2022 3rd International Conference on Intelligent Engineering and Management (ICIEM)

Detection by Clinicians Comparison of Intelligence

Bone Fracture Detection System with SIFT algorithm
for Identification of Bone Fracture
2022 3rd International Conference on Intelligent Engineering and Management (ICIEM) | 978-1-6654-6756-8/22/$31.00 ©2022 IEEE | DOI: 10.1109/ICIEM54221.2022.9853197

Ram Mukesh P. Dass

Research Scholar Project Guide
Department of Electronics and Communication Engineering Department of Electronics and Communication Engineering
Saveetha School of Engineering Saveetha School of Engineering
Saveetha Institute of Medical and Technical Sciences Saveetha Institute of Medical and Technical Sciences
Saveetha University Saveetha University
Chennai, India Chennai, India
[email protected] [email protected]

Abstract - The aim of the study was to detect by clinicians' published in Google Scholar on the topic of fracture
comparison of the innovative Intelligence Bone Fracture detection. In recent times bone fractures were explored
detection system (IBFDS) with a sift algorithm for the mostly for the detection of fractured parts[1]. Most of the
identification of bone fracture. Material and Methods: The research deals with the high percentage of fractured bones
input images of X-Rays are collected from Kaggle.com. The
and the technology of the fractured bones has increased and
sample size of 2 groups was 20. The data set contains bone
fractures which are considered as an input image. MATLAB has a high percentage of SIFT of the success rate [2],[15];
(2013) is used to compare an Intelligence bone fracture detection Detection of fractured bones cannot be easily detected. By an
system with a deep neural network for accurate fracture Intelligent Bone fracture detection system of an X-ray image,
detection. SPSS version 21 was used for the statistical analysis. we can detect the accurate fractured part of a bone fracture
A total of 20 samples were processed for the 2 groups to better image [3] but the analysis cannot be done. Nowadays
accurately detect the bone fracture using a g power of 80%. fractured bones are occurring due to accidents and many
Results: The comparison of IBFDS over the deep neural
reasons, and to detect the fractured part of an image,
network was done independent sample t-test using SPSS
software. The accuracy appeared for Innovative IBFDS was especially for the doctors, SIFT feature is much useful [5],[6];
91.67% and for the neural network was 83.56. Significance was and this process is used for the detection of a fractured area
observed for the comparison of parameter accuracy through and efficient use of the image processing and automated
SPSS Version 2.1 and the accuracy for SIFT was minimum fracture detection [7]. Using automated fracture detection,
(77.34 ± 6.40) followed by IBFDS (64.69±13.68). The there could be a way to automatically detect vertebral
Independent sample test revealed statistical significance and got fractures in patients of older people who have CT scans of
P (0.003). Conclusion: The SIFT algorithm has better accuracy their spines. This could help them get better care before they
than IBFDS for bone fracture detection. break [14]. The broken part of the X-ray image was shown.
To find fractures and improve accuracy, parts of bones can be
Keywords: X-ray images, SIFT algorithm, Innovative
Identification Bone Fracture Detection System, Fracture looked at [10]. Our teams have written about a wide range of
detection, Deep Neural Network. subjects in the past As a way to make this experience even
better, we've worked on the study now.
I. INTRODUCTION
Though new technologies are arising, some old techniques
Bone fractures may occur due to different reasons. are used mostly. Still, the size of the fracture is very small
Depending upon their fractured bones, Image processing and this could not be detected. Therefore, an innovative
techniques are used often in varied medical fields. The Scale-Invariant Scale Feature Transform was designed to
clinician’s images of bone fractures are detected by using detect bone fractures irrespective of their sizes.
SIFT and Innovative Identification Bone Fracture Detection
System [1],[3];The fractured part of the bone cannot be II. MATERIALS AND METHODS
detected even by obtaining a high accuracy rate. Several
fractured bones have the high fractured area in an X-ray Bone fracture detection is the goal of this project. The
image for such cases it is not easy to detect all fractured parts, research was conducted at the Saveetha School of
and by using a Scale Invariant Feature (SIFT) [8] the Engineering, Saveetha Institute of Medical and Technical
Clinicians images of fractured parts can be detected easily. Sciences, and Saveetha University's Department of
Most applications of this research are broken bones and Electronics and Communication Engineering. A total of 20
fracture detections [4],[5]; In the past five years, 53 articles samples were processed for the 2 groups to better accurately
were published in IEEE Explore and 128 articles were detect the bone fracture using a g power of 80%.

533
978-1-6654-6756-8/22/$31.00 ©2022 IEEE

Authorized licensed use limited to: Nirma University Institute of Technology. Downloaded on April 25,2024 at 06:09:55 UTC from IEEE Xplore. Restrictions apply.
2022 3rd International Conference on Intelligent Engineering and Management (ICIEM)

Utilizing clinical c.com, the alpha error-threshold by 0.05, 95 Input Image Accuracy of IBFDS by Accuracy by SIFT
percent confidence interval, and power of 80 percent were Clinicians Comparison algorithm
used to compute the sample size. This study compares the
accuracy of 20 samples per group based on prior research. INPUT 1 79.43 83.96
Kaggle.com was used to gather the samples for this study
[14]. INPUT2 69.38 75.49

INPUT3 83.56 91.67

MATLAB (2013) and Windows 10 -i3 core are the primary
tools utilized in this project because of the considerable (and INPUT4 75.84 89.4
tight) variety of image-making-ready instruments they have
produced. After converting the x-ray images of fractured INPUT5 68.55 77.54
from RGB to dark scale, the specified framework begins by
removing the commotion. An algorithm known as SIFT is INPUT6 60.99 67.84
used to analyse images of fractured bones in the first step of
INPUT7 76.93 84.29
the IBFDS to identify their fragmented areas.
INPUT8 51.95 59.68
The development of a bone fracture detection system using
X-ray pictures and the sift algorithm. An X-ray image would INPUT9 81.41 88.65
show the shattered portion of a bone image. Within the image
process section, the images are processed by a Matlab INPUT10 70.06 77.54
process. Image enhancement is of input and the image
INPUT11 58.78 64.25
segmentation is processed for the Innovative Intelligent Bone
fracture Detection System, and image extraction occurs for INPUT12 71.06 77.85
an X-ray image. The IBFDS process method helps to extract
the fractured part by a sift algorithm.[19]. [18] IBFDS is used INPUT13 65.73 71.90
to simulate the real-world situation. The dependent variables
are accuracy for IBFDS and the independent variables in this INPUT14 60.41 66.43
study are images with bone fractures.
INPUT15 68.97 73.65

III. STATISTICAL ANALYSIS INPUT16 77.45 85.68

IBM SPSS version 2.1 was used to conduct the statistical INPUT17 59.77 64.76
analysis for this investigation. The t-test for independent
samples was used to compare the means, standard deviations, INPUT18 79.86 88.43
and standard error means between the groups in this study.
INPUT19 56.87 60.73

IV. RESULTS INPUT 20 77.94 85.06

Table 1 represents the accuracy values of IBFDS and Neural

networks and all the data sets used in this work are suitable
for fracture detection. The accuracy values obtained from the
clinicians were considered as a dependent variable in the
fracture detection of the bones by a sift gives better results
than neural IBFDS. Figure 1 shows the X-ray image is a
fractured bone of the leg, and the input image processed by
Matlab, FIG. 2 a displays the original image, and the
following images are shown in detail: Fig. 2 b: Bone fracture
image enhancement, and Fig. 2 c: Image segmentation
Figure 1 X-Ray image of bone fracture
processed part, and Fig. 2 D: IBFDS feature extraction.

TABLE I. ACCURACY VALUES OF SIFT AND IBFDS. FOR THE THIRD INPUT
IMAGE, THE MAXIMUM ACCURACY OBTAINED IS 91.67, AND FOR ALL THE
INPUTS SIFT OCCURRED WITH MORE ACCURACY COMPARED TO THE
IBFDS.

534

Authorized licensed use limited to: Nirma University Institute of Technology. Downloaded on April 25,2024 at 06:09:55 UTC from IEEE Xplore. Restrictions apply.
2022 3rd International Conference on Intelligent Engineering and Management (ICIEM)

Figure 2 The input images processed by Matlab. (a) Shows the input image,
(b) shows the image enhancement of bone fracture, (c) shows the image
segmentation processed part, (d) represents the feature extraction of the
image processed, (e) Shows the IBFDS part of bone fracture.

Figure 3 represents using an IBFDS for the accurate fracture

of the bone. The fractured area of bone by using a SIFT is
shown in Fig. 4. This method detects the minor fractures that Figure 4 The detected bone fracture of X-ray image using SIFT algorithm
occurred at the ib bones.
The statistics for the group are shown in Table 2. IBFDS had
a standard deviation of 64.68, while SIFT had a standard
deviation of 6.40, in a statistical analysis of 20 samples, with
a standard error of 0.9723. The standard deviation of IBFDS
is less compared to SIFT.

TABLE II. GROUP STATISTICS: IT IS USED TO COMPARE THE NUMERIC

VALUES WITH RESPECT TO ONE OR MORE DIFFERENT VARIABLES. THE
ACCURACY OF IBFDS AND NEURAL NETWORKS WERE COMPARED.
Standard Std. Error
Group n Mean
deviation means

Accuracy IBFDS 20 77.349 6.40901 .97230

SIFT
Accuracy 20 64.683 13.68537 2.647
Algorithm

Table 3 shows the accuracy of the 2 groups and a significant

Figure 3 The detected bone fracture of X-ray image using IBFDS algorithm difference of 0.003 was observed between the two groups
from the independent sample test. As shown in Figure 5, SIFT
appears to have a higher degree of accuracy than the IBFDS,
and the SIFT standard deviation is slightly better than IBFDS.

TABLE III. SAMPLES FROM AN INDEPENDENT SOURCE CALCULATE THE STANDARD ERROR AND TEST FOR SIGNIFICANCE. 95 PERCENT OF THE TIME, A P-
VALUE OF LESS THAN 0.05 IS DEEMED STATISTICALLY SIGNIFICANT.

Levene's Test for Equality of variables t-Test for equality of means 95% confidence of difference

F sig. t df Sig (2-tailed) Mean difference Std Error difference lower upper

Equal variables to be
58.466 .003 6.987 10 .000 16.865543 2.09692 12.6755 22.24133
assumed

Equal variables not to be

6.987 8.470 .000 16.865543 2.09692 12.0847 22.41466
assumed

535

Authorized licensed use limited to: Nirma University Institute of Technology. Downloaded on April 25,2024 at 06:09:55 UTC from IEEE Xplore. Restrictions apply.
2022 3rd International Conference on Intelligent Engineering and Management (ICIEM)

Figure 5 A comparison of the precision of the SIFT and IBFDS algorithms. SIFT's standard deviation is slightly lower than that of the IBFDS, and its accuracy is
better than that of the IBFDS. Comparing SIFT Algorithm (X-axis) and IBFDS Algorithm (Y-axis).

V. DISCUSSION In the future, we can try with more samples for different
algorithms to improve the accuracy level.
In this method, the accuracy for SIFT features is high as
compared to IBFDS. The highest accuracy obtained for a VI. CONCLUSION
SIFT was 91.67 and for IBFDS 83.56, and also observed
statistically significant that is less than 0.04 which is less than The results of the study showed that the Scale in-variant
0.05. By the accuracy value of fractured and un-fractured feature transform has a better accuracy rate of 91.67% than
bone of different samples, proved that the exact region of the the Intelligence bone fracture detection system method for
fractured area of a Clinicians image is detected for 20 bone fracture detection.
samples. For a fractured image, the accuracy obtained for
various samples is high for the Scale-invariant feature while VII. ACKNOWLEDGEMENT
compared with the Intelligence Bone fracture Detection.
The authors would like to express their gratitude towards
To verify the effectiveness of the proposed method, accuracy Saveetha School of Engineering, Saveetha Institute of
is chosen as a parameter. The accuracy of the fractured bone Medical and Technical Sciences (Formerly known as
is high [6] and the exact area of the fractured part is not Saveetha University) for providing the necessary
detected [20]. The accuracy of SIFT appears to be the most infrastructure to carry out this work successfully.
consistent result within a meaning range from 75 to 90 [17].
When compared to X-ray images with un fractured bones, the REFERENCES
total percentage of broken bones is significant . Due to the
poor quality of the X-ray image, there was a high rate of [1] Anbu, R. Tamil, V. Suresh, Revathy Gounder, and Abinaya Kannan.
2019. “Comparison of the Efficacy of Three Different Bone
mistakes [18].
Regeneration Materials: An Animal Study.” European Journal of
Dentistry 13 (1): 22–28.
The factor that affects the accuracy and percentage of the [2] Avinash, Kavarthapu, Sankari Malaippan, and Jayakumar Nadathur
detection process is the quality of the image. Image quality Dooraiswamy. 2017. “Methods of Isolation and Characterization of
Stem Cells from Different Regions of Oral Cavity Using Markers: A
includes factors such as the quality of the X-ray image. As a
Systematic Review.” International Journal of Stem Cells 10 (1): 12–
limitation of the study, the modifications can be done by 20.
using different types of fractured bones of X-rays for a better [3] “[3]Bone Fracture Detection System Using Image Processing and
accurate part of fracture detection. The shape and size of bone Matlab.” 2019. International Journal of Innovative Technology and
Exploring Engineering. https://doi.org/10.35940/ijitee.l3950.1081219.
fracture can be used in the detection process, and it can be
[4] Bungartz, Matthias, Georg Matziolis, Frank Layher, Victoria Horbert,
applied for the exact area of bone fracture and adaptive use Alexander Maslaris, and Olaf Brinkmann. 2021. “Biodegradable
of image processing. Mainly it is easy for doctors to detect Cement Augmentation of Gamma Nail Osteosynthesis Reduces
fractures. The limitations of this study were that they can Migration in Pertrochanteric Fractures, a Biomechanical in Vitro
Study.” Clinical Biomechanics 84 (March): 105327.
compare SIFT with different deep learning algorithms but the
[5] Chen, Feng, Ying Tang, Yujia Sun, Vishnu Priya Veeraraghavan,
fracture does not occur as much. The novel algorithm was Surapaneni Krishna Mohan, and Chuanxin Cui. 2019. “6-Shogaol, a
implemented accuracy is improved by taking more samples. Active Constiuents of Ginger Prevents UVB Radiation Mediated
In the proposed study we are limited to 20 samples (Images). Inflammation and Oxidative Stress through Modulating NrF2

536

Authorized licensed use limited to: Nirma University Institute of Technology. Downloaded on April 25,2024 at 06:09:55 UTC from IEEE Xplore. Restrictions apply.
2022 3rd International Conference on Intelligent Engineering and Management (ICIEM)

Signaling in Human Epidermal Keratinocytes (HaCaT Cells).” Journal [21] Mäyränpää, Mervi K., Ilkka Helenius, Helena Valta, Mikko I.
of Photochemistry and Photobiology. B, Biology 197 (August): Mäyränpää, Sanna Tctures in Children: Accuracy of Vertebral Fracture
111518. Assessment.” Bone. https://doi.org/10.1016/j.bone.2007.05.012.
[6] Deng, Jiawen, and Wenteng Hou. n.d. “Prevention of Bone Loss and
Fractures Following Solid Organ Transplantations: Protocol for a
Systematic Review and Network Meta-Analysis.”
https://doi.org/10.1101/19013797.2017. “IBFDS: Intelligent Bone
Fracture Detection System.” Procedia Computer Science.
https://doi.org/10.1016/j.procs.2017.11.237.
[7] Dinesh, S., P. Kumaran, S. Mohanamurugan, R. Vijay, D. Lenin
Singaravelu, A. Vinod, M. R. Sanjay, Suchart Siengchin, and K.
Subrahmanya Bhat. 2020. “Influence of Wood Dust Fillers on the
Mechanical, Thermal, Water Absorption and Biodegradation
Characteristics of Jute Fiber Epoxy Composites.” Journal of Polymer
Research 27 (1). https://doi.org/10.1007/s10965-019-1975-2.
[8] Dinesh, S. P. Saravana, A. V. Arun, K. K. Shantha Sundari, Christine
Samantha, and K. Ambika. 2013. “An Indigenously Designed
Apparatus for Measuring Orthodontic Force.” Journal of Clinical and
Diagnostic Research: JCDR 7 (11): 2623–26.
[9] Downey, Colum, Stephen Flannery, Ben Murphy, Tiarnan Daly, Sarah
Conway, Mohammed Gaffar, Peter Dawson, et al. 2021. “A Multi-Site
Review of Second Hip Fractures across 6 Dublin Teaching Hospitals.”
Irish Journal of Medical Science, March.
https://doi.org/10.1007/s11845-021-02607-1.
[10] Gheena, S., and D. Ezhilarasan. 2019. “Syringic Acid Triggers
Reactive Oxygen Species-Mediated Cytotoxicity in HepG2 Cells.”
Human & Experimental Toxicology 38 (6): 694–702.
[11] Gomathi, A. C., S. R. Xavier Rajarathinam, A. Mohammed Sadiq, and
S. Rajeshkumar. 2020. “Anticancer Activity of Silver Nanoparticles
Synthesized Using Aqueous Fruit Shell Extract of Tamarindus Indica
on MCF-7 Human Breast Cancer Cell Line.” Journal of Drug Delivery
Science and Technology 55 (101376): 101376.
[12] Govindaraju, Lokhasudhan, Prasanna Neelakantan, and James L.
Gutmann. 2017. “Effect of Root Canal Irrigating Solutions on the
Compressive Strength of Tricalcium Silicate Cements.” Clinical Oral
Investigations 21 (2): 567–71.
[13] ara, Masahide, Tetsunori Saikawa, Mamoru Kurokawa, Toshiie Sakata,
and Hironobu Yoshimatsu. 2004. “Leg Fat Percentage Correlates
Negatively With Coronary Atherosclerosis.” Circulation Journal.
https://doi.org/10.1253/circj.68.1173.
[14] ose, Jerry, Ajitha, and Haripriya Subbaiyan. 2020. “Different
Treatment Modalities Followed by Dental Practitioners for Ellis Class
2 Fracture – A Questionnaire-Based Survey.” The Open Dentistry
Journal 14 (1):59-65.
[15] Kakkad, Subhash. 2005. “Hand Bones Fracture Frame.” Practical
Orthopaedics. https://doi.org/10.5005/jp/books/10662_31.
[16] Kamisetty, Supradeep Kumar, Jaya Krishanan Verma, Arun, Shanta
Sundari, Shyamala Chandrasekhar, and Aravind Kumar. 2015. “SBS
vs Inhouse Recycling Methods-An Invitro Evaluation.” Journal of
Clinical and Diagnostic Research: JCDR 9 (9): ZC04–08.
[17] Kaplunov, O. A., S. N. Biryukov, and E. Yu Nekrasov. 2016.
“Prevention of Thromboembolic Complications afterSurgical
Treatment for Lower Extremity BoneFractures: Experience in
Rivaroxaban Use in ClinicalPractice.” N.N. Priorov Journal of
Traumatology and Orthopedics.
https://doi.org/10.17816/vto201623257-61.
[18] Ke, Yang, Mohammed Saleh Al Aboody, Wael Alturaiki, Suliman A.
Alsagaby, Faiz Abdulaziz Alfaiz, Vishnu Priya Veeraraghavan, and
Suresh Mickymaray. 2019. “Photosynthesized Gold Nanoparticles
from Catharanthus Roseus Induces Caspase-Mediated Apoptosis in
Cervical Cancer Cells (HeLa).” Artificial Cells, Nanomedicine, and
Biotechnology 47 (1): 1938–46.
[19] Kodama, M. 2010. “P9-7 Leg Movement Durations and Periodicities
in Periodic Limb Movements of Three Patients with Brain Lesion,
Spinal Cord Lesion, and Leg Bone Fractures.” Clinical
Neurophysiology. https://doi.org/10.1016/s1388-2457(10)60637-2.
[20] iu, Jun, and Danwei Wang. 1992. “Image Compression Using Neural
Network.” Neural and Stochastic Methods in Image and Signal
Processing. https://doi.org/10.1117/12.130869.

537

Authorized licensed use limited to: Nirma University Institute of Technology. Downloaded on April 25,2024 at 06:09:55 UTC from IEEE Xplore. Restrictions apply.