Automatic Segmentation of Femur Bone Features and Analysis ...lnse.org/papers/44-IE2029.pdf ·...
Transcript of Automatic Segmentation of Femur Bone Features and Analysis ...lnse.org/papers/44-IE2029.pdf ·...
194DOI: 10.7763/LNSE.2013.V1.44
Abstract—This paper mainly focuses on diagnosing the
osteoporosis using X-Ray images. Osteoporosis is a metabolic
bone disease which is characterized by compromised bone
strength predisposing a person in to an increased risk of
fracture. It is most common in women after menopause where
it is called post menopausal osteoporosis. The condition of
Osteoporosis are decrease in the bone density, decrease in the
bone strength resulting in fragile bones and susceptibility to
fracture. In clinical practice, dual energy X-ray
absorptiometry (DEXA) scanning is the most common way of
measuring BMD levels and it is the current gold standard for
diagnosing and monitoring the osteoporosis. In DEXA
scanning, the method of comparing individual BMD
measurements to the normal is called the ‘T-score’ system.
This project work presents an efficient method to diagnose the
osteoporosis by measuring the bone mineral density (BMD) of
segmented femur bone parts with the help of X-Rays images
and also compare the bone mineral density obtained from the
DEXA scan with the density obtained from the conventional X-
rays.
Index Terms—Bone mineral density (BMD), dual-energy X-
ray absorptiometry (DEXA), t-score, osteoporosis.
I. INTRODUCTION
Bone is a living material which helps to form the skeleton,
thereby enabling the locomotion and giving protection to the
organism [1]. Human body has increasing complex levels of
organization progressing from cells to tissues to organs to
organ systems and finally to the organism. The principal
functions of the skeletal bone are giving mechanical support
and maintain the calcium homeostasis and haematopoiesis in
the bone marrow of an organism.
This can be disturbed by various skeletal disorders.
Osteoporosis is one of the systemic skeletal disorders,
Characterized by low bone density and micro architectural
deterioration of bone tissue with a Consequent increase in
bone fragility [2]. It is three times more common in women
than in men because women have a lower peak femur bone
mass and the hormonal changes that occur at the menopause
stage [3].Femur or thigh bone is most proximal bone of the
leg in tetra pods vertebrates capable of walking and jumping.
It is the largest, longest and strongest bone in our body
extends from the hip to the knee. Four main parts of the
human femur bone having bone minerals are neck, ward,
trochanter and shaft. Dual -energy X-ray Absorptiometry
Manuscript received
January
31,
2013; revised April
25, 2013.
This
work was
supported
in
part
by
the
R.M.K
Engineering
College,
Chennai,
India.
The authors are with R.M.K
Engineering
College,
Chennai,
India
(e-
mail:
[email protected], [email protected],
(DEXA) scans are one of the bone measuring techniques
which are used to diagnose the osteoporosis but it requires
high radiation dose to the patients. It is the current „gold
standard‟ for the diagnosis of osteoporosis [4] .In this
project work, normal x-ray image of femur bone is taken.
Four main parts in that femur bone image are automatically
segmented and measure the bone mineral content of each
part which helps to diagnose the Osteoporosis. The term
“bone mineral content” describes the amount of mineral in
the specific bone site is scanned, from which a value for
BMD can be derived by dividing the bone mineral content
by the area or volume. BMD values are expressed in relation
to the young adult mean called T-score or age-matched
controls called Z-score. . Finally, compare the normal X-ray
image bone mineral density and DEXA image bone mineral
density for their accuracy.
II. FEMUR BONE FEATURES
The femur or thigh bone is the largest, longest and
strongest bone in the human body. The upper end of femur
bone is a part of the hip joint (the largest joint in the body)
and the lower end of femur bone is a part of the knee joint. It
is almost perfectly cylindrical in the greater part. In the erect
posture, it is not look like a vertical cylinder. it is being
separated above from its fellow by a considerable interval,
which corresponds to the breadth of the pelvis but inclining
gradually downwards and medial wards, so its fellow is
approach towards its lower part for the purpose of bringing
the knee-joint near the line of gravity of the body. The
degree of this inclination varies from different persons and
is greater in the female than in the male because women
having pelvis of greater breadth. The femur bone is divisible
into a body, upper and lower extremities like other long
bones. The upper extremity consists of a head, a neck, a
greater and a lesser trochanter.
The Head: The head, which is globular in shape. Its
surface is coated with smooth cartilage in the fresh state
except the fovea captious femoris, which is situated a little
below and behind the center of the head .the head region is
partially embedded in pelvis region. In this project, Hough
filter is used to segment the head region in the femur bone
by detecting the spherical shape .The bone mineral density
of head part in normal femur bone must have 0.846994.
The Neck: The neck is a flattened pyramidal shape which
connects the head with the body and forms a wide angle in
the medial ward. The angle becomes lessened during
puberty. So it forms a gentle curve from the axis of the bone
body. In the adult stage, the neck forms an angle of about
125° with the body. It is inversely proportion to the
development of the pelvis region. For female, the femur
Automatic Segmentation of Femur Bone Features and
Analysis of Osteoporosis
P. Santhoshini, R. Tamilselvi, and R. Sivakumar, Member, IACSIT
Lecture Notes on Software Engineering, Vol. 1, No. 2, May 2013
195
neck forms more nearly a right angle with the body. The
angle decreases during the period of growth but after full
growth has been attained, it does not undergo any change,
even in old age. It varies in different persons who having
same age. In addition, the femur neck is also projecting 12°
to 14°upward and medial ward from the body. The bone
mineral density of neck part in normal femur bone must
have 0.59259.
The Trochanter: The greater Trochanter is a large,
irregular, quadrilateral eminence, situated at the junction of
the neck with the upper part of the body. It is directed a little
lateral ward and Backward. It has two surfaces and four
borders. The lesser Trochanter is a conical eminence which
varies in size. It projects from the lower and back part of the
base of the neck. The bone mineral density of trochanter part
in normal femur bone must have 0.7975.
The Body or Shaft: The body, almost cylindrical in
shape which is little broader than in the center. It is slightly
arched, the shaft part is convex in front and concave behind.
The shaft and the lower extremity is separated from one
another by a smooth shallow depression called the patellar
surface. The bone mineral density of shaft part in normal
femur bone must have 1.12771.The block diagram in Fig. 1,
clearly explains the various stages for measuring bone
mineral densities in femur bone.
Fig. 1. Block diagram.
In this work, Normal X-Ray image of pelvis femur bone
is taken. By using canny Edge-based image segmentation,
femur bone parts are automatically segmented and then
measure the T-score value of BMD of each femur bone parts
which is used to diagnose the Osteoporosis. Finally,
compare the accuracy between normal X-ray and DEXA
scan.
III. SEGMENTATION
Normal X-ray image of femur bone of a 75 aged women
is collected from CETIR centre medic, Barcelona, Spain
through E-mail. The Fig. 2 shows the X-Ray image of left
and right of the femur bone which is along with pelvis
region.
Before segmentation of femur bone x-ray image, the
threshold of each region is to be calculated. It is done by
thin plate spline Transformation (TPS).it has been widely
used for non-rigid Transformation model such as image
alignment and shape matching. The interpolation gets
smoothened with derivatives of any order for its energy
function [5]. Thin spline interpolation is done after the
initial estimation. It helps to calculate the area of each femur
bone parts. The area of each parts is not constant .It is varied
image to image. The rectangle box (red color) in Fig. 3
indicates the area of region wish to measure which is
selected by the user.After thresholding the selected
region,the image of femur bone becomes interpolated which
is shown .
Fig. 2. X-Ray image of femur bone.
Fig. 3. Thin plate spline interpolation.
Usually x-ray image having both strong and weak edges.
The Canny Edge detector is an optimal edge detection
algorithm that can be used to detect a wide range of strong
and weak edges in images .The canny algorithm contains a
number of adjustable parameters, which can reduce the
computation time. It gives good detection, good localization
and minimal response. It is very effective edge detecting
technique as it detects faint edges even when the image is
noisy. If the pixel value is less than the lower threshold then
it is set to 0 and if it is greater than the higher threshold then
it is set to 1. If a pixel falls in between the two thresholds
and is adjacent or diagonally adjacent to a high-value pixel
then it is set to 1, otherwise it is set to 0. The object to be
segmented is differs from the background image. It can be
detected by gradient operators of an image. The calculated
gradient image can be applied to create a binary mask
containing the segmented cell. The Fig. 4 shows the canny
edge detected output of femur bone which helps for
segmentation.
Fig. 4. Canny edge detection.
Segmenting the femur bone parts such as neck, ward,
trochanter and shaft regions. Initially measure the pixel
value inside the blue box for each part in the femur bone x-
ray image. The neck and ward region segmentation is shown
in Fig. 5. The Trochanter and shaft region segmentation is
shown in Fig. 6. It helps to make segmentation easier. After
Lecture Notes on Software Engineering, Vol. 1, No. 2, May 2013
196
segmenting out the regions, calculated the areas of each
femur bone part which is actually needed to measure the
BMD. As the edges are clear it is easier to measure the area.
The area obtained was in pixels, so we used pixel
conversion calculator to attain the area in centimeter square.
In x-ray images of femur bone, the areas defined for the
neck region is a rectangle, for the ward region it is a square
and for the shaft and the trochanter it is a triangle. This
geometrical shape makes comparison more flexible.
Fig. 5. Neck and ward region.
Fig. 6. Trochanter and shaft regions.
The area of region is obtained from the pixel conversion
calculation. The bone mineral content values of the normal
femur and the osteoporotic femur is collected from hospitals.
BMD is a measure of degree and heterogeneity of
mineralization in bone tissue [6]. which is calculated by
bone mineral content divided by area of each femur part.
IV. T-SCORE MEASUREMENT
The T-score is the number of standard deviations below
the average for a young adult at peak bone density [7]. The
prevalence of low bone density in the general population can
be assessed by means of the WHO diagnostic criteria.
According to these criteria, persons with bone density levels
of t-score more than 2.5 standard deviations and below the
young adult reference mean are considered to have
osteoporosis. Persons with bone density below this threshold
who also sustains a fracture meet the definition of
“established or severe osteoporosis” [8].
Fig. 7. BMD measurement.
Fig. 7 shows the boundary detection of femur bone x-ray
image. This boundary detection is not same for all images. It
will vary according to the osteoporotic affected bones [9].
Layout is created using the MATLAB GUIDE, Graphical
user interface development environment [10]. The Fig. 8
shows the first page of the program that can be created using
the GUI, listing down all the femur images.
Fig. 8. GUI layout.
Then we have placed the pushbutton named „Measure
BMD‟. First a particular femur image is selected, and then
clicking on the „Measure BMD‟ button gives the BMD
value of the particular femur. The Fig. 9 shows the GUI
layout page displaying the area, the BMC and the measured
BMD values of the osteoporotic femur. Fig. 10 shows the
GUI layout page displaying the area, the BMC and the
measured BMD values of the normal femur.
Fig. 9. GUI layout for osteoporotic bone.
Fig. 10. GUI layout for normal bone.
The head region is still manually removed by setting the
threshold value. But, in this work, it represents a method
which is potentially removed automatically by incorporating
a Hough filter for detecting spherical objects in head region
which is articulates in pelvis area [5]. Fig. 11 highlights the
head region of femur from pelvis which is done by matlab
using Hough filter.
Lecture Notes on Software Engineering, Vol. 1, No. 2, May 2013
197
Fig. 11. Hough filter output.
Then comparing the BMD values obtained from X ray
image and DEXA image. This is done for both normal
femur image and an osteoporotic femur image and estimates
their accuracy. The BMD values measured from X-ray
image is very close to the BMD values measured from
DEXA image. The BMD Measurement from X-ray image is
very much simpler compared to the DEXA method. This
makes the method much easier as well as efficient. The
Table I compares the BMD values.
TABLE I: BMD COMPARISON
NORMAL FEMUR
OSTEOPOROTIC
FEMUR
REGIONS
DEXA
IMAGE
X-RAY
IMAGE
DEXA
IMAGE
X-RAY
IMAGE
Neck 0.847 0.8469 0.8469 0.654917
Ward 0.593 0.5925 0.5925 0.479726
Trochanter 0.797 0.7975 0.7975 0.507397
Shaft 1.128 1.1277 1.1277 0.780815
The Fig. 12 shows the BMD values obtained from X ray
and DEXA images in chart, showing small variations for all
the regions namely, the neck, the ward, the trochanter and
the shaft of the osteoporotic femur. From the chart it can
conclude that, the BMD measured from X-ray image by this
method does not vary to a great extent for the osteoporotic
femur.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
neck ward troch shaft
DEXA
X ray
Fig. 12. BMD comparison chart.
V. CONCLUSION
Thin Spline Interpolation is mainly used for an intensity
based registration process using thin plate Spline registration
where the landmarks on the reference volume are defined by
the corresponding surface mesh. The WHO defines
osteoporosis as a BMD of 2.5 or more standard deviations
below that of a normal young healthy female (a T score of
≤ .2.5 SD) for postmenopausal women and men over 50
years as measured by DEXA at the femoral neck. Low BMD
is a major risk factor for fragility fracture and other risk
factors may act via their effect on bone mineral density. The
automatic segmentation of femur bone parts include neck,
ward, trochanter and shaft region and then measure the
BMD of proximal femur from X ray images. It is much
simpler than the DEXA method. The results show some
promising correlations with the DEXA scan results both for
normal and osteoporotic femur. The values of BMD from X
ray and DEXA images show very less variations. The
measurement is cost effective. As measurement of BMD is
possible with the conventional X ray image, hospitals need
not replace their X ray machines with high cost DEXA
equipment to diagnose osteoporosis, especially to measure
BMD. Moreover special training to personnel to handle the
DEXA equipment is also not needed. So, it is mainly helpful
for economically weaker people to diagnose the
osteoporosis.
REFERENCES
[1] C. Matsumoto, K. Kushida, K. Yamazaki, K. Imose, and T. Inoue,
“Metacarpal bone mass in normal and osteoporotic Japanese women
using computed X-ray densitometry,” Calcified Tissue International,
vol. 55, no. 5, pp. 324-329, 1994.
[2] WHO, Prevention and management of osteoporosis World Health
Organization Technical Rep., 921, 2003.
[3] The sex and age distributions of population. The 1994 revision of the
United Nations global population estimates and projections, New
York, NY, United Nations, 1995.
[4] Peripheral X-ray absorptiometry in the management of osteoporosis
edited by Dr Rajesh Patel, Head of Academic Bone Densitometry and
Clinical Studies, Imperial College London.
[5] A. Malich, “Early results of the comparison of bone mineral density
values assessed with digital X ray based radiogrammetry and double
energy x ray absorptiometry on patients suffering from rheumatoid
arthritis,” The Internet Journal of Radiology, vol. 5, no. 1, 2003.
[6] T. Whitmarsh, L. Humbert, M. De Craene, L. M. Del R. Barquero,
and A. F. Frangi, “Reconstructing the 3D Shape and Bone Mineral
Density Distribution of the Proximal Femur from Dual-Energy X-Ray
Absorptiometry,” IEEE Transactions on medical imaging, vol. 30, no.
12, December 2011.
[7] R. McKinney et al., Author affiliations appear on, pp. 2288–2289.
[8] E. Shane et al., Atypical Subtrochanteric and Diaphyseal Femoral
Fractures: Report of a Task Force of the American Society for Bone
and Mineral Research.
[9] J. A. Kanis et al., “Assessment of fracture risk and its application to
screening for postmenopausal osteoporosis,” Synopsis of a WHO
Report. Osteoporosis International, 1994, vol. 4, pp. 368–381.
[10] S. Simakov, “Introduction to MATLAB Graphical User Interfaces,”
Maritime Operations Division Defence Science and Technology
Organization DSTO–GD–044.
P. Santhoshini is pursuing M.E. Applied Electronics
Final Year, having 9.1 CGPA in R.M.K Engineering
College, Chennai, India .In 2011 she graduated in
Electronic and communication Engineering at C.Abdul
Hakeem College of Engineering and Technology,
Vellore, India with Distinction. During her College
days, She Won Third Prize for Presented a Paper
named as “A VLSI Architecture of an Invisible
Watermarking Unit for a Biometric Based Security System in a Digital
Camera” at the Student‟s National Level Technical Symposium
(IMPETUS‟10) in Mount Zion College of Engineering And Technology.
She attended two day workshop on “Signal and Image Processing” in VIT
University, Vellore. She Underwent Project Work Training on
“Development of Analog Data Acquisition for Controlling ESP Using AVR
Micro Controller” in BHEL, Ranipet
R. Tamilselvi is an Assistant Professor in the
Department of Electronics and Communication
Engineering at R.M.K Engineering College,
Tamilnadu, India. She has been teaching in the
Electronics and Communication field since 2004.
She is pursuing her PhD in College of Engineering
Guindy, Anna University Chennai. She obtained her
Master‟s degree from sastra university, Thanjavur.
Her research interests include Medical Image
Lecture Notes on Software Engineering, Vol. 1, No. 2, May 2013
198
Processing and wireless communication. She has published over 3 journal
and 8 international conference papers over the last few years. Tamilselvi is
a life member of Biomedical Engineering Society of India.
R. Sivakumar is a Professor and Head of
Department of Electronics and Communication
Engineering at R.M.K Engineering College,
Tamilnadu, India. He has been teaching in the
Electronics and Communication field since 1997. He
obtained his Master‟s degree and PhD from College
of Engineering Guindy, Anna University Chennai.
His research interests include Bio Signal Processing,
Medical Image Processing, wireless body sensor networks and VLSI. He
has published over 22 journal and 35 conference papers over the last
several years. He has taught a wide variety of Electronics courses including
Digital Image Processing, Multimedia Compression Techniques, VLSI
Design, Medical Electronics and Electronic Circuits. Dr. Siva is a life
member of the Indian Society of Technical Education, Senior Member of
IACSIT and a member of IEEE. Dr. Siva has been invited to Chair and
speak at various conferences; more recently, he was Conference Chair at
the ICCTS 2012 in Delhi & ICIAE 2012 in Bangalore.
Lecture Notes on Software Engineering, Vol. 1, No. 2, May 2013