the outflow
tract vessels are observed. The sweep consists
of the following views: four-chamber view,
five-chamber view, main pulmonary artery
or 3-vessel view, and the tracheal view. In a
fetal heart examination using ultrasound, the
abdominal view, the four-chamber view and
both (left and right) cardiac outflow tracts
should be obtained at screening level (Achiron
et al., 1992). Additional fetal cardiac views are
crucial for the sequential segmental analysis to
assess the connections and anatomical detail
(Jantarasaengaram, 2010).
The diagnosis of fetal heart requires highly
skilled operations and is often time-consuming
for doctors (Siqueria et al., 2005). From this
point of view, the state-of-the-art technology
in fetal cardiac ultrasound examination is that
computerized methods are proposed and they
are trying to assist the doctors in the diagnosis of
anomalies in fetal heart. The present scenario is
that the fetal cardiac images have been evaluated
manually. This might usually lead to human error
during observation. This can be overcome by
using image processing algorithms to set a standard to evaluate the images automatically. This
can result in more appropriate interpretation of
the images. A conventional 2-dimensional [2D]
ultrasound transducer without position sensing,
freehand acquisition using a conventional 2D
ultrasound transducer with position sensing, and
automated acquisition using dedicated mechanical volume probes) rely on the acquisition of a
series of 2D frames that are then reassembled
by the ultrasound equipment and displayed as
a 3D volume data set (Pretorius et al., 2001).
Two dimensional ultrasound methods have
traditionally relied on both static and real-time
imaging to understand fetal cardiac anatomy
and function (Belohlavek et al., 1993)
Vibhakar Shrimali et al (2009) carried out a
cross-sectional study on fetal ultrasound images
that were processed using morphological opera-
tors to obtain the shape of the femur. In Zayed
et al (2001) fuzzy based clustering algorithm
is applied for grouping together those pixels in
the image, which have similar features in the
feature space. Lassige et al (2000) used the
level set algorithm to detect the septal defects.
Siqueira et al [3] proposed to apply the selforganizing map to segment the fetal heart and
obtain the heart structure. These computerized
methods are mainly based on the information
of edge or region, which is not reliable for the
ultrasound data of early trimester fetal heart
(Deng et al., 2007)
This paper discusses the automated screening of fetal heart chamber based on the 2-D cine
loop sequences obtained from the ultrasonic
imaging. Of the different views such as four
chamber, five chamber and three vessel views,
apical four-chamber view is considered for
this study. A cine-loop sequence is similar to a
video which is recorded for a prolonged time
of several seconds to few minutes. Based on
the requirements, various cine-loop sequences
of 10 to 30 seconds duration encompassing the
diastole and systole phase of the heart is taken
for the study. The frame rate for the selected
cine-loop sequence is 25 frames per second
(fps) based on the hardware features of the
ultrasound machine used.
The procedure involves the application
of median and morphological filters as preprocessing step to remove the speckle noise
from the sequences. With the region of interest
(the chamber region alone) being focused, the
shape priors are generated to segment all the
four chambers using contours. The amorphous
snakes or active contours then identify the
edges of the chambers thereby yielding efficient segmentation of all the four chambers
for clinical diagnosis. For the analysis, 8 cineloop sequences belonging between 22 and 25
weeks of gestation are taken. The model of the
machine used here for acquiring the ultrasound
cine-loop sequence of the fetal heart is Voluson
E8 Expert from GE Healthcare.
Copyright © 2012, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
8 more pages are available in the full version of this
document, which may be purchased using the "Add to Cart"
button on the publisher's webpage:
www.igi-global.com/article/automated-screening-of-fetalheart-chambers-from-2-d-ultrasound-cine-loopsequences/86049
Related Content
Evidence-Based Medicine: A New Approach in the Practice of Medicine
Nilmini Wickramasinghe, Sushil K. Sharma and Harsha P. Reddy (2009). Medical
Informatics: Concepts, Methodologies, Tools, and Applications (pp. 2377-2386).
www.irma-international.org/chapter/evidence-based-medicine/26379/
Prospective Memory Impairment in Remembering to Remember in Mild
Cognitive Impairment and Healthy Subjects
Nobuko Ota, Shinichiro Maeshima, Aiko Osawa, Miho Kawarada and Jun Tanemura
(2011). Early Detection and Rehabilitation Technologies for Dementia: Neuroscience
and Biomedical Applications (pp. 98-106).
www.irma-international.org/chapter/prospective-memory-impairmentremembering-remember/53426/
Medical Education in the 21st Century
Stefane M. Kabene (2009). Medical Informatics: Concepts, Methodologies, Tools,
and Applications (pp. 178-185).
www.irma-international.org/chapter/medical-education-21st-century/26215/
Monitoring Hospital Patients Using Ambient Displays
Monica Tentori, Daniela Segura and Jesus Favela (2009). Mobile Health Solutions for
Biomedical Applications (pp. 143-158).
www.irma-international.org/chapter/monitoring-hospital-patients-usingambient/26770/
Model Simulating the Heat Transfer of Skin
Anders Jarløv and Tim Toftgaard Jensen (2014). International Journal of Biomedical
and Clinical Engineering (pp. 42-58).
www.irma-international.org/article/model-simulating-the-heat-transfer-ofskin/127398/