Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques

Abstract

Accurate medical image processing plays a crucial role in several clinical diagnoses by assisting physicians in timely treatment of wounds and mishaps. Medical doctors in the hospitals generally rely on examining bone x-ray images based on their expertise, knowledge and past experiences in determining whether a fracture exist in bone or not. Nevertheless, majority of fractures identification methods using X-rays in the hospitals is beyond human understanding due to variation in different attributes of fracture and complication of bone organization thereby making it difficult for doctors to correctly diagnose and proffer adequate treatment to patient ailments. The need for robust diagnostic image processing techniques for image segmentation for different bone structures cannot be overemphasized. This research implemented different image segmentation techniques on a bone x-ray image in order to identify the most efficient for timely medical diagnosis. Also, the strength and weaknesses of the diverse segmentation techniques were also identified. This will empowered researchers with appropriate knowledge needed to improve and build better image segmentation models which doctors can use in handling complex medical image processing problems. Also, miss rate in bone X-rays that contains multiple abnormalities can be lowered by using appropriate image segmentation techniques thereby improving some of the labor intensive work of medical personnel during bone diagnosis. MATLAB 9.7.0 programing tool was used for the implementation of the work. The results of X-ray bone segmentation revealed that active contour model using snake model showed the best performance in detecting boundaries and contours of regions of interest when used in segmenting Femur bone image than the other medical image segmentation approaches implemented in the work.

Full text

I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
Published Online October 2021 in MECS (http://www.mecs-press.org/)
DOI: 10.5815/ijigsp.2021.05.03
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
Segmentation of Medical X-ray Bone Image
Using Different Image Processing Techniques
Folasade Olubusola Isinkaye1, Abiodun Gabriel Aluko2, Olayinka Ayodele Jongbo3
1,2,3 Department of Computer Science, Ekiti State University, Ado-Ekiti, Nigeria.
Email: 1folasade.isinkaye@eksu.edu.ng, 2gabrielaluko29@gmail.com, 3yinkutech1@gmail.com
Received: 09 June 2021; Accepted: 28 July 2021; Published: 08 October 2021
Abstract: Accurate medical image processing plays a crucial role in several clinical diagnoses by assisting physicians
in timely treatment of wounds and mishaps. Medical doctors in the hospitals generally rely on examining bone x-ray
images based on their expertise, knowledge and past experiences in determining whether a fracture exist in bone or not.
Nevertheless, majority of fractures identification methods using X-rays in the hospitals is beyond human understanding
due to variation in different attributes of fracture and complication of bone organization thereby making it difficult for
doctors to correctly diagnose and proffer adequate treatment to patient ailments. The need for robust diagnostic image
processing techniques for image segmentation for different bone structures cannot be overemphasized. This research
implemented different image segmentation techniques on a bone x-ray image in order to identify the most efficient for
timely medical diagnosis. Also, the strength and weaknesses of the diverse segmentation techniques were also identified.
This will empowered researchers with appropriate knowledge needed to improve and build better image segmentation
models which doctors can use in handling complex medical image processing problems. Also, miss rate in bone X-rays
that contains multiple abnormalities can be lowered by using appropriate image segmentation techniques thereby
improving some of the labor intensive work of medical personnel during bone diagnosis. MATLAB 9.7.0 programing
tool was used for the implementation of the work. The results of X-ray bone segmentation revealed that active contour
model using snake model showed the best performance in detecting boundaries and contours of regions of interest when
used in segmenting Femur bone image than the other medical image segmentation approaches implemented in the work.
Index Terms: Image Processing, Image Segmentation, Thresholding, Edge-based, Region-based technique,
Deformable Model
1. Introduction
Image segmentation is one of the most difficult tasks in image processing. Conventionally, image segmentation is
generally done by a radiologist and it is taking as the gold standard. This approach is distinguished as it can exploit
expert knowledge. However, the challenges of the approach are, it wastes a lot of time, it is not very precise and it is
open to inter-observer and intra-observer variability [1]. In traditional image segmentation approach, image is
distinguished by the nature of the intensity, contour and texture. However, there are many cases in which some of the
approaches used in determining the constituent of the images will not be functional thereby making it difficult to infer
from one problem to another. Also, traditional image segmentation wastes a lot of time during processing which reduces
the performance accuracy of image measurements. In recent times, researchers have moved from traditional methods of
Image segmentation to more advanced methods of image segmentation such as edge based detection, thresholding,
region growing and deformable model [2]. These methods help to partition images into numerous parts based on some
specific features such as intensity value, color, texture and etc. Also, they are largely based on features which are
similarity and discontinuity. Methods based on similarity are called Region based approaches while methods based on
discontinuities are called boundary based approaches.
Medical image segmentation is the process of forming visual representations of the internal parts of a body [3,4]
needed in scientific analysis for making intelligent diagnostic decisions. Medical image processing seeks to reveal the
internal structures of the objects and analysis obtained from these discoveries can be used by physician in proffering
solution to patient’s ailment. medical image processing is an important area of interest in health care industries based on
the fact that majority of medical images contains vital information about patients cases which tends to improve
visualization of human anatomy [5,6]. Automated image processing is a way of improving the accuracy of image
segmentation procedures using the algorithm that take images as input and return image as output by extracting
necessary information from the image. In this process, image needs to be reduced to certain defining characteristics and
the analysis of these image attributes gives relevant information for making diagnostic decisions [7]. Segmentation aims

28 Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
to split image into regions, based on the characteristics of the image that are relatively constant in each region.
Generally, segmentation purpose is to extract important information from medical imaging. Medical image processing
in bone structures has improved majorly in the areas of medical diagnoses that can be used by medical personnel in
studying the anatomical structures of the image needed in determining the best treatment plan to be given to patient [8].
Femur X-ray medical image is useful for many medical studies such as diagnosis, surgery and treatment. However,
segmentation of Femur bone X-ray images is a crucial task in automatic study of medical images and orthopedic
inspections. It is more challenging than segmenting computed Tomography (CT) and Magnetic Resonance (MR)
images because some of the lower density similar tissues are difficult to differentiate from the femur in X-ray images.
Despite many years of studying medical bone images, its segmentation still remains an open issue in different areas.
Although, there exist recent segmentation techniques to address medical image segmentation problems, there is
still no fit all among them for segmenting medical images. Each imaging technique has its peculiar constraints.
Therefore, this work tried to implement different image segmentation techniques to identify the best for segmenting
femur X-ray image. This will assist medical experts to be able to use suitable technique to quickly extract valuable
details from medical images and hence, improves the precision of clinical diagnosis as well as timely treatment of
patients.
The rest of the paper is organized as follows; section 2 gives the details of related work and techniques used in
image processing for medical images. Section 3 presents the methodology as well as result and discussions, while the
rest of the section gives the conclusion, recommendation and suggestions for future work.
2. Related Work
This section presents literatures of existing methods used in image segmentation in recent years. Most of this work
showed promising results based on techniques used. For example, in Osama et al [9], an algorithm to segment organs of
human body from CT scans using automated image processing technique was proposes. The authors employed Binary
Mask generations and Otsu thresholding to convert grayscale images with colors intensities from 0 to 255 to white and
black pixels. Mathematical morphology, distance transform, marker-controlled technique, and watershed transform was
also used in their approach, result proved the efficiency of image segmentation in enhancing the procedures and the
workflow of the radiological examination. Zhao et al [10] developed a versatile framework for medical image
processing using deep learning approach. The study employed RSNA dataset generative adversarial network (GAN)
and class active map was used for image processing. Result of mean average error (MAE) of 5.991 and 6.263 months
on male and female cohorts was achieved, comparable to the state-of-the-art performance on a large-scale dataset which
can be effectively applied to medical image processing task. Also, A study was carried out in [11,12] on image
segmentation using hybrid segmentation approach. The work used a locally adaptive thresholding technique that
removes background noise by using local mean and standard deviation. Result showed that Niblack algorithm is better
than the Sauvola algorithm in removing background noise from an image
Farmaha et al [13] carried out their study on image segmentation using clustering algorithms based on artificial
neural network (ANN) for identifying bio-medical images that can automate wound area selection. Result showed that
clustering algorithm could accurately be used to determine the precise assessment and measurement of wound in
determining the effective treatment for medical diagnosis. However, due to high dependence of artificial neural network
on training data, the heterogeneity of training images affects the allocation of features and segmentation. For better
results, more homogeneous images are required. Bansal et al [14] developed a hybridization approach for segmentation
of brain tumor to locate the digital pixel in the brain from MRI image. Using swarm optimization (PSOA) algorithm
and swarm ant lion method to improve the PSNR value for early detection of brain tumor that could enhance patient
lives. Parameters like image quality (PSNR), error rate (MSE) and accuracy rate (ACC) were used in model evaluation.
Accuracy of 98.58 %.was achieved using hybridization algorithm for detection of brain tumor. In the work of Tian et al
[15], they proposed image segmentation method based on deep reinforcement learning algorithm. Segmentation process
is formulated as a Markov decision process and solved by a deep reinforcement learning (DRL) algorithm, which trains
an agent for segmenting ROI in images. The agent performs a serial action to delineate the ROI. Experimental result
revealed that the proposed method has 7.24% improved to the state-of-the-art method on three prostate MR data sets
and has 3.52% improved on one retinal fundus image data set. Kant and Bala [16] focused on generalize encoder-
decoder model called dense dilated inception network (DDI-Net) for medical image segmentation. The study used
dense path to replace the skip connection in the middle of the encoder and decoder to make the model deeper and
replace the U-net basic convolution blocks with a Multi-scale dilated inception module in making the model wider.
Experiment result showed that the proposed approach had a better result that can be used for image segmentation with a
Dice score of 0.82 and 0.95 for brain tumor and heart segmentation respectively.
Ouyang et al [17] developed a self-supervised few-shot segmentation framework for medical imaging to address
the problem of existing techniques due to lack of annotations. The study employed adaptive local prototype pooling
module plugged into prototypical networks, to solve foreground-background imbalance problem in medical image
segmentation. The general applicability of the proposed study was carried out using abdominal organ segmentation as
well as cardiac segmentation for CT and MRI. Haider, et al [18] presented a hybrid method for edge continuity based on

Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques 29
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
pixel Neighbors pattern analysis. The authors used Sobel edge detector in detecting edges with noise-suppression
property after which, Otsu thresholding technique was used for localization of background and foreground pixels.
Experiments result revealed that the proposed technique outperform NN-based approach to segmentation in terms of
accuracy and processing time. Samet et al [19] proposed a new Fuzzy Rule-based image segmentation technique to
segment thin images in rock. The study used RGB image of rock thin segment as input and mineral image as output and
applied Fuzzy C Means on rock thin images for image processing Result showed that the proposed approach was better
than the existing algorithms used in this area. Khokher et al [20] presented a new approach of image segmentation using
Fuzzy Rule based system and Graph Cut to segment the gray scale, color, and texture images. Fuzzy rules were used in
assigning weight to the features of the images. Evaluation was carried out using Mean, Standard Deviation, and PPV
value. Result showed that proposed approach achieved a better result of 0.85 to 0.95 for S.D and PPV respectively. Li et
al [21] suggested an image segmentation technique that improvises the performance parameters of (CCQPSO)
algorithm to overcome the performance of CCQPSO algorithm having lesser convergence rate and slow searching
speed. The study used partitioned and cooperative quantum based PSO technique to overcome these issues by coupling
the two techniques for efficient segmentation. This approach improvises convergence and avoids trapping into local
optima. Result revealed that the proposed method improves segmentation with multiple thresholds.
Rouhi et al [22] classified breast tumors using region growing and CNN segmentation techniques for generating
adaptive thresholds and templates for conserving tumor boundaries. Five classification algorithms such as random forest,
support vector machine, KNN, MLP and naïve Bayes were used for prediction. The proposed approach was tested on
publicly available (DDSM and MIAS) dataset consisting of 219 images of malignant and benign patient respectively.
Experimental result yield accuracy of 96.47% and showed that CNN segmentation algorithm help in efficient
classification of breast tumors. Tyan et al [23] identified several image segmentation methods for effective detection of
ischemic stroke or embolus in brain. The author focused on image in frequency-time domain and applied various
measures for detection of emboli in brain. Comparisons were made among the model to determine the best performing
model. Evaluation result revealed that SM modeling has 84.2% acceptance rate for estimating stroke area. An approach
of segmenting medical images was proposed in [24] using unsupervised learning and calculating local center of mass of
an image signal. In the study, pixels were assembled into regions based on their center of mass. The authors compared
their approach with other existing methods such as watershed method, SLIC and GMM-HMRF. Result showed that
their approach produces promising result and better boundaries between regions by attaining the highest optimal dice
score which outperformed other existing unsupervised methods used in segmenting medical images.
Different image segmentation approaches have been used by different authors to segment diverse image parts and
have each reported to be successful according to the reviews. In this paper, we investigated different image
segmentation techniques on femur bone image in order to identify the most promising out of them all that can handle
femur image segmentation problems accurately.
2.1 Medical Image Segmentation Techniques.
Dar and Padha [25] highlighted various approaches that can be used in image processing for medical image
segmentation as follows:
i. Segmentation by Clustering: Clustering algorithm helps to improve the performances accuracy of the models
used in medical image segmentation. The used of clustering technique is still a challenging issue in image
processing as this approach cannot be used to solve all segmentation problems. However, many studies have
used this approach to solve classification problems as highlighted in Table 1.
Table 1. Clustering techniques used in image segmentation
S/N
Author/year
Method
Result
Limitation
Future improvement
1
Li et al, [26]
fuzzy clustering with
cellular automata (CA)
and features weighting
Fast convergence speed,
strong anti-noise property,
and robustness. Ability to
effectively segment
common images and long-
term sequence satellite
remote sensing images and
has good applicability.
The determination of
fuzzy membership is
tasking.
More image features
to be utilized in the
system, identification.
of optimal feature
combination, more
efforts to improve the
segmentation speed
and efficiency.
2
(Lei et al,
[27]
Automatic fuzzy
clustering framework
for image
segmentation by
integrating super pixel
algorithms, density
peak clustering, and
prior entropy.
Accurate numbers of
clusters were obtained
through the system. It
produced the best
segmentation results than
the state of the art.
It is computationally
intensive.
Convolutional neural
networks can be used
to extract image
features and feature
learning algorithm can
be explored to achieve
better automatic
image segmentation.

30 Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
3
(Qureshi
and Ahamad,
[28]
Clustering technique
with Neutrosophy was
used to solve the
problem of
indeterminacy factor
of image pixels
Better results on synthesis
image and real images
with/without noise.
The approach may
not work well on a
non- global clusters
Neutrosophy can be
applied to other image
processing problems
such as feature
extraction and
classification
4
Bora and
Gupta [29]
K- means clustering
with cosine distance
measure.
An efficient result of color
image segmentation
The approach may
not adequately detect
blurred image
An improved
clustering technique
with a good
determines number of
clusters may be
considered for better
segmentation result.
5
Hassan et al,
{30]
K-means clustering
combined with RGB
and HSV color spaces
Accurate segmentation
result of RGB and HSV
color spaces compared to
segmentation of single
color space
The algorithm may
not be able to detect
all colors.
Modification of the
proposed algorithm by
working on various
color space such as
CMYK,
L*a*b,YCBCR and
HSL to determine the
best segmentation
result.
6
Saravanan et
al, [31]
K-means clustering
and local thresholding
techniques
Good segmentation
performance is achieved
using K-means clustering
than thresholding method
on mammographic images.
Difficult in predicting
the value of K with
fixed number of
clusters
Improved k-means
algorithms and other
clustering methods
can be applied for
more accurate
classifications in
mammographic
images
7
Panda [32]
Comparative analysis
of K-means clustering
and thesholding
technique.
Low computational time,
increase in clusters size
and more accurate
segmentation quality using
K-means
Low convergence
Time
K-means clustering
can be applied to
applications such as
video retrieval and
face recognition
system
8
Dubey et al
[33]
K- means clustering
Low computational cost
and robust technique for
defected area of fruits
segmentation
High memory
management
Machine learning
algorithms can be
used to determine the
number of clusters to
segment the defected
fruits more
effectively.
9
Inbarani et
al, [34]
Hybrid
histogram-based soft
covering rough k-
means clustering
(HSCRKM) technique
The algorithms accurately
segment the nucleus with
improved accuracy using
logistic regression and
neural network algorithms
for detection.
Require more
processing time for
multiple color
images.
Future work can use
Bio-inspired
algorithms to improve
the number
of clusters.
10
Wang and
Zhang [35]
K-Means algorithms
based on internet of
things
Robust techniques for
segmentation of plant
disease in leaf images,
The algorithm may
not work well on a
non-global cluster
leaf images
The number of
clusters can be
improved using
clustering algorithms
that improve
segmentation result.
Clustering algorithms is an efficient technique that is robust and computationally faster when the clusters are
global. The technique is used in image processing for its simplicity of implementation and convergence speed. The
limitations of the algorithms are based on its time complexity which is high and does not work well on non-global
clusters. Sometimes, it is difficult in predicting the value of K with fixed numbers of clusters which may affect
segmentation outputs.
ii. Region Growing: This method of segmentation is used to extract a specific region from pre-existing image
based on some characteristics such as intensity level inhomogeneity or edges in an image among others
[21]. This method requires prior information for selecting a seed pixel. The seed point or pixel is selected by
an operator and then pixels that share a unique characteristic such that we can grow a seed pixel in an image
till an edge is detected. The region growing technique is never used alone but it is usually combined with set
of image processing operations for visualization of small, simple and delicate regions in tumors and lesions.
Sometimes region growing method can be sensitive to noise which may affect the accuracy of image
segmentation. Various studies conducted using this approach is shown in Table 2.

Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques 31
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
Table 2.Region growing techniques used in image segmentation
S/N
Author/year
Method
Result
Limitation
Future Work
1
Shimodaira
[36]
Seeded region growing
and merging image
segmentation centered
on square elemental
regions.
Reduction in time
complexity and good
segmentation output with
a small number of
regions.
The resolution of the
segmentation outputs does
not capture objects of
regions smaller than
single square elemental
regions.
Future work can be done
using a segmentation
technique that will
capture all regions of the
image rather than
dividing it into sub
regions
2
Yuan [37]
3D segmentation of
human femur using
region growing
algorithms.
Less computational time
with stable results and
improvement in image
segmentation processes
The segmentation process
may not actually capture
all ROIs due to its
resolution.
Developing a high-
resolution image
segmentation method that
can capture all hidden
joints in bone regions
3
Kaur and
Jindal [38]
Region growing
algorithm inside GPU
using Parallel Best
Fitting and Parallel
Local Mutual Best
Fitting techniques.
A better segmented
image with the best
performance obtained
using Parallel Local
Mutual Best Fitting
techniques
The algorithms cannot
handle a wide range of
weighting function on
each image region.
Combining region
growing with other
algorithms such as N-cut
method of pixel based
segmentation to reduce
the time complexity of
image segmentation
processes.
4
Jaber et al
[39]
Region growing
segmentation using
double filtering
techniques for data
preprocessing
A robust segmentation
techniques for breast
cancer detection
It is time consuming
An efficient feature
extraction technique with
a more accurate
segmentation algorithm
can be considered.
5
Kansal and
Jain [40]
Automatic seed
selection algorithm
using region growing
segmentation
It produces original
image having clear edge
with good segmentation
results.
The computational cost is
high.
K-mean algorithm can be
used as a robust
technique for improving
segmentation
performance.
6
Shewale and
Patil [41]
Region growing
segmentation
Fast and accurate
segmentation for brain
tumor identification.
It may lead to over
segmentation if the image
is noisy
A robust algorithm of
region growing
segmentation combined
with histogram
thresholding technique
can be used for better
segmentation
7
Biratu et al
[42]
Modification of region
growing algorithm using
deep learning approach.
The algorithm can
identify brain tumor
locations and extract the
best region of interests.
The proposed technique
cannot adequately select
thresholding points for
region-growing algorithm
Future work can be done
using a deep learning
algorithm that can
overcome the limitations
of existing approaches.
8
Udayakumar
et al [43]
Comparison of region
growing algorithm with
K means clustering
technique.
Proposed Region
growing algorithm gives
better accuracy than K-
Means clustering
segmentation for
newborn baby MRI.
High computational
complexity.
Future work can be done
using canny edge based
algorithm and other
convolutional algorithms
to have more accurate
detection.
9
Jain and
Susan [44]
Adaptive single-seed
based region growing
segmentation algorithm.
A robust algorithm with
good segmentation result
for a wide-ranging
realistic images
High computational cost
and time consuming
A fast and more accurate
algorithm using deep
learning approaches
could be considered.
10
Thukral et al
[45]
Region growing
algorithm and Median
filter method combined
with Recurrent neural
network algorithm
A robust technique for
identifying the region of
interests with 97.12%
accuracy for lung cancer
detection.
Training time for Neural
network is long and time
consuming
An ensemble approach of
machine learning
techniques can be
considered to improve
detection accuracy.

32 Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
Region growing segmentation is an efficient technique that works based on partitioning of image into regions, the
algorithms are easy to implement, fast, robust, connected regions are guaranteed and clear object boundaries are
generated. The drawback lies in the areas of its high computational complexity in memory and time management and
may sometimes lead to over segmentation/ under segmentation of images if the object is noisy and not well
preprocessed.
iii. Edge based/ boundary based: This segmentation approach deals with identifying and locating boundaries in
an image such as edges. The edges are sharp discontinuities which are. Intensity values in an image. This
technique is helpful in recognition, disclosure and segmentation of image artifacts. The edge detectors are
called ‘masks’ or ‘filters’ which are super-imposed over an image to detect discontinuities or boundaries.
Review of literatures carried out on this algorithm is highlighted in Table 3.
Table 3. Edge based image segmentation techniques
S/N
Author/year
Method
Result
Limitation
Future improvement
1
Padmapriya et al
[46]
Edge based image
Segmentation using 2D
ultrasound image.
Reduction in
computational time and
improvement in overall
efficiency of the system for
detecting bladder
boundaries
Sensitivity to noise
A 3D ultrasound image may be
used on classical models of the
proposed algorithm to improve
accuracy of detection
2
Kharoffa [47]
Performance evaluation of
seven edge detection
techniques by measuring
the structural content of
the original image.
Canny edge detection
algorithm achieved better
result as compared to other
methods
It is computationally
intensive
Future work can consider an
algorithm that is less sensitive to
noise to improve segmentation
result.
3
Cao et al, [48]
Edge Detection
segmentation based
on the Otsu-canny
operator on the hadoop
platform
Improved time reduction
and better edge detection
technique than existing
traditional edge detection
algorithms
It is time consuming
An efficient algorithm that can
handle real-time large-scale image
edge detection segmentation could
be considered.
4
Mittal et al, [49]
Edge detection algorithm
using multiple
thresholding (B-edge)
techniques
Ability to detect robust and
thin edges with reduced
noise proportion
The algorithms
cannot efficiently
work on blur images
Deep learning approach can be
proposed to reduce the
computational time of the
algorithms
5
Yao [50]
MM-Sobel edge detector
technique
Efficient algorithms
capable of detecting edges
of the image with low
computational time
The algorithm is
sensitive to noise
Canny edge detection could be
used to improve the accuracy of
detection
6
Aslam et al, [51]
Improved Sobel edge
detection algorithm
combined with image
dependent thresholding
approach.
A better performance of
image segmentation over
conventional segmentation
algorithms
Difficult in detecting
closed contour
regions of the
images.
Future work can be explored using
an approach to improve closed
contour algorithm in order to
increase the region area and
reduce the thickness of boundary
lines of regions
7
Varadarajan et al,
[52]
A distributed Canny edge
detection segmentation
technique
Scalable fast detection
algorithm with reduction in
latency capable of
supporting images and
videos
High computational
cost and time
consuming
A combined algorithms of
segmentation techniques can be
used in real time processing to
improve accuracy of detection
8
Chithambaram1
and Perumal [53]
Modified canny edge
detection and artificial
neural network algorithm.
Fast, Accurate and robust
segmentation algorithms
for detection of brain
tumor
High computational
cost in detecting
accuracy
Future research can be explored
using different machine learning
approaches for improving the
accuracy of segmentation
algorithms.
9
Asmaidi [54]
Sobel edge detection
technique
Accurate segmentation
with different mean
squared error results for
flowers image.
Highly sensitive to
noise
Algorithms for improving image
acquisition qualities for improving
segmentation accuracy can be
proposed.
10
Ratnam et al [55]
Canny edge detector
algorithm
Robust and reliable
segmentation output for
detecting brain tumor in
MR image
It is time consuming
The algorithm can be extended to
identification of 3D image
processing.

Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques 33
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
Edge based image segmentation is a method that is centered on discontinuities detection .The algorithms are
simple to implement and can give an accurate segmentation with good results even in a noisy environment and work
well on low quality images than other segmentation algorithms. The limitation of the technique is its high
computational complexity in terms of time and memory management.
iv. Threshold Based Segmentation: This is the commonest and easiest approach used in medical image
segmentation. This method is used to convert a grey-scale image to binary image. In this approach threshold
value is specified and the image is fragmented into group of pixels having value less than or equal to the
Specified threshold value. Threshold based segmentation can be used for images having light objects over
darker background. Global and Local thresholding are the two most commonly used thresholding approaches
[22]. Threshold technique has a fast processing speed. This method will work well when object and
background have high contrast. The drawback of this technique is that it is unable to give accurate result when
mage has no major grey scale difference or image with overlapping grey scale. Recent studies conducted
using this segmentation approach is highlighted in Table 4.
Table 4. Threshold based segmentation techniques.
S/N
Author/year
Method
Result
Limitation
Future Work
1
Dash and
Bhoi [56]
Otsu tresholding was
used for blood vessel
segmentation using
images obtained from
DRIVE and STARE
databases
Robust segmentation with
less computation time and
easy implementation..
Highest accuracy of 0.956
was achieved using DRIVE
database.
Wrong selection of
threshold may lead to
over-segmentation
Future work can be
explore using fuzzy
based algorithm for
blood vessel
segmentation
2
Wang et al
[57]
Two-dimensional Otsu
based on estimation of
distribution
algorithm using guided
filtering
Improved segmentation and
less computational time with
exponential growth.
Limited in handling
images with the same
gray scale range but not
efficient for object with
large gray scale
distribution
Efficient segmentation
techniques could be
adopted so as to
handle more complex
images
3
Mapayi et al,
[58]
Thresholding technique
based on gray level co-
occurrence matrix-
energy information for
retinal vessel
segmentation.
A robust and time efficient
segmentation with efficient
accuracy rates
It has assumption of
uniform illumination
The use of soft
Computing and
heuristics approaches
may be adopted in
detecting more thin
vessels.
4
Jang et al,
[59]
Global thresholding
algorithm using
boundary blocks for
extracting a bimodal
histogram.
Robust image segmentation
method for images with
noise and small objects
The method may not
accurately to handle
large scale images
5
Vijay and
patil [60]
Otsu thresholding using
Iteration and Custom
approach
Custom approach showed
better result in segmenting
foreground from background
images.
High sensitive to noise
Histogram thresholding
may be considered for
future research
6
Telgad et al,
[61]
Global thresholding
algorithms
Simple, fast and accurate
segmentation for fingerprint
minutiae extractions
Texture of the image if
not well preprocessed
may affect
segmentation result
Otsu method and PCA
algorithms may be used
to accurately de-noised
the image.
7
Gurung and
tamang [62]
Heuristic approach for
image segmentation to
determine multilevel
thresholds by sampling
the histogram of a digital
image.
Decrease in CPU
computational time for image
segmentation than Otsu
method with robust images
The approach may not
efficiently be used to
detect all boundaries in
an image
Clustering algorithms
could be applied to detect
image efficiently wilt
less time complexity
8
Srinivas et
al, [63]
Otsu thresholding
algorithm
It gives better accuracy in
segmenting the image than
existing algorithms
Highly sensitive to
noise
K-means algorithm could
be proposed for future
studies
9
Pai et al,
[64]
Threshold-gradient
based Segmentation
method
Proposed approach
performed efficiently than
existing algorithms
It neglects spatial
information of an image
Convolutional neural
network algorithms
could be used to improve
the results
10
Patil and
Shaikh [65]
Otsu thresholding using
L*a*b color space
Good segmentation result for
flower images as compared
to the state of the art.
It is sensitive to noise
and selecting wrong
threshold can hindered
segmentation results.
Feature extraction such
as color shape and
texture can be performed
on segmented image to
improve accuracy of
detection.

34 Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
Thresholding is a popular method of image segmentation used in discriminating background from foreground
images; it is a simple, fast, easy technique with low computational time and does not normally require prior information
of the image before segmentation. The algorithm is efficient and robust in its detention. The weakness of this algorithm
is that it does not work fast with wide range of images. Threshold selection is a crucial step of this method, selecting the
wrong threshold may leads to over segmentation of the algorithm and cause inaccurate segmentation results.
3. Methodology
The process of femur bone image processing as depicted in Figure 1 consists of the image segmentation techniques
employed in this study which includes edge based detection, thresholding, region growing and deformable model. The
performance of these algorithms were investigated on the input image and comparison were made between the four
segmentation techniques to determine the best performing among them that could be used in bone image segmentation
for making intelligent decisions.
Fig. 1. The process of femur bone image segmentation
3.1 Experimental Description
The experiment was carried out in MATLAB. 9.7.0 Programming environment. Femur X-Ray image was used as
the input image for image segmentation. The femur bone image as depicted in Fig. 2 was taken at 53 kV and digitized at
7 bit/pixel using a charge-coupled device (CCD) camera with a size of 410 by 500 resolutions. Thereafter it was
observed that the intensity level of the image was not widely dispersed. To correct this anomalies caused by uneven
illumination in the image, histogram generation was used to enhanced the appearance of the images by mapping the
input image to the output image such that its histogram is uniform after mapping. This pre-processing step partially
eliminates the intensity variations between images.
Fig. 2. X-ray Femur Image (input Image)
As part of the preprocessing techniques needed to remove noise that can affect the performance of image
segmentation algorithms, the image was converted into binary form where each gray level image was quantized into
bits consisting of ’0’ as black pixel and ‘1’ as white pixel by separating femur shaft image from soft tissue shade pixel.
A non-linear digital image filtering approach known as median filter was used to suppress isolated noise while
preserving the femur bone border edge. After image pre-processing stage, the input image was processed with four
different segmentation algorithms whose performance were measured to determine the best performing approach and
classical model that can robustly handle femur bone image segmentation problem.

Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques 35
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
4. Results of the Different Segmentation Techniques on Femur Bone Image
The following section shows the results of the four segmentation techniques used in the study which are Edge
detection, Thresholding, Region growing and deformable model.
i. Thresholding: Global image thresholding in equation 1 and 2 respectively were applied on the input image to
determine the optimal value to distinguish the region of interest (the bone image) from the background (X-
ray image using Otsu’s method of thresholding. This technique was based on the interclass variance
maximization so that threshold classes will have a well discriminated intensity values.
         (1)
Where T denotes the threshold value, a, b represented the coordinate of the threshold value point. q(a, b), r(a, b)
symbolized the points of the gray level images. Threshold image h(a,b) can further be expressed as denoted in equation
4

 (2)
Fig. 3 depicts result obtained after applying Otsu method of thresholding on the input image.
Fig. 3. Otsu method of thresholding
ii. Edge based segmentation: This algorithm uses edge detectors to find edges in the image. The study adopted
five based edge detectors techniques such as, canny operator, prewitt operator, sobel operator, Roberts
operators and log (laplacian) operator which were applied on the input image and comparison were made
among the edge based segmentation algorithms used in the study to ascertain the best edge detector
techniques. Fig. 4 depicts the output of various edge based segmentation techniques used in the study.
Analysis of the visual inspection of the experimental results obtained from the images denotes that Prewitt
operator gives a similar image as obtained in Sobel and Roberts. Canny and Log also showed similar image
which perfectly detects all the edges in the image (flesh and bones) and can be seen as the most effective
method for detecting edges of the input image using edge based segmentation.
Fig. 4. Edge based segmentation algorithms output

36 Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
iii. Region-growing segmentation: The use of region growing segmentation adds neighboring pixels to the
regions of femur bone with similar image features, thereby growing the regions. The anatomical structure of
femur bone using this approach is depicted in Fig. 5.
Fig. 5. Region growing segmentation on input image
iv. Active contour model: Snake model of active contour model in medical image segmentation works by
identifying the region of interest (ROI) under consideration (Femur bone) [66,26]. This model uses the
application of spline to decrease the energy in ROI by different internal and external forces that is affecting
the image based on suitable contour features. It propagates through the region of intrest of input image to
lessen the energy function and dynamically move to the local minimum as expressed in equation 3.
      (3)
Where the coordinate of the two dimensional curve is denoted by q and r, given that g denotes the spline
parameters that ranges from [0-1, m symbolize the linear parameter [0,1] and o represent the time parameter  [0, ∞].
The total energy (
) used by the snake model in identification of the image features is expressed as represented in
equation 4;

= + +  (4)
Where  denotes the internal energy of the snake model which relies on the degree of the spline connecting to the
shape of the target image which explains piecewise smoothness factors in the contour,  represent the external energy
specified by the user and  is the energy of the femur bone image under consideration that moves valuable data on the
illumination of the spline signifying the target object.
The internal energy of snake active contour model for detecting region of interest (Femur bone image) in the study
can further be expressed as depicted in equation 5.
 
  
 (5)
Given that α denotes how far the snake will be protracted and the capacity of elasticity possible for the snake while
β is use to determine the rigidity level for the snake needed in bone processing. Fig. 6 represent the output of active
contour model using snake model as it perfectly detected the smooth shape in the femur image.
Fig. 6. Active contour model (snake model)

Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques 37
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
4.1 Discussions of the Results
From the results obtained above, Thresholding technique of bone image processing is one of the simplest
segmentation techniques, but it is difficult to threshold noisy images as the background intensity and the foreground
intensity could not be distinctly separated making it difficult to estimate the threshold for each sub-image. Edge based
used different edge detector techniques, but it was observed from the study that the edges extracted by edge-based
algorithms were disjointed and could not completely represent the boundary of an object. However, region growing
algorithms as used in the study proved to be efficient and hence, a reliable approach that could be used in femur bone
image processing for detecting ROI. It was also observed that these techniques have over-segmentation tendency in
processing the input image and are sensitive to noise. Deformable techniques were less sensitive to noise if well pre-
processed than the other model considered in the study. This makes them more suitable for complex medical bone
image segmentation problems. Based on the results of the implementations, it has shown that Thresholding, Edge-based,
Region based techniques have the capabilities to solve simple medical bone image segmentation problems. However, in
case of complex bone images segmentation problems, which cannot be tackled effectively by classical image
segmentation approaches, deformable model is the most suitable approach.
5. Conclusion
In this study, we have identified the strengths and weaknesses of different segmentation techniques and we have
also implemented the segmentation techniques on femur bone image to detect the most promising among them and their
results have been presented. For the examination of the segmentation techniques, femur bone image was used as input.
All implementations were carried out in MATLAB 9.7.0 programming environment. From the visual results, Active
contour model using snake model came out as the most efficient technique that can be used in segmenting input data
(Femur image) than the other medical image segmentation approaches implemented in the study. This is because it was
able to perfectly detect the smooth shape in the femur image presented to it. Therefore, automatic segmentation of
medical bone images for diagnosis and assessment can be accomplished using active contour model. Also, it was
observed that the effectiveness of diverse segmentation techniques depends on the nature of image modality,
characteristics of region of interest and application.
However, in image processing and computer vision, segmentation is still a challenging issue for many real life
applications and hence more innovative work is required. Future work can be explored in the areas of combining
geometric techniques for active contours model and deep learning approach to overcome the limitations of some of the
existing approaches.
References
[1] Saini, S., & Arora, K. (2014). A study analysis on the different image segmentation techniques. International Journal of
Information & Computation Technology, 4(14), 1445-1452.
[2] Shiv Gehlot, John Deva Kumar,"The Image Segmentation Techniques", International Journal of Image, Graphics and Signal
Processing(IJIGSP), Vol.9, No.2, pp.9-18, 2017.DOI: 10.5815/ijigsp.2017.02.02.
[3] Kumar, S., Lenin , F., Muthukumar, s., Ajay , K., and Sebastian , V. (2018). A voyage on medical image segmentation
algorithms. Computational Life Sciences and Smarter Technological Advancement, Special Issue: 75-87.
[4] L.Sankari,C.Chandrasekar,"A New Enhanced Semi Supervised Image Segmentation Using Marker as Prior Information",
IJIGSP, vol.4, no.1, pp.51-56, 2012..
[5] Dorgham, O., Osama , M., Stephen , D., Laycock, O., and Mark, H. (2012). GPU accelerated generation of digitally
reconstructed radiographs for 2-D/3-D image registration. IEEE Transactions on biomedical engineering, 59(9), 2594-
2603.
[6] Mandeep , K., and Jindal, G. (2011). Medical Image Segmentation using Marker Controlled Watershed Transformation.
International journal of Computer Science and Technology, 2(1), 1-6.
[7] Huiyu, Z., Jiahua, W., and Jianguo, Z. (2010). Digital Image Processing Part II, 2nd edition. London, United Kingdom: Ventus
Publishing APS.
[8] Farzaneh Nikroorezaei, Somayeh Saraf Esmaili, " Application of Models based on Human Vision in Medical Image Processing:
A Review Article", International Journal of Image, Graphics and Signal Processing (IJIGSP), Vol.11, No.12, pp. 23-28,
2019.DOI: 10.5815/ijigsp.2019.12.03.
[9] Osama , D., Mohammad , J., Mohammad , H., and Ammar , A. (2018). Proposed Method for Automatic Segmentation of
Medical Images. International Arab Confrence on Infornation Technology, Lebanon: IEEE.pp. 1-9
[10] Zhao, C., Han, J., Jia, Y., Fan, L., and Gou, F. (2018). Versatile Framework for Medical Image Processing and Analysis with
Application to Automatic Bone Age Assessment. Journal of Electrical and Computer Engineering, 1(2), 1-14.
[11] Senthilkumaran , N., and Vaithegi , S. (2016). Image Segmentation by Using Thresholding Techniques for Medical Images.
International Journal of Computer Science and Engineering, 6(1), 1-13.
[12] Anithadevi, D., and Perumal, K. (2016). A Hybrid Approach Based Segmentation Technique for Brain Tumor in MRI Images.
International Jpournal of Signal and Image Processing, 17(1), 1-10.
[13] Farmaha, I., Banas, M., Lukashchuk, V., and Farmaha, T. (2019). Wound image segmentation using clustering based
algorithms. New Trends in Production Engineering, 2(1), 570-578.

38 Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
[14] Bansal, S., Kaur, S., and Kaur, N. (2019). Enhancement in Brain image segmentation using Swarm Ant Lion Algorithm.
International Journal of Innovative Technology and Exploring Engineerin, 8(10), 1-6.
[15] Tian, Z., Si, X., Zheng, Y., Chen, Z., and Li, X. (2020). Multi-step medical image segmentation based on reinforcement
learning. Journal of Ambient Intelligence and Humanized Computing, 1(2), 1-12.
[16] Kant, S., and Bala, S. (2020). Dense Dilate Inception Network for Medical Image Segmentation. international Journal of
Advanced Computer Science and Applications, 11(11), 785-793.
[17] Ouyang, C., Biffi, C., Chen , C., Kart, T., Qiu, H., and Rueckert, D. (2020). Self-Supervision with Superpixels: Training Few-
shot Medical Image Segmentation without Annotation. Computer Vision and Pattern Recognition, 2(1) 1-19.
[18] Haider, W., Malik, M., Raza, M., Wahab, A., Khan, I., Zia, U., et al. (2012). A hybrid method for edge continuity based on
Pixel Neighbors Pattern Analysis (PNPA) for remote sensing satellite. International Journal of Communications, Network and
System Sciences, 5(1), 624-630.
[19] Samet, R., Amrahov, E., and Ziroglu, A. (2012). Fuzzy rule-based image segmentation technique for rock thin section images.
3rd International Conference on Image Processing Theory, Tools and Applications (pp. 402-406). IEEE: Istanbul,Turkey.
[20] Khokher, M., Ghafoor, A., and Siddiqui, A. (2012). Image segmentation using fuzzy rule based system and graph cuts. 12th
International Conference on Control Automation Robotics & Vision, Guangzhou, China : IEEE. 1148-1153.
[21] Li, Y., Bai, X., Jao, L., and Xue, Y. (2017). Partitioned-cooperative quantum-behaved particle swarm optimization based on
multilevel thresholding applied to medical image segmentation. Journal of Applied Soft Computing, 56(1), 345-356.
[22] Rouhi, R., Jafari , M., Kasaei, S., and Keshavarzian, P. (2014). Benign and malignant breast tumors classification based on
region growing and CNN segmentation. Expert Systems With Application, 42(3), 990-1002.
[23] Tyan, Y., Wu , M., Chin, C., Kuo, Y., Lee, M., and Chang, H. (2014). Ischemic stroke detection system with a computer-aided
diagnostic ability using an unsupervised feature perception enhancement method. International Journal of Biomedical Imaging,
12(2), 1-6.
[24] Aganj, I., Harisinghani, M., Weisslede, R., and Fischi, B. (2018). Unsupervised Medical Image Segmentation Based on the
Local Center of Mass. Journal of Scientific Report, 8(1), 1-8.
[25] Dar, S. A., and Padha, D. (2019). Medical Image Segmentation: A Review of Recent Techniques, Advancements and a
Comprehensive Comparison. International Journal of sComputer Sciences and Engineering, 7(7), 114-124.
[26] Li, C., Liu, L., Sun, X., Zhao, J., and Yin, J. (2019). Image segmentation based on fuzzy clustering with cellular automata and
features weighting. EURASIP Journal on Image and Video Processing, 2019(1), 1-11.
[27] Lei, T., Liu, P., Jia, X., Zhang, X., Meng, H., and Nandi, A. K. (2019). Automatic fuzzy clustering framework for image
segmentation. IEEE Transactions on Fuzzy Systems, 28(9), 2078-2092
[28] Qureshi, M. N., and Ahamad, M. V. (2018). An improved method for image segmentation using K-means clustering with
neutrosophic logic. Procedia computer science, 132, 534-540.
[29] Bora, D. J., and Gupta, A. K. (2015). A novel approach towards clustering based image segmentation. arXiv preprint
arXiv:1506.01710.
[30] Hassan, M. R., Ema, R. R., and Islam, T. (2017). Color image segmentation using automated K-means clustering with RGB
and HSV color spaces. Global Journal of Computer Science and Technology.
[31] Saravanan, M., Kalaivani, B., and Geethamani, R. (2017). Image Segmentation Using K-means clustering based thresholding
algorithm. International journal of advanced technology in Engineering and science, 5(2).
[32] Panda, S. (2015). Color Image Segmentation Using K-means Clustering and Thresholding Technique. Interntional journal of
ESC, 1132-1136.
[33] Dubey, S. R., Dixit, P., Singh, N., and Gupta, J. P. (2013). Infected fruit part detection using K-means clustering segmentation
technique.
[34] Inbarani H, Hannah, and Ahmad Taher Azar. "Leukemia image segmentation using a hybrid histogram-based soft covering
rough k-means clustering algorithm." Electronics 9, no. 1 (2020): 188.
[35] Wang, Yun Shi Xuqi, and Shanwen Zhang. "Plant Disease Leaf Image Segmentation Using K-Means Clustering Based on
Internet of Things." (2016).
[36] Shimodaira, H. (2017). Automatic color image segmentation using a square elemental region-based seeded region growing and
merging method. arXiv preprint arXiv:1711.09352.
[37] Yuan, Y., Chen, D., and Yan, L. (2015). Interactive three-dimensional segmentation using region growing algorithms. Journal
of Algorithms & Computational Technology, 9(2), 199-213.
[38] Kaur, G., and Jindal, S. Region growing image segmentation on large datasets using gpu. International journal, 15(14).
[39] Jaber, I. M., Jeberson, W., and Bajaj, H. K. (2006). Improved region growing based breast cancer image segmentation.
International Journal of Computer Applications, 975, 888
[40] Kansal, S., and Jain, P. (2015). Automatic seed selection algorithm for image segmentation using region growing. International
Journal of Advances in Engineering & Technology, 8(3), 362.
[41] Shewale, T. P., and Patil, S. B. (2016). Detection of brain tumor based on segmentation using region growing method. Int. J.
Eng. Innov. Res, 5(2), 173-176.
[42] Biratu, E. S. S., Schwenker, F., Debelee, T. G. G., Kebede, S. R. R., Negera, W. G. G., and Molla, H. T. T. (2021). Enhanced
Region Growing for Brain Tumor MR Image Segmentation. Journal of Imaging, 7(2), 22.
[43] Udayakumar, E., Santhi, S., Gowrishankar, R., Ramesh, C., and Gowthaman, T. (2016). Region growing image segmentation
for newborn brain MRI. Biotechnology: An Indian Journal Trade Sci Inc, 12, 1-8.
[44] Jain, P. K., and Susan, S. (2013, December). An adaptive single seed based region growing algorithm for color image
segmentation. In 2013 Annual IEEE India Conference (INDICON) (pp. 1-6). IEEE.
[45] Thukral, T., Verma, S., Gaur, S., and Tyagi, N. Segmentation of Lung Cancer using Mark Region Growing and Median Filter.
International Journal of Computer Applications, 975, 8887.
[46] Padmapriya, B., Kesavamurthi, T., and Ferose, H. W. (2012). Edge based image segmentation technique for detection and
estimation of the bladder wall thickness. Procedia engineering, 30, 828-835.

Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques 39
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
[47] Kharrofa, S. H. A. (2018). Segmentation of Images by Using Edges Detection Techniques. Iraqi Journal of Information
Technology. V, 8(3).
[48] Cao, J., Chen, L., Wang, M., and Tian, Y. (2018). Implementing a parallel image edge detection algorithm based on the Otsu-
canny operator on the Hadoop platform. Computational intelligence and neuroscience, 2018.
[49] Mittal, M., Verma, A., Kaur, I., Kaur, B., Sharma, M., Goyal, L. M., ... and Kim, T. H. (2019). An efficient edge detection
approach to provide better edge connectivity for image analysis. IEEE Access, 7, 33240-33255
[50] Yao, Y. (2016). Image segmentation based on Sobel edge detection. In 2016 5th International Conference on Advanced
Materials and Computer Science (ICAMCS 2016). Atlantis Press.
[51] Aslam, A., Khan, E., and Beg, M. S. (2015). Improved edge detection algorithm for brain tumor segmentation. Procedia
Computer Science, 58, 430-437.
[52] Xu, Q., Varadarajan, S., Chakrabarti, C., and Karam, L. J. (2014). A distributed canny edge detector: algorithm and FPGA
implementation. IEEE Transactions on Image Processing, 23(7), 2944-2960.
[53] Chithambaram, T., and Perumal, K. (2016). Edge detection algorithms using brain tumor detection and segmentation using
artificial neural network techniques. International Research Journal of Advanced Engineering and Science, 1(3), 135-140.
[54] Asmaidi, A., Putra, D. S., and Risky, M. M. (2019). Implementation of sobel method based edge detection for flower image
segmentation. Sinkron: Jurnal dan Penelitian Teknik Informatika, 3(2), 161-166.
[55] Ratnam, S. S., Kumar, D. S., and Madhukar, K. (2015). Implementation of Edge Detection Technique for Identification and
Evaluation of Brain Tumor Metrics in MR Image through Lab VIEW.
[56] Dash, J., and Bhoi, N. (2018, January). Retinal blood vessel segmentation using Otsu thresholding with principal component
analysis. In 2018 2nd International Conference on Inventive Systems and Control (ICISC) (pp. 933-937). IEEE
[57] Wang, W., Duan, L., and Wang, Y. (2017). Fast image segmentation using two-dimensional Otsu based on estimation of
distribution algorithm. Journal of Electrical and Computer Engineering, 2017.
[58] Mapayi, T., Viriri, S., and Tapamo, J. R. (2015). Adaptive thresholding technique for retinal vessel segmentation based on
GLCM-energy information. Computational and mathematical methods in medicine, 2015.
[59] Jang, J. W., Lee, S., Hwang, H. J., and Baek, K. R. (2013, October). Global thresholding algorithm based on boundary selection.
In 2013 13th International Conference on Control, Automation and Systems (ICCAS 2013) (pp. 704-706). IEEE.
[60] Vijay, P. P., and Patil, N. C. (2016). Gray scale image segmentation using OTSU Thresholding optimal approach. Journal for
Research, 2(05).
[61] Telgad, R., Siddiqui, A. M., and Deshmukh, P. D. (2014). Fingerprint image segmentation using global thresholding.
International Journal of Current Engineering and Technology, 4(1), 216-9.
[62] Gurung, A., and Tamang, S. L. (2019). Image segmentation using multi-threshold technique by histogram sampling. arXiv
preprint arXiv:1909.05084.
[63] Srinivas, C. H. V. V. S., Prasad, M. V. R. V., and Sirisha, M. Remote Sensing Image Segmentation using OTSU Algorithm.
International Journal of Computer Applications, 975, 8887.
[64] Pai, P. Y., Chang, C. C., Chan, Y. K., Tsai, M. H., and Guo, S. W. (2012). An Image Segmentation-Based Thresholding
Method. Journal of Imaging Science and Technology, 56(3), 30503-1.
[65] Patil, A. B., and Shaikh, J. A. (2016). OTSU thresholding method for flower image segmentation. International Journal of
Computational Engineering Research (IJCER), 6(05).
[66] Hemalatha, R., Thamizhvani, T., Dhivya, J., Joseph, J., Babu, B., and Chandrasekaran, R. (2018). Active Contour Based
Segmentation Techniques forMedical Image Analysis. Robert Koprowski, Intech Open.
http://dx.doi.org/10.5772/intechopen.7457.
Authors’ Profiles
Folasade O. Isinkaye received her BSc in Computer Science from Ondo State University (EKSU), Ado-Ekiti, MSc
and PhD in Computer Science from the University of Ibadan, Oyo State, Nigeria. She is a Senior Lecturer in the
Department of Computer Science, Ekiti State University, Ado-Ekiti, Nigeria. Her research interests include
Recommender Systems, Data Mining, Machine Learning and Information Systems. She is a member of professional
bodies which include, Computer Professionals (Registration Council of Nigeria (CPN)) and Association for
Computing Machinery (ACM). She was a visiting PhD scholar at the Laboratory for Knowledge Management,
Politecnico di Bari, Italy.
Abiodun G. Aluko had a BSc degree in Computer Science from Ekiti State University, Ado-Ekiti, Nigeria. He is
an experienced Information Technologist with different IT roles which include Data Protection on Windows and
Linux machine, Microsoft Office 365, Application Lifecycle Management, IT Management, IT Support, IT Service
Monitoring with technical knowledge in Cloud Computing. He has worked for some private companies which
include Tee Vision Technologies, Fidelity Bank Plc and currently, he is working with Tek Experts as a Software
Support Engineer on behalf of its client Micro Focus (Formerly HP/Hewlett Packard Enterprise), He has various
certifications in Microsoft, Linux, Oracle, and Microfocus.

40 Segmentation of Medical X-ray Bone Image Using Different Image Processing Techniques
Copyright © 2021 MECS I.J. Image, Graphics and Signal Processing, 2021, 5, 27-40
Olayinka A. Jongbo obtained both his BSc (Ed) and MSc in Computer Science from the prestigious Ekiti State
University, Ado-Ekiti, Ekiti State, Nigeria. His research interest lies within the areas of Machine Learning, Data
Mining, Image Processing and Big Data Analytics. He has publications in reputable peer-reviewed journals.
How to cite this paper: Folasade Olubusola Isinkaye, Abiodun Gabriel Aluko, Olayinka Ayodele Jongbo, " Segmentation of
Medical X-ray Bone Image Using Different Image Processing Techniques", International Journal of Image, Graphics and Signal
Processing(IJIGSP), Vol.13, No.5, pp. 27-40, 2021.DOI: 10.5815/ijigsp.2021.05.03