An Image-Based Bone fracture Detection Using AForge Library
1 like1,410 views
1) The document discusses a technique for detecting bone fractures in x-ray images using edge detection methods like Gaussian and Canny edge detection. 2) It involves preprocessing the x-ray image, applying Gaussian filtering to remove noise, using Canny edge detection to identify edges, and inverting the image to make fractures more visible. 3) The method is implemented using the AForge library and is found to accurately detect bone fractures in x-ray images for use in medical applications like aiding doctors' diagnoses.
![ISSN: 2312-7694
Tahira et al, / International Journal of Computer and Communication System Engineering (IJCCSE), Vol. 2 (5), 2015, 682-685
683 | P a g e
© IJCCSE All Rights Reserved Vol. 02 No.05 Oct 2015 www.ijccse.com
images are frequently degraded by different unknown type
of noise which should be removed for accurate results. But
before noise removal first we will convert our x-ray
image into Black & White image by applying suitable
Threshold value. This black and white image will be
converted into gray scale for further processing.
2.Noise Removal:
As mostly x-ray images are degraded by noise so it is
important to remove the noise and smoothen the digital
image so by using Guassian filter image can be smoothen
and by removing its unnecessary details and noise.
Guassian filter is two dimensional filter and degree of
smoothening operator performs a weighted average of
surrounding pixels where surrounding depends upon the
Guassian distribution in a digital image.
This technique proves itself more effective to smoothen
the digital image because it has basis for human visual
sensitivity.
3.Canny edge detection :
An edge is the limit between an item and the foundation.
Edge recognition is basically distinguishing the objects
from their boundary where pixel intensity sharply gets
changed.[4]
To detect the edges of broken bone in x-ray image we
have used Canny edge detection technique.
Canny edge detection techniques is highly significant
because it filters out unnecessary details of image by
reduces the amount of data while preserving necessary
details of edges.
Canny edge detection technique is very much helpful for
accurate results because it includes series of useful steps :
Canny edge technique is more flexible to use
because it gives flexibility to determine the edge
thickness according the user requirements.
Canny edge detection is good for accuracy
because it Sobel operator to find edge strength.
Sobel operator uses 2-D spatial gradient to find
edge strength and 3x3 matrix to calculate both x
and y gradients which ensures more precision
(KURNIAWAN, PUTRA, GEDE, & SUDANA,
2014).
This will gives only one response for each edge,
which eliminate the confusion while edge
detection.
Important parameters which can effect every
resultant image are
i) Threshold value
ii) Guassian Gradient value
Steps of Canny Edge Detection filter are:
1) First removal of noise because it may possible that
Guassian noise removal may have used simple mask.
2) After noise removal, image gets smoothen so the next
step of canny edge detection is find out gradient using
Sobel operator.
3) Finding out the total gradient value using formula
(KURNIAWAN, PUTRA, GEDE, & SUDANA,
2014) |G|=|Gx|+|Gy|
4) Now we will find the edge direction using given
formula (KURNIAWAN, PUTRA, GEDE, &
SUDANA, 2014)
5) After finding out the direction of edges, non
maximum suppression must be applied, which is used
to differentiate edge pixels from non edge pixels. Non
edge pixels are set as 0 in order to get un necessary
emerging edges. This will give slimmer edge line.
6) At last breaking points in edges are founded out.
These breaking points may come when operator
output fluctuate above and below the threshold value.
4. Image negative :
In last step we will invert output image from canny edge
detection so that the image must look clear enough to use
in analysis of routine cases.
System overview:Gradient (X) Gradient (y)](https://image.slidesharecdn.com/rp10155927-160107194258/85/An-Image-Based-Bone-fracture-Detection-Using-AForge-Library-2-320.jpg)
![ISSN: 2312-7694
Tahira et al, / International Journal of Computer and Communication System Engineering (IJCCSE), Vol. 2 (5), 2015, 682-685
685 | P a g e
© IJCCSE All Rights Reserved Vol. 02 No.05 Oct 2015 www.ijccse.com
Conclusion:
The paper presents the Guassian and Canny methods to
analyse an x-ray image to detect the fractures. The
program is applied on real images and it provided almost
96.9% accurate results. And remaining accuracy is just
effected by the poor quality of x-ray images. So more
better results image quality must be improved so that the
accuracy should assist the radiologists to detect fractures
in an x-ray image.
References
[1] Aishwariya, R., Geetha, M. K., & Archana, M. (n.d.).
Computer-Aided Fracture Detection Of X-Ray Images.
[2] Chai, H. Y., Wee, L. K., Swee, T. T., Salleh, S.-H.,
Ariff, A., & others. (2011). Gray-level co-occurrence
matrix bone fracture detection.
[3] Ciang, C. C., Lee, J.-R., & Bang, H.-J. (2008).
Structural health monitoring for a wind turbine system:
a review of damage detection methods. Measurement
Science and Technology , 19 (12), 122001.
[4] KARTIKA, S., & others. (2013). DIGITAL IMAGE
PROCESSING USING SOBEL EDGE DETECTION
ALGORITHM IN FPGA. Journal of Theoretical &
Applied Information Technology , 58 (1).
[5] Kumar, S., & Pandey, P. (n.d.). Implementation of X-
Ray Image Segmentation by Using Edge Detection
Based On Sobel Edge Operator.
[6] KURNIAWAN, S. F., PUTRA, D., GEDE, I. K., &
SUDANA, A. K. (2014). BONE FRACTURE
DETECTION USING OPENCV. Journal of
Theoretical & Applied Information Technology , 64
(1).
[7] Pathan, M., Jusoff, K., Alias, M., Razali, Y., Qureshi,
B., & others. (2011). Implication of image processing
algorithm in remote sensing and GIS applications.
Journal of Theoretical and Applied Information
Technology , 34 (1), 34-41.
[8] PETRONAS, U. (2011). MEAN AND STANDARD
DEVIATION FEATURES OF COLOR
HISTOGRAMUSING LAPLACIAN FILTER FOR
CONTENT-BASED IMAGE RETRIEVAL. Journal
of Theoretical and Applied Information Technology ,
34 (1).
[9] Sachin R.Mahajan, P. (2012). Review of An Enhance
Fracture Detection Algorithm Design Using X-Rays
Image Processing. International Journal of Innovative
Research in Science, Engineering and Technology .
[10] Swathika.B1, A. B. (n.d.). Radius Bone Fracture
Detection Using Morphological Gradient Based Image
Segmentation Technique. Swathika.B et al, / (IJCSIT)
International Journal of Computer Science and
Information Technologies, Vol. 6 (2) , 2015, 1616-
1619 .
[11] Umadevi, N., & Geethalakshmi, S. (2012). Bone
Structure and Diaphysis Extraction Algorithm for X-
Ray Images. International Journal of Advanced
Research in Computer Science and Software
Engineering (IJARCSSE) , 2 (2).
[12] KURNIAWAN, S. F., PUTRA, D., GEDE, I. K., &
SUDANA, A. K. (2014). BONE FRACTURE
DETECTION USING OPENCV. Journal of
Theoretical & Applied Information Technology , 64
(1).
[13] Grier, S., Turner, A., & Alvis, M. (1996). The use of
dual-energy x-ray absorptiometry in animals.
Investigative Radiology , 31 (1), 50-62.
[14] Abbas, Waseem, Nasim Abbas, and Uzma
Majeed. "PERFORMANCE ENHANCEMENT
OF END-TO-END QUALITY OF SERVICE IN
WCDMA WIRELESS NETWORKS." Science
International 26.2 (2014).](https://image.slidesharecdn.com/rp10155927-160107194258/85/An-Image-Based-Bone-fracture-Detection-Using-AForge-Library-4-320.jpg)
An Image-Based Bone fracture Detection Using AForge Library
- 1. ISSN: 2312-7694 Tahira et al, / International Journal of Computer and Communication System Engineering (IJCCSE), Vol. 2 (5), 2015, 682-685 682 | P a g e © IJCCSE All Rights Reserved Vol. 02 No.05 Oct 2015 www.ijccse.com An Image-Based Bone fracture Detection Using AForge Library Tahira younas Department of Software Engineering, Fatima Jinnah Women University Rawalpindi, Pakistan Tahira564@gmail.com Abstract- The paper discusses the technique of accurate edge detection which is important to find out bone fractures from an x-ray image using digital image analysis accurately using Guassian and Canny edge detection methods. This system is built using AForge library combined with Guassian and Canny edge detection methods. Use of proposed technique demonstrates its wide application in medical field. Index Terms- Edge, Fracture, Edge detection, filter, Invert, Threshold, Guassian , Canny. I. INTRODUCTION Now a day digital image processing is an extended region with applications, especially in dynamic transmission of pictures, feature coding (video chatting), advanced libraries, picture database, remote detecting, and other specific connected utilization. This reason leads digital image analysis as helpful tool in medical, education, art and crime prevention etc (Pathan, Jusoff, Alias, Razali, Qureshi, & others, 2011) In medical field doctors are able to collect quantitative and qualitative information about physiology and anatomy using medical images and by applying digital image processing’s techniques. Similarly X-ray, MRI , digital radiography and ultrasound are used now a days for decision making. These images are used by applying different techniques of digital image processing (KURNIAWAN, PUTRA, GEDE, & SUDANA, 2014). One of the most established and as often as utilized gadget to catch human bones is X-Ray. A x-beam makes pictures of any bone in the body and is primarily used to recognize problem in human bones. Similarly when we use x-ray image of bone to detect bone fracture, digital images are segmented and different algorithms are applied to identify the broken edges of bone. The techniques to identify edges are canny, sobel , prewitt , and robert but in this paper we are using Gussian filter for noise removal and canny edge detection which locate the edges by using the boundary of bone detection (Aishwariya, Geetha, & Archana). Bone fracture can be detected by applying various techniques such as SAMUEL used Canny edge detection method using OPENCV to detect bone edges (KURNIAWAN, PUTRA, GEDE, & SUDANA, 2014). Similarly Jaskirat kaur et al. ‘2looks at picture division of X-beam picture utilizing different edge identification strategies and found that best division results were gotten utilizing careful edge identification . Subodh kumar et al performed X-beam picture division utilizing sobel edge recognition technique . Satanage et al has connected diverse edge location strategies on X-beam pictures. furthermore, watched that the ordinary edge recognition strategies likewise give best division results . Tian et al. (2003) has executed the strategy for recognizing femur cracks in x-beam pictures by processing the edge between the pole hub and ‘the neck pivot. Donnelley et al. (2008) have made a CAD framework for the long bone break recognition. Which is used to improve the manual examination of X-ray images (Chai, Wee, Swee, Salleh, Ariff, & others, 2011). Ouyang et al. (1998) had proposed surface investigation of spinal trabecular bone structure by utilizing higher request measurement investigation. Materka et al. (2000) worked on bone mineral thickness assessed by method for Dualphoton Absorptiometry (DXA) (Chai, Wee, Swee, Salleh, Ariff, & others, 2011). Manual fracture detection by area specialists is the most exact and accurate but it has lengthy algorithm on the other hand auto edge detection is very fast but it requires much more effort for accuracy (Sachin R.Mahajan, 2012). In this paper we used four steps to detect the fractured bone edges. In first step we have performed pre- processing then Gaussian sharpen filter is applied to remove the noise then canny edge detection to detect the image edges and then by inverting image we will have final image in which fractured part of image will be much more clearer. II. METHODOLOGY 1.Pre-processing: x-ray images are used for bones analysis. They don’t give better results for muscles and tissues. Some x-ray images may be dark, some may light. Moreover mostly the x-ray
- 2. ISSN: 2312-7694 Tahira et al, / International Journal of Computer and Communication System Engineering (IJCCSE), Vol. 2 (5), 2015, 682-685 683 | P a g e © IJCCSE All Rights Reserved Vol. 02 No.05 Oct 2015 www.ijccse.com images are frequently degraded by different unknown type of noise which should be removed for accurate results. But before noise removal first we will convert our x-ray image into Black & White image by applying suitable Threshold value. This black and white image will be converted into gray scale for further processing. 2.Noise Removal: As mostly x-ray images are degraded by noise so it is important to remove the noise and smoothen the digital image so by using Guassian filter image can be smoothen and by removing its unnecessary details and noise. Guassian filter is two dimensional filter and degree of smoothening operator performs a weighted average of surrounding pixels where surrounding depends upon the Guassian distribution in a digital image. This technique proves itself more effective to smoothen the digital image because it has basis for human visual sensitivity. 3.Canny edge detection : An edge is the limit between an item and the foundation. Edge recognition is basically distinguishing the objects from their boundary where pixel intensity sharply gets changed.[4] To detect the edges of broken bone in x-ray image we have used Canny edge detection technique. Canny edge detection techniques is highly significant because it filters out unnecessary details of image by reduces the amount of data while preserving necessary details of edges. Canny edge detection technique is very much helpful for accurate results because it includes series of useful steps : Canny edge technique is more flexible to use because it gives flexibility to determine the edge thickness according the user requirements. Canny edge detection is good for accuracy because it Sobel operator to find edge strength. Sobel operator uses 2-D spatial gradient to find edge strength and 3x3 matrix to calculate both x and y gradients which ensures more precision (KURNIAWAN, PUTRA, GEDE, & SUDANA, 2014). This will gives only one response for each edge, which eliminate the confusion while edge detection. Important parameters which can effect every resultant image are i) Threshold value ii) Guassian Gradient value Steps of Canny Edge Detection filter are: 1) First removal of noise because it may possible that Guassian noise removal may have used simple mask. 2) After noise removal, image gets smoothen so the next step of canny edge detection is find out gradient using Sobel operator. 3) Finding out the total gradient value using formula (KURNIAWAN, PUTRA, GEDE, & SUDANA, 2014) |G|=|Gx|+|Gy| 4) Now we will find the edge direction using given formula (KURNIAWAN, PUTRA, GEDE, & SUDANA, 2014) 5) After finding out the direction of edges, non maximum suppression must be applied, which is used to differentiate edge pixels from non edge pixels. Non edge pixels are set as 0 in order to get un necessary emerging edges. This will give slimmer edge line. 6) At last breaking points in edges are founded out. These breaking points may come when operator output fluctuate above and below the threshold value. 4. Image negative : In last step we will invert output image from canny edge detection so that the image must look clear enough to use in analysis of routine cases. System overview:Gradient (X) Gradient (y)
- 3. ISSN: 2312-7694 Tahira et al, / International Journal of Computer and Communication System Engineering (IJCCSE), Vol. 2 (5), 2015, 682-685 684 | P a g e © IJCCSE All Rights Reserved Vol. 02 No.05 Oct 2015 www.ijccse.com The systematic overview of this program is explained in detail in given flowchart. The flow chart gives step by step operations performed on an image to get require results. Fig 1.Flowchart of proposed bone fracture detection system Step by step explanation of the whole procedure is given below. 1. First of all user ,must select an x-ray image of bone for analysis. 2. System will perform pre-processing by applying suitable threshold value to convert this image in two colours. 3. After applying threshold system will convert this image into eight bit image to apply Guassian filter to remove the image noise. 4. Now system will apply GuassianSharpen filter to remove unnecessary details of the image. 5. In next step system will identify edges by applying Canny edge detection filter. 6. After applying Canny filter system will invert the image so that the output should be more clearer to the user. 7. In last step user will identify broken edges from the output image. Results and Analysis : In this part of the paper corresponding results are analyzed. Fig 2.operational view of X-ray image for output Image 1 and image 2 are x-ray images taken from internet for analysis. And the system produced accurate results. image 3 is taken from (KURNIAWAN, PUTRA, GEDE, & SUDANA, 2014) for comparison purpose which gave 100% results, similarly image 4 is taken from (Swathika.B1)which also gave 98% accuracy. Input x-ray image Applying threshold Applying Guassian filter Applying Canny filter Image negative Output image Fracture extraction
- 4. ISSN: 2312-7694 Tahira et al, / International Journal of Computer and Communication System Engineering (IJCCSE), Vol. 2 (5), 2015, 682-685 685 | P a g e © IJCCSE All Rights Reserved Vol. 02 No.05 Oct 2015 www.ijccse.com Conclusion: The paper presents the Guassian and Canny methods to analyse an x-ray image to detect the fractures. The program is applied on real images and it provided almost 96.9% accurate results. And remaining accuracy is just effected by the poor quality of x-ray images. So more better results image quality must be improved so that the accuracy should assist the radiologists to detect fractures in an x-ray image. References [1] Aishwariya, R., Geetha, M. K., & Archana, M. (n.d.). Computer-Aided Fracture Detection Of X-Ray Images. [2] Chai, H. Y., Wee, L. K., Swee, T. T., Salleh, S.-H., Ariff, A., & others. (2011). Gray-level co-occurrence matrix bone fracture detection. [3] Ciang, C. C., Lee, J.-R., & Bang, H.-J. (2008). Structural health monitoring for a wind turbine system: a review of damage detection methods. Measurement Science and Technology , 19 (12), 122001. [4] KARTIKA, S., & others. (2013). DIGITAL IMAGE PROCESSING USING SOBEL EDGE DETECTION ALGORITHM IN FPGA. Journal of Theoretical & Applied Information Technology , 58 (1). [5] Kumar, S., & Pandey, P. (n.d.). Implementation of X- Ray Image Segmentation by Using Edge Detection Based On Sobel Edge Operator. [6] KURNIAWAN, S. F., PUTRA, D., GEDE, I. K., & SUDANA, A. K. (2014). BONE FRACTURE DETECTION USING OPENCV. Journal of Theoretical & Applied Information Technology , 64 (1). [7] Pathan, M., Jusoff, K., Alias, M., Razali, Y., Qureshi, B., & others. (2011). Implication of image processing algorithm in remote sensing and GIS applications. Journal of Theoretical and Applied Information Technology , 34 (1), 34-41. [8] PETRONAS, U. (2011). MEAN AND STANDARD DEVIATION FEATURES OF COLOR HISTOGRAMUSING LAPLACIAN FILTER FOR CONTENT-BASED IMAGE RETRIEVAL. Journal of Theoretical and Applied Information Technology , 34 (1). [9] Sachin R.Mahajan, P. (2012). Review of An Enhance Fracture Detection Algorithm Design Using X-Rays Image Processing. International Journal of Innovative Research in Science, Engineering and Technology . [10] Swathika.B1, A. B. (n.d.). Radius Bone Fracture Detection Using Morphological Gradient Based Image Segmentation Technique. Swathika.B et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 6 (2) , 2015, 1616- 1619 . [11] Umadevi, N., & Geethalakshmi, S. (2012). Bone Structure and Diaphysis Extraction Algorithm for X- Ray Images. International Journal of Advanced Research in Computer Science and Software Engineering (IJARCSSE) , 2 (2). [12] KURNIAWAN, S. F., PUTRA, D., GEDE, I. K., & SUDANA, A. K. (2014). BONE FRACTURE DETECTION USING OPENCV. Journal of Theoretical & Applied Information Technology , 64 (1). [13] Grier, S., Turner, A., & Alvis, M. (1996). The use of dual-energy x-ray absorptiometry in animals. Investigative Radiology , 31 (1), 50-62. [14] Abbas, Waseem, Nasim Abbas, and Uzma Majeed. "PERFORMANCE ENHANCEMENT OF END-TO-END QUALITY OF SERVICE IN WCDMA WIRELESS NETWORKS." Science International 26.2 (2014).